Assessment of display performance for medical imaging...

Assessment of display performance for medical imaging systems:Executive summary of AAPM TG18 reportChairman:

Ehsan Sameia!

Duke Advanced Imaging Laboratories, Departments of Radiology, Physics, and Biomedical Engineering,Duke University, DUMC 3302, Durham, North Carolina 27710

TG18 Members:

Aldo BadanoFDA, CDRH

Dev ChakrabortyUniversity of Pennsylvania (currently with University of Pittsburgh)

Ken ComptonClinton Electronics (currently with National Display Systems)

Craig CorneliusEastman Kodak Company (currently a consultant)

Kevin CorriganLoyola University

Michael J. FlynnHenry Ford Health system

Bradley HemmingerUniversity of North Carolina, Chapel Hill

Nick HangiandreouMayo Clinic, Rochester

Jeffrey JohnsonSarnoff Corp, NIDL (currently with Siemens Corporate Research)

Donna M. Moxley-StevensUniversity of Texas, Houston, M.D. Anderson Cancer Center

William PavlicekMayo Clinic, Scottsdale

Hans RoehrigUniversity of Arizona

Lois RutzGammex/RMI

Ehsan SameiDuke University Medical Center

Jeffrey ShepardUniversity of Texas, Houston, M.D. Anderson Cancer Center

Robert A. UzenoffFujifilm Medical Systems USA

Jihong WangUniversity of Texas, Southwestern Medical Center (currently with University of Texas, Houston)

Charles E. WillisBaylor University, Houston, Texas Children’s Hospital (currently with University of Texas, Houston)

sReceived 27 August 2004; revised 4 December 2004; accepted for publication 23 December 2004;published 31 March 2005d

1205 1205Med. Phys. 32 „4…, April 2005 0094-2405/2005/32 „4…/1205/21/$22.50 © 2005 Am. Assoc. Phys. Med.

1206 Ehsan Samei et al. : Performance assessment of medical displays 1206

Digital imaging provides an effective means to electronically acquire, archive, distribute, and viewmedical images. Medical imaging display stations are an integral part of these operations. There-fore, it is vitally important to assure that electronic display devices do not compromise imagequality and ultimately patient care. The AAPM Task Group 18sTG18d recently published guide-lines and acceptance criteria for acceptance testing and quality control of medical display devices.This paper is an executive summary of the TG18 report. TG18 guidelines include visual, quantita-tive, and advanced testing methodologies for primary and secondary class display devices. Thecharacteristics, tested in conjunction with specially designed test patternssi.e., TG18 patternsd,include reflection, geometric distortion, luminance, the spatial and angular dependencies of lumi-nance, resolution, noise, glare, chromaticity, and display artifacts. Geometric distortions are evalu-ated by linear measurements of the TG18-QC test pattern, which should render distortion coeffi-cients less than 2%/5% for primary/secondary displays, respectively. Reflection measurementsinclude specular and diffuse reflection coefficients from which the maximum allowable ambientlighting is determined such that contrast degradation due to display reflection remains below a 20%limit and the level of ambient luminancesLambd does not unduly compromise luminance ratiosLRdand contrast at low luminance levels. Luminance evaluation relies on visual assessment of lowcontrast features in the TG18-CT and TG18-MP test patterns, or quantitative measurements at 18distinct luminance levels of the TG18-LN test patterns. The major acceptable criteria for primary/secondary displays are maximum luminance of greater than 170/100 cd/m2, LR of greater than250/100, and contrast conformance to that of the grayscale standard display functionsGSDFd ofbetter than 10%/20%, respectively. The angular response is tested to ascertain the viewing conewithin which contrast conformance to the GSDF is better than 30%/60% and LR is greater than175/70 for primary/secondary displays, or alternatively, within which the on-axis contrast thresh-olds of the TG18-CT test pattern remain discernible. The evaluation of luminance spatial uniformityat two distinct luminance levels across the display faceplate using TG18-UNL test patterns shouldyield nonuniformity coefficients smaller than 30%. The resolution evaluation includes the visualscoring of the CX test target in the TG18-QC or TG18-CX test patterns, which should yield scoresgreater than 4/6 for primary/secondary displays. Noise evaluation includes visual evaluation of thecontrast threshold in the TG18-AFC test pattern, which should yield a minimum of 3/2 targetsvisible for primary/secondary displays. The guidelines also include methodologies for more quan-titative resolution and noise measurements based on MTF and NPS analyses. The display glare test,based on the visibility of the low-contrast targets of the TG18-GV test pattern or the measurementof the glare ratiosGRd, is expected to yield scores greater than 3/1 and GRs greater than 400/150for primary/secondary displays. Chromaticity, measured across a display faceplate or between twodisplay devices, is expected to render au8,v8 color separation of less than 0.01 for primary displays.The report offers further descriptions of prior standardization efforts, current display technologies,testing prerequisites, streamlined procedures and timelines, and TG18 test patterns. ©2005 Ameri-can Association of Physicists in Medicine. fDOI: 10.1118/1.1861159g

Key words: medical display, liquid crystal display, cathode ray tube, image quality, qualityassurance, quality control, acceptance testing, picture archiving and communication systemsPACSd

andn-andns,s fo

menesstoyr of

in

pro-anydical

ionfortests

s ofd be

e-r ac-playing

dmin-

I. INTRODUCTION

The adoption of digital detectors and Picture ArchivingCommunication SystemssPACSd has provided healthcare istitutions an effective means to electronically archiveretrieve radiological images. Medical display workstatioan integral part of PACS, are used to display these imagediagnostic and clinical purposes. Considering the fundatal importance of image quality to the overall effectivenof a diagnostic imaging practice, it is vitally importantassure that electronic display devicessalso termed softcopdisplaysd do not compromise image quality as a numbestudies have suggested.1–3

According to the American Association of Physicists4
MedicinesAAPMd professional guidelines,the performance
Medical Physics, Vol. 32, No. 4, April 2005

r-

assessment of electronic display devices falls within thefessional responsibilities of medical physicists. While mprior publications have addressed some aspect of medisplay performance,5–15prior evaluation and standardizatefforts have fallen short of providing an unified approachtesting the performance of display devices such that thewould take into consideration all the important aspectdisplay performance, be specific to medical displays, anrelatively easy to implement in a clinical setting.

AAPM Task Group 18sTG18d recently completed a rport which suggests standard guidelines and criteria foceptance testing and quality control of medical disdevices.16 The intended audience of the report is practicmedical physicists, engineers, researchers, radiology a
istrative staff, manufacturers of medical displays, radiolo-

tioncur-ray

tech

um-nceuto-olo-smef refou

ion,ke

oce-playdingdis-e otes

thisrt in

n thce tedu

f thed toed,

d inicana-izedaryed

cer-sy

dicaetiveomt fors o

ly, arfor-

en

alityablealitys 3.1f theular

t pat-rderom-

of thee-le II

crip-the

thewith

dis-oulddis-

ngel,nd040houldenship

per-soft-cess-and

ationonic

in

blesireds are18dur-orkill be


gists, and students interested in display quality evaluaThe report is developed such that while addressing therent dominant medical display technologies, cathodetubessCRTsd and liquid crystal displayssLCDsd, many of thetests and concepts could be adapted to future displaynologies.

The report is divided into six sections. Section one smarizes prior standardization efforts in the performaevaluation of medical display devices. Section two is a trial on the current and emerging medical display techngies. Section three sets forth prerequisites for the assesof the display performance and includes a description oquired instrumentation and TG18 test patterns. Sectionis the main body of the report containing the descriptquantification methods, and acceptance criteria for eachdisplay characteristic. Sections five and six outline prdures for acceptance testing and quality control of disdevices. The report further includes appendices proviguidelines for evaluating the performance of “closed”play systems, requirements for equivalent appearancmonochrome images, a full tabular description of TG18patterns, and a selected bibliography.

Considering the significant extent of the TG18 report,paper aims to provide an executive summary of the repoa more condensed format. This paper focuses mainly otesting procedures and criteria of the most direct relevanacceptance testing and quality control procedures. Thecational, advanced, and detailed descriptive portions oreport are not included. Interested individuals are referrethe full report for a complete description of the eliminatsummarized, and referenced sections.

II. GENERAL PREREQUISITES FOR DISPLAYASSESSMENTS

II.A. Classification of Display Devices

In recognition of the currently accepted practice anaccordance with the guidelines set forth by the AmerCollege of Radiology17 and the Food and Drug Administrtion, display devices for medical imaging are characterin the TG18 report as either primary or secondary. Primdisplay systems are those used for the interpretation of mcal images. They are typically used in radiology and intain medical specialties such as orthopedics. Secondarytems are those used for viewing medical images by mestaff or specialists other than radiologists after an interprreport is rendered. The operator’s console monitors cmonly used to “adjust” the images before they are seninterpretation are treated as a primary display in termcontrast response but secondary otherwise.

II.B. Required Tools

II.B.1. Instrumentation

Although many display tests can be performed visualmore objective and quantitative evaluation of display pemance requires special test tools. The required instrum
vary in their complexity and cost depending on the context

.-

-

nt-r

y

ft

eo-

i-

s-l

-

f

ts

of the evaluationsresearch, acceptance testing, or qucontrold and how thorough the evaluation needs to be. TI summarizes the required instruments for display quevaluation. The readers are advised to consult Sectionand 4-6 of the TG18 report to determine the subset otools and their performance requirements for the partictests being performed.

II.B.2. Test Patterns

The TG18 report recommends the use of specific testerns for performance evaluation of display devices in oto facilitate comparisons of measurements. The recmended patterns are designated with a nomenclatureform TG18-xyz, wherex, y, andz describe the type and drived variants of a pattern. The patterns are listed in Taband a few examples are illustrated in Fig. 1. The full destion of the patterns are in Sec. 3.2 and Appendix III ofTG18 report.

While the electronic copy of the TG18 report providespatterns in multiple formats, they may also be generatedthe aid of the information provided in the report. Whenplaying the patterns, no special processing functions shbe applied. The 16-bit version of the patterns should beplayed with a window width and level set to cover the rafrom 0 to 4095swindow width, WW=4096, window leveWL=2048d, except for the TG18-PQC, TG18-LN, aTG18-AFC patterns, where a WW of 4080 and WL of 2should be used. For 8-bit patterns, the displayed range sbe from 0 to 255sWW=256,WL=128d. For some of thpatterns, it is also essential to have a one-on-one relatiobetween the image pixels and the display pixels.

II.B.3. Software

Though not essential, software tools can facilitate theformance assessment of display devices. They includeware for semiautomated generation of test patterns, proing software for assessment of resolution and noise,spreadsheets for recording and manipulating the evaluresults. Some tools are provided along with the electrcopy of the TG18 report. Further information is availableSec. 3.3 of the report.

II.C. Initial Steps for Display Assessment

II.C.1. Availability of Tools

Before starting the tests, the availability of the applicatools and test patterns should be verified. Lists of detools for acceptance testing and quality control purposeprovided in the following section of this paper. The TGtest patterns should be stored on the display workstationing installation, or otherwise be accessible from a netwarchive. This approach ensures that the same pattern w
utilized for all future testing.

ion,

nts


TABLE I. Instrumentation used for display quality evaluation.

Instrument Desired requirements Purpose

Near-range luminance meter • Calibration traceable to NIST Luminance and luminance uniformitymeasurements• 0.05–1000 cd/m2 luminance range

• Better than 5% accuracy

• Better than 10−2 sideally 10−3d precision

• Aperture rangeø5 deg

• Better than 3% compliance with theCommission Internationale de L’EclairagesCIEdstandard photopic spectral response

Telescopic luminance meter • Those listed above for near-range meter Luminance, luminance uniformity, reflectangular response, and glare measurements• Acceptance angleø1 deg

• Ability to focus to an areaø6 mm

Illuminance meter • Calibration traceable to NIST Reflection and ambient lighting measureme

• 1–1000 lux illuminance range

• Better than 5% accuracy

• Better than 3% compliance with theCIE standard photopic spectral response

• 180 deg cosinesLambertiand responseto better than 5% out to 50° angulation

Colorimeter • Calibration traceable to NIST Chromaticity measurements

• 1–1000 cd/m2 luminance range

• Better than 0.004su8 ,v8d accuracy

Digital camera • Low noise and wide dynamic range Quantitative resolution and noisemeasurements• 1–500 cd/m2 luminance range

• .5123512 matrix size

• 10- to 12-bit depth

• Equipped with a focusable macro lens

• Variable frame rate/integration times up to 1 s

• Digital interface to a computer

• Calibrated for camera luminance, flat-fieldresponse, noise, and MTF

• Equipped with a stable stand or tripod withdirectional adjustments

Light source • Uniform luminance.200 cd/m2 Quantitative specular reflection measurement

• Small enough to subtend 15° fromcenter of display

Illumination device • See TG18 report Sec. 3.1.3 Quantitative diffuse reflection

Baffle • Light absorbing characteristics Glare and luminance measurements

• 5–15 mm opening

Cone • Light absorbing characteristics Glare and luminance measurements

• 5 mm opening andø60 deg angulardivergence

Light absorbing cloth or hood • Light absorbing characteristics Display evaluation in the areas that haveno control over the level of ambient lighting


vicete. I

be

ini-h ashervedsed

d byarea

eckthe

medtemhebetheand

thAnyade

orbousmeponefor

d be.

withel ofis-ilitytionmentlight

n itst tohisr toam-thedis-op-

f aightital

d

as-d, andeerlyased

istics-eg the


II.C.2. Display Placement

Prior to testing, the proper placement of a display deshould be verified and adjustments made as appropriathe placement of a display device, the following shouldconsidered:

1. Display devices should always be positioned to mmize specular reflection from direct light sources succeiling lights, film illuminators, or surgical lamps. Treflection of such light sources should not be obseon the faceplate of the display in the commonly uviewing orientations.

2. Many display devices, such as CRTs, are affectemagnetic fields; they should not be placed in anwith strong magnetic fieldssi.e., vicinity of MRI scan-nersd, unless properly shielded.

3. Displays should be placed ergonomically to avoid nand back strain at reading level with the center ofdisplay slightly below eye level.

II.C.3. Start-up Procedures

Prior to evaluation, the display device should be warup for approximately 30 min. In addition, the general sysfunctionality should be verified by a quick review of tTG18-QC fFig. 1sadg test pattern. The pattern shouldevaluated for distinct visibility of the 16 luminance steps,continuity of the continuous luminance bars at the rightleft of the pattern, the absence of gross artifacts, andproper size and positioning of the active display area.adjustments to vertical and horizontal size must be mprior to performing the luminance measurements.

Dust and smudges on the face of the display will absreflect, or refract emitted light possibly resulting in erronetest results. In addition, newly installed displays are sotimes covered with a protective plastic layer, which uremoval can leave residual marks on the faceplate. B

TABLE I.

Instrument Desired re

Measuring microscope or magnifier • Magnificationù25–50x

• Equipped with a metric rø0.05 mm divisions

• Focusing capabilities

• Allow a working distance

Flashlight • None

Lint-free cleaning tissue glass-cleaningsolution

• Recommended by the d

Two rulers and angle measurement device • 1 m in length

Tape measure • Flexible and 20–30 cm

testing a display device, the cleanliness of the faceplate


n

e

,

-

e

should be verified. If the faceplate is not clean, it shoulcleaned following the manufacturer’s recommendations

II.C.4. Ambient Lighting Level

The artifacts and loss of image quality associatedreflections from the display surface depend on the levambient lighting. As shown in Table III, illumination of dplay device surfaces in various locations of a medical facmay vary by over two orders of magnitude. The reflecmeasurement described in a later section of this docudelineates a method to determine the maximum ambientlevel appropriate for any given display device based oreflection and luminance characteristics. It is importanverify that the ambient lighting in the room is below tmaximum. The condition for the tests should be similathose under normal use of the equipment. By recordingbient light levels at a reference point at the center offaceplate and noting the location and orientation of theplay devices at acceptance testing, it will be possible totimize repeatability of testing conditions in the future. Idisplay device is equipped with a photocell for ambient ldetection, its use should be in compliance with the DigImaging and Communication in MedicinesDICOMd gray-scale standard display functionsGSDFd as further discussebelow.

II.C.5. Pretest Luminance Settings

Before the performance of a display system can besessed, proper display area size should be establishethe maximum luminanceLmax and the minimum luminancLmin must be checked to verify that the device is propconfigured. The desired values should be determined bon the desired luminance ratio, the reflection characterof the system, and the ambient lighting levelssee the reflection and luminance sections belowd. Using a luminancmeter, the luminance values should be recorded usin

ntinued.d

ments Purpose

Visual resolution measurements

with

12.5 mm

Allow inspections in dark

y manufacturer Used for cleaning the faceplate, if needed

Angular response and specular reflectioncoefficient measurements

ngth Geometric distortions measurements

sCo

quire

eticle

ofù

ispla

in le

TG18-LN8-01sor TG18-LN12-01d test pattern forLmin and

.e reec-

-g thiseper

ance

calirat-

-

oughturn-pen-ualdicalOtherlec-

formsitua-ver-Forsuchr theof

er of


TG18-LN8-18 sor TG18-LN12-18d for Lmax, respectivelyFor these measurements, ambient illumination should bduced to negligible levels using a dark cloth shroud if nessary. If the measured values forLmax andLmin are not appropriate, the proper values should be established usinbrightness and contrast controls of the display. Otherwthe display device should be serviced before testing itsformance. The TG18 report further recommends compliof medical display systems with the DICOM GSDF.15 Beforeinitiating the testing procedures, the device should bebrated or otherwise its calibration verified within its opeing luminance range defined byLmax andLmin.

II.C.6. Personnel

The acceptance and quality controlsQCd testing of a dis

TABLE II. Test patterns recommended for display quality evaluation. Tsize and in either DICOM or tiff format. Some patterns are available i

Set Series Type

Multipurposes1 k and 2 kd TG18-QC Vis./QntTG18-BR Visual

TG18-PQC Vis./Qnt.

Luminances1 k onlyd TG18-CT VisualTG18-LN Quant.TG18-UN Visual

TG18-UNL Quant.TG18-AD Visual

TG18-MP Visual

Resolutions1 k and 2 kd TG18-RH Quant.

TG18-RV Quant.

TG18-PX Quant.TG18-CX Visual

TG18-LPH Visual

TG18-LPV Visual

Noise s1 k onlyd TG18-AFC VisualTG18-NS Quant.

Glare s1 k onlyd TG18-GV VisualTG18-GQ Quant.TG18-GA Quant.

Anatomicals2 k onlyd TG18-CH VisualTG18-KN VisualTG18-MM Visual

play system must be performed by an individualssd having


-

e,-

-

appropriate technical and clinical competencies. Even ththe vendor is expected to perform some testing beforeing a display system over to the user, the user must indedently test the systemssd. For acceptance testing and annQC evaluation, the tests should be performed by a mephysicists trained in display performance assessments.staff including biomedical engineers, in-house service etronic technicians, or trained x-ray technologists can persome of the tests described herein; however, in suchtions, a qualified medical physicist should accept full osight responsibilities and final approval of the results.monthly or quarterly QC, the tests can be delegated toqualified professionals as well as long as they work undedirect supervision of the medical physicist. The daily QCa display system should be performed by the operator/us

tterns are divided into six sets. Most patterns are available in 102431024 s1 kd8048 s2 kd size.

Images Description

1 Resolution, luminance, distortion, artifacts1 Briggs pattern, low contrast detail vs luminance1 Resolution, luminance, contrast transfer for prints

1 Luminance response18 DICOM grayscale calibration series2 Luminance and color uniformity, and angular response2 Same as above with defining lines1 Contrast threshold at low luminance for evaluating

display reflection1 Luminance responsesbit depth resolutiond

3 Five horizontal lines at three luminance levels for LSFevaluation

3 Five vertical lines at three luminance levels for LSFevaluation

1 Array of single pixels for spot size1 Array of Cx patterns and a scoring reference for

resolution uniformity3 Horizontal bars at 1 pixel width, 1/16 modulation, three

luminance levels3 Vertical bars at 1 pixel width, 1/16 modulation, three

luminance levels

1 4AFC contrast-detail pattern, four CD values3 Similar to RV/RH, five uniform regions for noise

evaluation

2 Dark spot pattern with low contrast object3 Dark spot pattern for glare ratio measurement8 Variable size dark spot patterns

1 Reference anatomical PA chest pattern1 Reference anatomical knee pattern2 Reference anatomical mammogram pattern

he pan 20432

the system. Radiology staff using electronic displays should

ctedestsn-ical the


be familiar with the daily testing procedure and experesults. All personnel responsible for performing QC twill require initial training specific to their level of resposibility and periodic retraining and mentoring by med

Fig. 1. Examples of TG18 test patterns: TG18-QCsad, TG18-PQCsbd, TG1sgd, TG18-UN80 shd, TG18-UNL10 sid, TG18-MP sjd, TG18-RV89skd, TGTG18-GQBsqd, TG18-CHsrd, TG18-KN ssd, TG18-MM1 std, and TG18-M

physics staff.


II.C.7. Specific Prerequisites for AcceptanceTesting

Acceptance testing requires close communication with

scd, TG18-LN8-01sdd, TG18-LN8-08sed, TG18-LN8-18sfd, TG18-UNL80RH50sld, TG18-CX smd, TG18-AFC snd, TG18-GV sod, TG18-GA30spd,ud.

8-CT18-M2s

vendor for understanding and documenting the operational

recg thvicefac-ac

temthe

whitionroom-

romn to

blishuringtab-tilizeplay


features and dedicated QC utilities of the system. Anyommended service and/or calibration schedule, includinservices provided, tests performed, and the sercalibration intervals, must be obtained from the manuturer, ideally as part of the purchasing process. Prior toceptance testing, the characteristics of the display sysdelivered should be verified against those specified inpurchase agreement. A database should be establishedincludes information such as display type, size, resolumanufacturer, model, serial number, manufacture date,number, display identificationsif applicabled, associated dis

Fig. 1.

play hardwarese.g., display controllerd and test patterns


-e/

-s

ch,

available on the systems. All delivered documentation fthe vendor should also be reviewed with special attentiothe testing results performed at the factory.

II.C.8. Specific Prerequisites for Quality Control

The initial acceptance testing data are used to estaand maintain expected performance. Data acquired droutine QC testing must be compared to the limits eslished around the baseline values. It is also essential to uthe same pattern for repeat evaluations of a given dis

tinued).
sCon
device. The use of worksheets and checklists will help in

reclec-sup

play

inep-rol,e arforfer-d reagerforow

medtion

dolor in

ay btes

asth

ed.

n as-nlinoulme

cer-PHthe

r dis-rowerter of

cantthe

gridould

cur-istor-ces-tion

.

anti-thebestuld

sCon


establishing and monitoring the baselines. It is stronglyommended to record and maintain this information in etronic databases. Most commercial calibration packagesport automated recording, tracking, and analysis of disQC results.

III. ASSESSMENT OF DISPLAY PERFORMANCE

The performance assessment of a display deviceclinical setting might be performed in the context of acctance testing, prior to first clinical use, or quality contthroughout the life of the device. Tables IV and V providlist of the tests, the required tools, and the expected pemance for the two types of procedures with specific reence to the TG18 report. Depending on the interest ansources, additional advanced tests are further encourFor QC tests, hardware features and reproducible pemance can reduce the need for very frequent testing. Hever, it is recommended that initially the tests be performore frequently. If stability is maintained, a determinacan be made to decrease the frequency of testing.

The sections below provide the assessment methogies. It is generally ideal to perform the tests in the ordewhich they are discussed as some of the latter tests minfluenced by parameters that are addressed in earlierFull descriptions of the specific display characteristicswell as advanced testing procedures are provided inTG18 report,16 to which the interested readers are referr

III.A. Geometric Distortions

Geometric distortions of displayed images are ofteconcern in cathode-ray tubesCRTd display devices. The ditortions can be in concave, convex, skewed, or other noear forms. The magnitude and type of such distortions shbe evaluated and, if deemed inappropriate, adjusted to

Fig. 1.

certain minimum requirements as noted below.


-

-

a

-

-d.--

-

ets.

e

-det

III.A.1. Visual Evaluation

The geometric distortion of a display system is astained visually using the TG18-QC or the TG18-LPV/Ltest pattern. The patterns should be maximized to fillentire usable display area. For displays with rectangulaplay areas, the patterns should cover at least the nardimension of the display area and be placed at the centhe area used for image viewing. The patternssd should beexamined from a viewing distance of 30 cm.

The patterns should appear straight without signifigeometric distortions, and should be properly scaled toaspect ratio of the video source pixel format so that theof the TG18-QC pattern appears square. The lines shappear straight indicative of proper linearity without anyvature or waviness. Some small barrel and pincushion dtions are normal for CRT devices but should not be exsive. For the TG18-LPV and TG18-LPH patterns, in addi

to straightness, the lines should appear equally spaced

III.A.2. Quantitative Evaluation

Spatial accuracy for geometric distortions can be qufied using the TG18-QC test pattern, maximized to fillentire display area. Using a straight edge as a guide for afit and with the aid of a flexible plastic ruler, distances sho

TABLE III. Typical ambient lighting levels.

Area Illuminationsluxd

Operating rooms 300–400Emergency medicine 150–300Hospital clinical viewing stations 200–250Staff offices 50–180Diagnostic reading stationssCT/MR/NMd 15–60Diagnostic reading stationssx-raysd 2–10

tinued).

rticaandeo ths, anionsen toulrel

an

houldtheanceere-

entsnceainsthe

fullyhing

tions re

n

c. 4.2.3.1


be measured in square areas in the horizontal and vedirections in each of the four quadrants of the patternwithin the whole patternsFig. 2d. It is important to assure thlocations of the cross hatches be viewed perpendicular tdisplay’s faceplate. In each quadrant, between quadrantwithin the whole pattern, the maximum percent deviatbetween the measurements in each direction and betwemeasurements in the horizontal and vertical directions shbe determined. The percentages should be calculated intion to the smallest of the values being compared.

The measured spatial deviations shall be less than 2%

TABLE IV. Tests, tools, and acceptance criteria for acceptance testingthe TG18 report.

Test Major required tools ProcedureEquipment Patterns

Geometricdistortions

Flexible ruler ortransparenttemplate

TG18-QC See Sec. 4.1.4

Reflectiona Measuring ruler,light sources,luminance andilluminancemeters,illuminator

TG18-AD See Secs. 4.2.3

Luminanceresponse

Luminance andilluminancemeters

TG18-LNTG18-CTTG18-MP

See Secs. 4.3.4

Luminancedependenciesb

Luminancemeter,luminanceangular responsemeasurementtool

TG18-UNLTG18-LNTG18-CT

See Secs. 4.4.3

Resolutionc Luminance metermagnifier

TG18-QCTG18-CXTG18-PX

See Secs. 4.5.3

Noisec None TG18-AFC See Sec. 4.6.3

Veiling glare Baffled funnel,telescopicphotometer

TG18-GVTG18-GVNTG18-GQs

See Secs. 4.7.3

Chromaticity Colorimeter TG18-UNL80 See Sec. 4.8.4

Note: Acronyms:Lamb= ambient luminance,Lmin= minimum luminance,Lma

measured and GSDF contrast, Cx= Cx score, RAR= resolution-addreaIn the absence of illumination devices, this acceptance testing can bbAngular tests are not required as a part of annual quality control.cMore objective resolution and noise measurements can be performe

5% for primary or secondary displays, respectively. If a dis-


l

ed

heda-

d

play device does not meet these criteria, adjustments sbe made to the distortion control of the device. Often, asarea of the display is increased or decreased, the luminwill also increase or decrease in a nonlinear fashion. Thfore, it is important to make and finalize such adjustmprior to testing and adjustment of the display luminacharacteristics. In addition, if a display workstation contmore than one display device, it is important to havevertical and horizontal sizes of the active areas carematched within 2%. This facilitates the subsequent matcof their luminance response characteristics.

annual quality control of electronic display systems. The section notafer to

Acceptance criteriasfor two classes of displaysd

Suggested actionsif unacceptabled

Primary Secondary

Deviationø2% Deviationø5% Readjustment,repair, orreplacement forrepeated failures

4.2.4 Lminù1.5Lamb

sideallyù4Lambd

Lminù1.5Lamb

sideallyù4Lambd

Results areused to adjustthe level ofambient lighting

4.3.3 Lmaxù170cd/m2LRù250kdø10%DLmaxø10%

Lmaxù100cd/m2

LRù100DLmaxø10%kdø20%

Readjustment,recalibration,repair, orreplacement forrepeated failures

4.4.4 Nonunif.ø30%LRdu8 ù175kduø30%

Nonunif. ø30%LRdu8 ù70kduø60%

Readjustment,repair orreplacement forrepeated failures;Angular resultsused to defineacceptable viewingangle cone

4.5.4.1.2 0øCxø4RAR=0.9−1.1ARø15

0øCxø6 Focus adjustment,repair, orreplacement forrepeated failures

All targetsvisible exceptthe smallest

Two largestsizes visible

Reverification ofluminanceresponse,otherwisereplacement

4.7.4 ù3 targets visible,GRù400

ù1 targetvisible,GRù150

Reverification ofluminanceresponse,otherwisereplacement

Dsu8 ,v8d None Replacementø0.01

aximum luminance, LR=luminance ratio,kd= maximum deviation betweeility ratio, AR= astigmatism ratio, GR= glare ratio.formed only visually using TG18-AD and the method described in Se.

described in Secs. 4.5.4 and 4.6.4 using a digital camera.

and

and

and

and

and

and

x= mssabe per

d as

reagein

-

bersno

eris-rde

tingpro-tras

playe o

ain-ouldan

re-high

tternssuchpo-e. Ifon athe

frved

oftatio

etricashedare atsed for


III.B. Display Reflection

Electronic display devices have specular and diffuseflection that can reduce image contrast and affect imquality. Ambient light reflections are more pronounceddisplay devices with thick faceplatesse.g., CRTsd comparedto those with thinner faceplatesse.g., LCDsd. They are generally reduced by the application of antireflectivesARd coat-ing on the faceplate and/or the addition of light absorwithin the faceplate of the display, but these means docompletely eliminate reflections. The reflection characttics of a medical display device should be evaluated in oto establish the maximum allowable level of ambient lighat which the device can be operated without overly commising the desired luminance performance and conthreshold.

III.B.1. Visual EvaluationIII.B.1.a. Specular Reflection CharacteristicsAn effec-

tive and simple visual test for specular reflection of a disdevice is to observe the device in the power-save modturned off. The ambient lighting in the room should be mtained at levels normally used. The display’s faceplate shbe examined at a distance of about 30–60 cm within angular view of 615 deg for the presence of specularlyflected light sources or illuminated objects. Patterns of

TABLE V. sad Tests for daily quality control of electronic display systemelectronic display systems performed by a medical physicist or by a Qto the TG18 report. For acronyms see Table IV.

Test Major required tools ProcedurEquipment Patterns

Overall visualassessment

None TG18-QC oranat. images

See Secs.or 4.10.6

Geometricdistortions

None TG18-QC See Sec.

Reflection Luminance andilluminancemeters

TG18-AD See Secs.and 4.2.4

Luminanceresponse

Luminance andilluminancemeters

TG18-LNTG18-CTTG18-MP

See Secs.and 4.3.3

Luminancedependencies

Luminance meter TG18-UNTG18-UNL

See Secs.and 4.4.4

Resolution Magnifier TG18-QCTG18-CX

See Sec. 4

contrast on the viewer’s clothing are common sources o


-

t

r

t

r

-

reflected features. In general, no specularly reflected paof high contrast objects should be seen. If light sourcesas that from a film illuminator or window are seen, thesition of the display device in the room is not appropriathigh contrast patterns such as an identification badgewhite shirt or a picture frame on a light wall are seen,ambient illumination in the room should be reduced.

III.B.1.b. Diffuse Reflection CharacteristicsThe effect odiffusely reflected light on image contrast may be obse

rformed by the display user.sbd Tests for monthly/quarterly quality controlhnologist under the supervision of a medical physicist. The section nons refer

Acceptance criteriasfor two classes of displaysd

Suggested actionsif unacceptabled

Primary Secondary

.1 See Secs.4.10.1/4.10.6

See Secs.4.10.1/4.10.6

Further /closerevaluation

.1 See 4.1.3.2 See 4.1.3.2 Further/closerevaluation

Lminù1.5Lamb

sideallyù4Lambd

Lminù1.5Lamb

sideallyù4Lambd

Readjust the levelof ambient lighting

Lmaxù170cd/m2


Lmaxù100cd/m2


Readjustment,recalibration,repair, orreplacement forrepeated failures

Nonunif. ø30% Nonunif.ø30% Readjustment,repair, orreplacement forrepeated failures

0øCxø4 0øCxø6 Focus adjustment,repair, orreplacement forrepeated failures

FIG. 2. Spatial measurements for the quantitative evaluation of geomdistortions using the TG18-QC test pattern. The small squares with dlines s- - -d define the four quadrants of the pattern, and the large squthe center encompassing the luminance patches is the one to be u

, peC tec

e

sad4.10

sbd4.1.3

4.2.3

4.3.4

4.4.3

.5.3

fgeometric distortion characterization within the whole image.

thermainvicotalwingbed inersncectio

-uredde-ouldrcebe

andthe

urceomandtlyured

mi-m.deflecum

c.ni-is

cana

andtion, 28,and

ting,mi-

acingace-

themi-einw-

epres-tingdark

tio,d.sing

il-portthetheeter

dententertele-mi-aper-not

truc-

tion-e

anf a


by alternately viewing the low-contrast patterns inTG18-AD test pattern in near total darkness and in noambient lighting, determining the threshold of visibilityeach case. A dark cloth placed over both the display deand the viewer may be helpful for establishing near tdarkness. The pattern should be examined from a viedistance of 30 cm. The threshold of visibility should notdifferent when viewed in total darkness and when vieweambient lighting conditions. If the ambient lighting rendthe “dark-threshold” not observable, the ambient illuminaon the display surface is causing excess contrast reduand the room ambient lighting needs to be reduced.

III.B.2. Quantitative EvaluationIII.B.2.a. Specular Reflection CharacteristicsThe specu

lar reflection coefficient for a display device can be measwith a small-diameter source of diffuse white light asscribed in Sec. 3.1.3 of the TG18 report. The display shbe in the power-save mode or turned off. The light sousubtending 15° from the center of the display, shouldpositionedd1 centimeters from the center of the displaybe pointed toward the center at an angle of 15° fromsurface normal. The reflected luminance of the light soshould then be measured with a telescopic photometer frdistance ofd2 centimeters from the center of the displaysimilarly angled at 15° to the normal. Finally, the direcviewed luminance of the light source should be measwith the same photometer from a distance ofd1+d2 centime-ters. The specular reflection coefficientRs is the ratio of thereflected spot luminance to the directly viewed spot lunance. All measurements should be made in a dark roo

As the artifacts associated with specular reflectionspend on the ambient lighting, the measured specular retion coefficient should be used to establish the maximallowable ambient lightingE as

E ø spCtLmind/s0.9 Rsd, s1d

where the contrast thresholdCt=DL /L ssee Fig. 3 and Se4.3.1 of TG18 reportd, corresponds to its value at the mimum luminanceLmin. For convenience, this relationship


l

e

n,

,

a

--

tabulated in Table VI so that the maximum room lightingbe identified ifRs andLmin are known. As an example, fortypical CRT with antireflectivesARd coatingsRs=0.004d op-erated at minimum luminance values of 0.5, 1, 1.5,2.0 cd/m2, the ambient lighting based on specular reflecconsideration should be less than approximately 14, 21and 31 lux, respectively. Note that in the adjustmentmeasurement of the appropriate level of ambient lighilluminance in the room should be measured with the illunance meter placed at the center of the display and foutward, so the proper amount of light incident on the fplate can be assessed.

III.B.2.b. Diffuse Reflection CharacteristicsThe lumi-nance from diffuse reflections adds to that produced bydisplay device. The ambient illumination produces a lunance ofLamb=RdE, whereE is ambient illuminance on thdisplay surface, andRd is the diffuse reflection coefficientunits of cd/m2 per lux or 1/sr. In the dark areas of a locontrast image, the change in luminanceDLt will produce arelative contrast ofDLt / sLmin+Lambd. For some devices, thluminance response can be calibrated to account for theence of a known amount of luminance from ambient lighLamb and produce equivalent contrast transfer in bothand bright regions. However, ifLamb is sufficiently large inrelation toLmin, even if the device has a high contrast rathe overall luminance ratio of the device is compromise

The diffuse reflection coefficient may be measured ustandardized illumination of the display surface with theluminator device described in Sec. 3.1.3 of the TG18 resFig. 4d. The illuminance should then be measured incenter of the display device using a probe placed oncenter of the display surface. The sensitive area of the mshould be held vertically to measure the illuminance incion the display faceplate. The induced luminance at the cof the display surface should then be measured with ascopic luminance meter as illustrated in Fig. 4. The lunance measurement should be made through the smallture at the back of the containment device so as toperturb the reflective characteristics of the containment s

FIG. 3. Contrast threshold for varied visual adaptasad and fixedsbd visual adaptationsRef. 19d. The contrast thresholddL/L for a just noticeable differencsJNDd depends on whether the observer has fixedsbd orvariedsad adaptation to the light and dark regions ofoverall scene.dL/L is the peak-to-peak modulation osmall sinusoidal test pattern.

° offthe

e in

arynces obthemen

rt

tionlux

tingmi-

nonlaymagcrensi-

mi-

theually

snt

hip-

ple,odernrater, thede-

allyhouldilitytrastanceisualll bem isgions.tternrrectrgetst toions.alu-udestivee ex-thelev-8-bitis-

ffuse


ture. The viewing aperture must be located from 8° to 12to the side from the normal so as to not interfere withmeasurement result. The diffuse reflection coefficientRd iscomputed as the ratio of the luminance to the illuminancunits of sr−1.

As diffuse reflection reduces the contrast, for primclass display devices, the level of ambient illuminashould be set to insure that the contrast in dark regionserved with ambient illumination will be at least 80% ofcontrast observed in near total darkness. This requiretranslates toLamb,0.25Lmin, or

E ø s0.25Lmind/Rd. s2d

Table VII identifies the ambient lighting for whichLamb is0.25 ofLmin as a function ofRd andLmin. As an example, foa typical CRT with AR coatingsRd=0.02 sr−1d operated aminimum luminance values of 0.5, 1, 1.5, and 2.0 cd/m2, theambient lighting based on diffuse reflection considerashould be less than approximately 7, 12, 19, and 25respectively. In situations where the level of ambient lighcan be strictly controlled and taken into account in the lunance calibration of the display device, a largerLamb can betoleratedsLamb,Lmin/1.5d as noted in the next section.

III.C. Luminance Response

The human visual system perceives brightness in alinear fashion.18 Ideally, the luminance response of a dispdevice should match this nonlinear response such that ivalues are displayed in equally perceptible luminance inments. While limited due to variations in the contrast se

TABLE VI. Maximum allowable ambient illuminanspecific minimum luminanceLmin and a specific swhich maintains specular reflections from hightabulated.

Lmax–Lmin

scd/m2d Ct Rs=0.002 Rs=0

5000–20 0.010 3492500–10 0.011 1921000–4 0.015 105500–2 0.018 63250–1 0.024 42

FIG. 4. Typical illuminating device used for the measurement of the direflection coefficient of a display device.


-

t

,

-

e-

tivity of the human eye in scenes with wide ranges of lunance levelsse.g., medical imagesd,19,20 the DICOM gray-scale standard display functionsGSDFd15 offers a way toapproach this goal by applying a specific look-up-table todisplay values, such that the display values present eqdiscriminable levels of brightness.

The intrinsic luminance responsesi.e., luminance versudisplay valued of most display devices is markedly differefrom the GSDF. It usually follows a power-law relationsfor CRTs, and a linear one for LCDs.5 In addition, the luminance response may vary over time. In CRTs, for examthe phosphor efficiency decreases as the device ages. Mdisplay devices also have utilities that automatically calibthe luminance response of the device to GSDF. Howevefunctionality and accuracy of these utilities should be inpendently verified by the user.

III.C.1. Visual Evaluation

The luminance response of a display device is visuinspected using the TG18-CT test pattern. The pattern sbe evaluated from a viewing distance of 30 cm for visibof the central half-moon targets and the four low-conobjects at the corners of each of the 16 different luminregions. Since this pattern is viewed in one state of vadaptation, it is expected that the contrast transfer wibetter at the overall brightness for which the visual systeadapted as opposed to the darkest or the brightest reWith experience, the visual characteristics of this test pacan be recognized for a system with quantitatively columinance response. In general, the low contrast tashould be visible in all regions. A common failure is nobe able to see the targets in one or two of the dark reg

The bit-depth resolution of the display should be evated using the TG18-MP test pattern. The evaluation inclascertaining the horizontal contouring bands, their relalocations, and grayscale reversals. The pattern should bamined from a viewing distance of 30 cm. In general,relative location of contouring bands and any luminanceels should not be farther than the distance between themarkersslong markersd. No contrast reversal should be dcernible.

sed on specular reflection: For a display device with alar reflection coefficientRs the ambient illumination

rast objects below the visual contrast thresholdsCtd is

mum room illuminancesluxd

Rs=0.008 Rs=0.020 Rs=0.040

87 35 1748 19 1026 10 516 6 310 4 2

ce bapecucont

Maxi

.004

17596523121

de 18ital

sval-

bi-bynce

ter’smend thcesthemethaned.

sea-

-sav

useini-the

nor-

e

ythouh am-for

with%mi-r’s

thao or-

ral,ues

re-ctive

thn

bi-unt

rgeran

besuf-

be-uredDF

e., theusing

thethind

-t,ex

val-ua-


III.C.2. Quantitative Evaluation

In the quantitative method, luminanceLspd is measureusing a calibrated luminance meter at the center of thTG18-LN test patterns, corresponding to 18 distinct digdriving levels p. The measurement ofLspd using patternother than the TG18-LN patterns may result in differentues due to the influence of veiling glare. The effect of ament illumination should be reduced to negligible levels,using a dark cloth if necessary. If a telescopic luminameter is used, in order to minimize the influence of meflare on the low-luminance measurements, the measuremay need to be made through a cone or baffle to shielinstrument from the surrounding light. For display deviwith non-Lambertian light distribution, such as a LCD, ifmeasurements are made with a near range luminancethe meter should either have an aperture angle smaller tdeg or display-specific correction factors should be appli21

The ambient luminance on the display faceplatesLambdshould either be estimated from the measuredRd values aLamb=ERd or measured directly. In the case of direct msurement, the display device should be put in the poweror blank screen-save modesotherwise turned offd. A tele-scopic luminance meter normal to the display surface iswith a light-absorbing mask placed behind the meter to mmize specular reflection from the display. Otherwiseroom lighting is set to the conditions established for themal use of the equipmentssee Sec. III B aboved. The valuesfor L8spd including Lmax8 andLmax8 are then computed by thaddition ofLamb to the measuredLspd values.

The recommended value forLmax8 is typically specified bthe vendor as the highest value that can be used wicompromising other performance characteristics, suclifetime or resolution.Lmax8 should be greater than 171 cd/2

for primary displays17 and 100 cd/m2 for secondary displays, and should be within 10% of the desired valueboth classes of display. Furthermore, for workstationsmultiple monitors,Lmax8 should not differ by more than 10among monitors.Lmax8 should be such that the desired lunance ratioLR8=Lmax8 /Lmin8 is obtained. If the manufacturerecommendations are not available, it is recommendedthe luminance ratio of a display device be set equal tgreater than 250 for all primary class devices.19 For second

TABLE VII. Maximum room lighting based on diffuluminanceLmin and a specific diffuse reflection cillumination which maintains 80% contrast in dacalculated as 0.25Lmin/Rd.

MaxiLmax–Lmin

scd/m2d Rs=0.005 Rs=0.010

5000–20 1000 5002500–10 500 2501000–4 200 100500–2 100 50250–1 50 25


tse

er,5

e

d

ts

t

ary class devices, LR8 should be no less than 100. In geneLmin8 should be within 10% of the nominally desired valfor both classes of display.

As ambient lighting can impact the low luminancesponse of a display device and reduce the device’s effeluminance ratio, a limit onLamb is further indicated. For boclasses of display devices,Lamb should ideally be less tha0.25Lmin sor 0.2Lmin8 d. In situations where the level of ament lighting can be strictly controlled and taken into accoin the luminance calibration of the display device, a laLamb can be tolerated, butLamb should always be less thLmin/1.5 sor Lmin8 /2.5d. If necessary, arrangements shouldmade to reduce the room lighting in order to achieve aficiently smallLamb.

In evaluating the luminance response of the displaytween the maximum and minimum extremes, the measluminance values should be related to the DICOM GSluminance response in terms of the contrast response, i.slope of the measured luminance response. To do so,the DICOM’s table of just noticeable differencesJNDd indi-ces versus luminance, the JND indices for the measuredLmin8andLmax8 should first be identified. The JND indices forintermediateL8 values should then be evenly spaced withe JND range and linearly related to the actualp values useas

Ji = Jmin +PisJmax− Jmind

DP, s3d

whereJ indicates the JND indicesse.g., Fig. 5d. The measured data are then expressed as the observed contrasdi, ateach luminance stepLi8, as a function of mean JND indvalue associated with that step

di =2sLi8 − Li−18 d

sLi8 + Li−18 dsJi − Ji−[email protected] + Ji−1d. s4d

The expected response from DICOM GSDF luminanceuesdi

d is also similarly computed using the following eqtion:

flection: For a display device with a specific minimumientRd in units of cd/m2 per lux or 1/sr, the ambientgions is tabulated. The maximum room illuminance is

room illuminancesluxd

Rs=0.020 Rs=0.040 Rs=0.060

250 125 83125 62 4250 25 1725 12 812 6 4

se reoefficrk re

mum

ntra

classent

.eriaace

once alarlynce

ofas a

ingndd thessed

be. Nocm

bepat-ilityof

houldn in

dis-F.

n lin-

es.

utedtrast0%


did =

2sLid − Li−1

d dsLi

d + Li−1d dsJi − Ji−1d

@0.5sJi + Ji−1d. s5d

The difference between the measured and GSDF coresponses at any given pointkd=Maxsudi −di

dud, should beless than 10% and 20% for the primary and secondarydisplay devices, respectivelysFig. 6d. This criterion appliespecifically to contrast evaluated from the 18 measuremof luminance made at uniformly spacedp-value intervalsThe failure of a display device to meet the above critshould prompt adjustment, recalibration, repair, or replment of the device.

III.D. Luminance Dependencies

The luminance response evaluations described abovepertain to the luminance characteristics of a display devione location on the display faceplate viewed perpendicuHowever, display devices often exhibit spatial lumina


st

s

s

-

lyt.

non-uniformities and variation in contrast as a functionviewing angle, both of which should be characterizedpart of display evaluation protocol.

III.D.1. Visual EvaluationIII.D.1.a. NonuniformityThe visual method for assess

display luminance uniformity involves the TG18-UN10 aTG18-UN80 test patterns. The patterns are displayed anuniformity across the displayed pattern is visually assefrom a viewing distance of 30 cm. The patterns shouldfree of gross nonuniformities from center to the edgesluminance variations with dimensions on the order of 1or larger should be observed.

III.D.1.b. Angular DependenceAngular response mayevaluated visually using the TG18-CT test pattern. Thetern should first be viewed on axis to determine the visibof all half-moon targets. The viewing angle at which anythe on-axis contrast thresholds are rendered invisible sthen be determined by changing the viewing orientatio

FIG. 5. Example of the measured luminance for 18play levels is plotted in relation to the DICOM GSDThep-values used to measure luminance have beeearly scaled to JND indices with the values atLmax8 andLmin8 set to be equal to the JND corresponding indic

FIG. 6. Example of the contrast response compfrom 18 gray levels is related to the expected conresponse associated with the DICOM GSDF with 1tolerance limits indicated.

rmy beaxiaon o

s oac-

d b

dfive

en-umd aum

e-

ofhesethegle

nts,ing

l anld b

t recepithin

gulae oancngleits o

,

iceing.uffi-. Fold b

ction

thed toof

r thanits

singthed soost

playsfers isouldcor-

e pro-and

rec-theso bee. Theandthe

xam-betest

ele-

CXents

limit

ouldri-ncyns.the

ed-tallyltingof a

Theg ofdis-the

n thelowits

and

ideents.izon-


polar and azimuthal orientations. Alternatively, a unifotest pattern with uniformly embedded test targets maused. The viewer distance at which all targets along theor diagonal axes are visible may be used as an indicatithe angular response performance of the display in termthe viewing angle cone within which the performance isceptable. The acceptable viewing angle cone shoulclearly labeled on the display device.

III.D.2. Quantitative EvaluationIII.D.2.a. NonuniformityUsing the TG18-UNL10 an

TG18-UNL80 test patterns, luminance is measured atlocations over the faceplate of the display devicescenter andfour cornersd using a calibrated luminance meter with atttions to the precautions noted in Sec. III C. The maximluminance deviation for each display pattern is calculatethe percent difference between the maximum and minimluminance values relative to their average value, 200*sLmax

−Lmind/sLmax+Lmind. The value for an individual display dvice should be less than 30%.

III.D.2.b. Angular DependenceThe luminance of a LCDdisplay may be quantitatively evaluated as a functionviewing angle. This can be done with two basic approactheconoscopicand thegonioscopicmethods, as noted in thTG18 report. A basic quantitative test should includeevaluation of luminance ratio as a function of viewing anusing the TG18-LN test patterns. For these measuremeis useful to have a subjective understanding of the viewangle dependence to determine the specific horizontavertical angles at which quantitative measurements shoumade.

Ideally, the angular response of a display should noduce the luminance ratio by more than 30%. Thus, an acable viewing angle is defined as an angular cone wwhich LR8 is greater than 175s25030.7d for primary dis-plays and 70s10030.7d for secondary displays.22 If the lu-minance in midluminance values is measured, the anluminance results should be evaluated the same as thaxis measurements described above in terms of conformto the GSDF. The contrast response for any viewing ashould not be greater than three times the expected limaxis skdø3310%=30% for primary displays andkdø3320%=60% for secondary displaysd. For a display deviceboth LR8 andkd requirements should be met.

The viewing angle limitation for medical use of a devshould be clearly labeled on the device for optimum viewIf multiple devices of the same design are used, it is scient to assess the viewing angle limits on one devicesuch systems, the acceptable viewing angle cone shouused to arrange the monitors for minimum contrast redudue to the angular dependencies of luminance.

III.E. Display Resolution

Resolution is the ability of a display device to presentspatial details of a displayed image. This ability is relateboth the number of pixelsand the actual spatial extent
each pixel. Because of various optical and electronic pro

lff

e

s

:

it

de

-t-

rn-e

n

re

cesses, a display pixel can have a breadth that is largeits nominal value, degrading the display resolution fromideal level.

III.E.1. Visual Evaluation

Display resolution can be evaluated by visually assesthe appearance of the “Cx” patterns in the TG18-QC orTG18-CX test patterns. The patterns should be displayethat each image pixel is mapped to one display pixel. Mimage viewers have the function to accomplish this dismode. In order not to be limited by the modulation tranfunction sMTFd of the eye, the use of a magnifying glasrecommended. In the TG18-QC pattern, the examiner shinspect the displayed “Cx” targets at the center and fourners of the pattern and score the appearance using thvided scoring scale. The line-pair patterns at Nyquisthalf-Nyquist frequencies in the horizontal and vertical ditions should also be evaluated in terms of visibility oflines. The average brightness of the patterns should alevaluated using the grayscale step pattern as a referencdifference in visibility of test patterns between horizontalvertical patterns should be noted. The relative width ofblack and white lines in these patches should also be eined using a magnifier. The resolution uniformity mayascertained across the display area using the TG18-CXpattern and a magnifier in the same way that the “Cx”ments in the TG18-QC pattern are evaluated.

In the visual inspection of the TG18-QC and TG18-patterns on primary class display systems, the Cx elemshould be scored between 0 and 4 at all locations. Thiscoincides with a resolution-addressability ratiosRARdø1.15.8 For secondary class displays, the Cx scores shbe between 0 and 6sRARø1.47d. For both classes, the hozontal and vertical line-pair patterns at Nyquist frequeshould be discernible at all locations and for all directioThe TG18 report further includes a method to determineextent of the display pixelssi.e., RARd using the TG18-PXtest pattern.

III.E.2. Quantitative Evaluation

Quantification of the MTF requires the use of a displayimage digitizing system, such as a digital camera, to digicapture a portion of the display and to analyze the resuimages. The lens flare should be reduced with the usehigh f number and the aid of a cone or funnel device.magnification of the lens should result in over-samplinthe display with at least 64 camera pixels covering oneplay pixel. The camera needs to be well focused onscreen of the display under test. This is best done whelens aperture is opened to its maximum level to achievedepth-of-focus. Afterward, the lens aperture is set tosmallest level in order to achieve a large depth-of-focusminimum flare.

The TG18-RV, TG18-RH, and TG18-NS patterns provline inputs as target patterns for the MTF measuremThese patterns allow the assessment of MTF in the hor
-tal and vertical directions at three luminance levels and five

merrontThtheverixeled io bepixedis-capTheigna100h tortde

bemantionea-

ageputeld bheree rForare

nt inthetherker

mi-

S toctVelfor

sub-

Iob-

the

ere-

c. Insym-with. Val-ld besec-ptablecusde-

agey inusu-, is

d itsise isRTs,ise;the

dis-otice-

thetternosi-t are

ts arethe

antst-sizelacethemberom-

cepttheince

round, i.e.,.ndentund.e anbilitylu-

by-


locations on the display area. At each location, the cashould be securely positioned in the normal direction in fof the target area of the display and focused on the line.magnification should be determined in accordance withdisplay pixel size, camera matrix size, and the desired osampling. The camera field of view should include the pmarkers in the pattern. While the camera should be placnormal direction with respect to the faceplate, it needs trotated parallel to the faceplate such that the cameraarray is angled at 2–5 deg with respect to that of theplayed image. Images from all six patterns should betured before moving the camera to the next location.exposure time should be selected such that the digital sof the camera exceeds the dark signal by a factor ofFurthermore, the exposure time should be long enougpermit integration over multiple display frames, but shenough with respect to instabilities of the scanning andflection circuits. Ultimately the integration time shouldappropriate with respect to the integration time of the hueye, for which the experiments are conducted. Integratimes between 0.2 and 1 s are appropriate to use. The msurements should be made in a darkened room.

The 30 images should be acquired without any imcompression. The data should be transferred to a comfor data processing. The captured line patterns shoureduced to orthogonal MTFs using Fourier analysis. Tare several processing steps in the calculations, and thsults are expected to vary slightly with the methods.standardization and simplicity, the following stepssuggested:23

1. Determine the size that the image pixels represeterms of the spatial dimension on the display usingknown physical distance of the pixel markers onpatterns and the measured pixel distance of the main the captured images.

2. Linearize the image data with respect to display lunance using the luminance response of the display.

3. Add the mean value of the image from the TG18-Nthat of the TG18-RVsor TG18-RHd pattern, and subtrathe TG18-NS image pixel by pixel from the TG18-Rsor TG18-RHd image in order to remove display pixstructure. Averages of multiple images may be usedmore complete removal of structured noise. Thetracted image is used for further processing.

4. Identify a central rectangular region of interestsROIdextending along the image of the line.

5. Determine the angle of the line.6. Reproject the two-dimensionals2Dd data within the RO

along the direction of the line into subpixel bins totain the composite line spread functionsLSFd.

7. Smooth the LSF if it expresses excessive noise.8. Find the Fourier transform of the LSF, and normalize

resulting MTF.9. Divide the MTF by thesinc function associated with th

width of the LSF subpixel bins, and correct for the pviously characterized MTF of the camera systemssee
Sec. 3.1.2 of the reportd.

a

e

-

n

l

-

l.o

-

re

e-

s

Note that in some cases the LSF might be asymmetrithose cases, each side of the LSF is used to form twometric LSFs. The resultant MTFs are reported, alongtheir average, as representative of the display resolutionues of the measured MTF at the Nyquist frequency shouat least 35% for primary display devices and 25% forondary devices. Measured responses outside the accerange should prompt corrective actions in the form of foor dithering adjustments, repair, or replacement of thevice.

III.F. Display Noise

Display noise refers to statistical fluctuations in the imthat either vary spatially, so-called spatial noise, or vartime, so-called temporal noise. Temporal noise, which isally dominant in the dark regions of displayed imagesdifficult to characterize outside of a laboratory setting anperceptual influence is less well understood. Spatial nodominant in the brighter areas of displayed images. In Cphosphor granularity is the main contributor to spatial nowhile in LCDs, the dominant noise is that associated withpixelated background.

III.F.1. Visual Evaluation

The visual method to quantify the spatial noise of aplay system is based on the method to determine just nable luminance differences as a function of size usingTG18-AFC test pattern. Each quadrant of the test pacontains a large number of regions with varying target ption. In each quadrant, the contrast and size of the targeconstant. The contrast-size values for the four quadran20-2, 30-3, 40-4, and 60-6. The observer should viewpatterns from a viewing distance of 30 cm. The quadrcan be subjectively evaluated to establish the contrasrelationships for which the observer can confidently pthe position of all targets. The target visibility in each oftarget regions may also be quantified by counting the nuof targets readily visible in each of the quadrants and cputing the percent correct.

The visual evaluation should render all the targets exthe smallest one visible for primary class displays andtwo largest sizes visible for secondary class displays. Sthe mean value and the standard deviation of the backgare each linearly dependent on the luminance, their ratiosignal-to-noise, remains independent of luminance24,25

Therefore, the results of the noise evaluation are indepeof the absolute luminance value of the pattern’s backgroHowever, the failure of a device in this test can also bindication of an improper luminance response, the possiof which can be eliminated by first verifying the properminance response of the device.

III.F.2. Quantitative Evaluation

Spatial noise of a display system can be quantifiedeither single-pixel signal-to-noise ratios24 or by the normal
ized noise power spectrumsNPSd. Both methods require the

ageraethe

wayclesual

oniceld-d fod onote

n bethreoneThern.with

siredtua-uredsureageda

ixelPSsses inForug-

nt inthetherker

mi-

into

theon

ionst 20. T

tipleternping

ach

ptio

agesmay

PSonal

NPSmea-fixed

aluesse of

tem.

playn theiling

n thering.of

com-rns.f thevis-ofmanightedbe

maskundon-

n thesiblecor-is at

m-the

Thef thenceter offfled

th aesek re-e


use of a scientific-grade digital camera to capture an imof a uniform pattern displayed on the device. The camlens should be set to a highf number in order to reducveiling glare in the camera. Also, the magnification oflens should result in over-sampling of the display in athat allows sampling of spatial frequencies up to 40 cyper degree, which is the resolution limit of the human vissystem at the maximum luminance of most electrdisplays.26 The camera images should also be flat-ficorrected, compensated for gain variations, and restorethe degradation of the MTF of the camera optics basethe prior performance evaluation of the camera system, nearlier.

The central region of the TG18-NS test patterns caused as the target uniform pattern for measurements atluminance levels. The camera should be securely positiin front of the target area of the display and focused.field of view should include the pixel markers in the patteThe magnification should be determined in accordancethe display pixel size, camera matrix size, and the deover-sampling. To eliminate the effects of temporal fluctions in the luminance output, images should be captwith an integration time of about one second. The meaments should be performed in a darkened room. The imshould be transferred uncompressed to a computer forprocessing.

The quantification of the display noise by the single-psignal-to-noise ratio is noted in the TG18 report. For the Ndetermination, the captured uniform patterns are proceby Fourier analysis. There are multiple processing stepvolved and the methods can vary the results slightly.standardization and simplicity, the following steps are sgested:

1. Determine the size that the image pixels represeterms of the spatial dimension on the display usingknown physical distance of the pixel markers onpattern and the measured pixel distance of the main the captured image.

2. Linearize the image data with respect to display lunance.

3. Divide the central 3 /4 region of the captured imagemultiple, nonoverlapping regions, 1283128 or 2563256 in size. The size of these regions determinesampling interval of the resulting NPS. Dependingthe exact level of magnificationsoversamplingd and thematrix size of the camera, between nine to 64 regmay be identified. It is recommended that at leasregions be used for the assessment of the NPSachieve this, it might be necessary to acquire mulimages from the central patch of the TG18-NS patby orienting the camera toward another, nonoverlaparea of the central area of the displayed pattern.

4. Apply a two-dimensional fast Fourier transform on eregion to yield the 2D NPS.

5. Average the 2D NPS from all regions.6. Correct for the camera noise. Based on the assum

that the camera noise and the display spatial noise ar


e

r

d

ed

-sta

d-

s

o

n

uncorrelated, the NPS based on sampled camera imwithout exposure using the same integration timebe subtracted from the results.

7. Derive the orthogonal NPS from the calculated 2D Nby band averaging, excluding the data on the orthogaxes.

Since there are currently only a few examples of actualmeasurements made, and since no correlation of thesurements and diagnostic accuracy is ascertained, nocriteria are recommended at this time. However, noise vassociated with the display device should not exceed thotypical radiological images that are viewed with the sys

III.G. Veiling Glare

Veiling glare is a light-spreading phenomenon in a disdevice that leads to the degradation of image contrast ipresence of strong surrounding brightness. In CRTs, veglare is caused by internal light-scattering processes idevice’s faceplate, light leakage, and electron backscatteIn LCDs, electronic cross-talk can be viewed as a formveiling glare.

III.G.1. Visual Evaluation

The visual assessment of veiling glare can be acplished using the TG18-GV and TG18-GVN test patteThe display size must be adjusted so that the diameter owhite region is 20 cm. The observer should discern theibility of the low-contrast objects in sequential viewingthe TG18-GVN and TG18-GV patterns. Because the huvisual systems will change adaptation if it views the brfield, it is imperative that the bright field is fully blockfrom view and that no reflected light from the bright fieldobservable. This may be accomplished by the use of aor cone, which shields the human eye from the surroluminance of the pattern. No significant reduction in the ctrast of the target objects should be observed betweetwo patterns. At least three objects should be readily viin either pattern for primary class display devices. Theresponding object for secondary class display devicesleast ones5thd target.

III.G.2. Quantitative Evaluation

The quantitative evaluation of veiling glare is accoplished using a highly collimated luminance meter andTG18-GQ, TG18-GQB, and TG18-GQN test patterns.display size must be adjusted so that the diameter owhite region is 20 cm. Furthermore, the bright luminasurrounding the central measurement point at the centhe test patterns should be blocked using either a baluminance meter or a telescopic luminance meter wilight-blocking baffled funnel or cone. Using either of thdevices, the luminance in the center of the central dargion of the TG18-GQ patternL, the white luminance in th
ecenter of the white region of the TG18-GQB patternLB, and

18lay

uld20%

nceini-dis-ini-larefor

pat-e ingingt beoman

sicen

RTsary

c-t isbe

sta-

edyedandosserenableviceclasclas

dis-r iseeacucede-ed tdis-rage

e

er oftchingon.crossd thisond-

bedt mayalua-ar-

talk.harac-

cifictemitativeof theatterneachcm.fo-basis

h anlua-s are

t pat-ance

eralls orackous

s oftternsplay


the background luminance value in the center of the TGGQN patternLN, are recorded. The glare ratio for the dispis then computed as

GR= sLB − LNd/sL − LNd. s6d

The veiling glare for a high fidelity display system shonot change the contrast of a target pattern by more thanwith and without a bright surrounding. Thus, the luminafrom veiling glare should not be more than 25% of the mmum luminance for the normal operating settings of theplay. Since the ratio of the maximum luminance to the mmum luminance should be about 250, this implies a gratio of 1000, which is typical of measurements madetransilluminated film. However, the recommended testtern presents a scene with significantly more veiling glarthe target region than is encountered in medical imascenes. Though not as strict criteria which may noachievable by certain display technologies, TG18 recmends a glare ratio greater than 400 and 150 for primarysecondary display devices, respectively.27,28

III.H. Display Chromaticity

In display devices, chromaticity refers to the intrinmagnitude and uniformity of color tint of the device whdisplaying a monochrome image. In monochrome Ccolor tint is dictated by the phosphor type, and can vslightly from monitor to monitor. In LCDs, color tint is ditated by the color temperature of the backlight. Color tinusually considered a preference issue. However, it cancause of distraction, especially in multiple monitor worktions where the color tints are mismatched.

III.H.1. Visual Evaluation

The visual assessment of color uniformity is performusing the TG18-UN80 test pattern. The pattern is displaon all the display devices associated with a workstation,the relative color uniformity of the displayed pattern acrthe display area of each display device and across diffdisplay devices is discerned. No significantly perceivcolor differences should be present among display deand across the display area of each device for primarydevices. No requirements are specified for secondarydisplays.

III.H.2. Quantitative Evaluation

The TG18-UNL80 test pattern is displayed on all theplay devices associated with a workstation. A colorimetethen used to measure thesu8 ,v8d color coordinates at thcenter and at the four corners of the display area ofdisplay device, and these coordinates averaged to prodmeansu8 ,v8d chromaticity measurement for the displayvice. The measurements on all display devices are uscompute the color uniformity index as the maximumtance insu8 ,v8d space between any possible pair of ave

2 2 1/2
su8 ,v8d points usingD=ssu18−u28d +sv18−v28d d . If the colo-

-

-d

,

a

t

sss

ha

o

rimeter used outputs the color coordinate in the oldersx,ydspace, the values can be converted tosu8 ,v8d space using thfollowing transformations:

u8 = 4x/s− 2x + 12y + 3d, v8 = 9y/s− 2x + 12y + 3d; s7d

or

x = 27u8/s18u8 − 48v8 + 36d,

y = 12v8/s18u8 − 48v8 + 36d. s8d

Based on clinical experience, a color uniformity paramet0.01 or less is necessary to assure acceptable color maof primary class grayscale display devices of a workstati29

The distance between any pair of color coordinates athe display area of each device should also not exceelimit. No quantitative requirements are specified for secary class displays.

III.I. Miscellaneous Tests

In addition to the primary display attributes descriabove, there are a number of secondary attributes thaneed to be addressed in a full display performance evtion. Those include video artifacts, moiré artifacts, colortifacts, physical defects, flicker, and electronic crossBrief descriptions and assessment methods for these cteristics are outlined in the TG18 report.

III.J. Overall Evaluations

In addition to the testing of a display device for a speperformance characteristic, the overall quality of a syscan be assessed using a comprehensive visual/quantapproach. Overall assessment can be based on anyTG18-recommended multipurpose test patterns. Each pshould be displayed with one display pixel representingimage pixel and examined from a viewing distance of 30The findings can be correlated with the results of morecused testing methods specified above and serve as afor quality control assessments. The frequency of sucevaluation is discussed in Sec. 6 of the full report. Evations based on TG18-QC and TG18 anatomical patternoutlined below.

III.J.1. Evaluations using TG18-QC Pattern

The appearance of the elements in the TG18-QC testern fFig. 1sadg can be used to assess the overall performof a display system. The following are recommended:

1. General image quality and artifacts: Evaluate the ovappearance of the pattern. Note any non-uniformitieartifacts, especially at black-to-white and white-to-bltransitions. Verify that the ramp bars appear continuwithout any contour lines.

2. Geometric distortion: Verify that the borders and linethe pattern are visible and straight and that the paappears to be centered in the active area of the di
device. If desired, measure any distortions.

allureandDF.

uatets aout

ano thaircor

umi-nta

geportose

eriaThepond-horateIII,igh-ante re-

inghain.it isppro-per-alityl re-inescificrpose

s

sels


3. Luminance, reflection, noise, and glare: Verify that16 luminance patches are distinctly visible. Meastheir luminance using a luminance meter if desired,evaluate the results in comparison to the DICOM GSVerify that the 5% and 95% patches are visible. Evalthe appearance of low contrast letters and the targethe corners of all luminance patches with and withambient lighting.

4. Resolution: Evaluate the Cx patterns at the centercorners of the pattern and grade them compared treference score. Also verify the visibility of the line-ppatterns at the Nyquist frequency at the center andners of the pattern, and if desired, measure the lnance difference between the vertical and horizohigh-modulation patterns.

III.J.2. Evaluations using Anatomical Images

A radiologist should evaluate the overall clinical imaquality of the display using patient images. The TG18 resuggests four specific anatomical images for this purpTG18-CH, TG18-KN, TG18-MM1, and TG18-MM2fFigs.

TABLE VIII. Criteria for evaluating the TG18 anatomical images.

Test pattern Evaluation criteria

TG18-CH Degree of difficultyOverall contrastOverall sharpnessSymmetrical reprodbetween the medialMedial borders of thReproduction of theVisually sharp reproSharp reproductionSharp reproductionSharp reproductionVisibility of the retroVisibility of subdiaphVisibility of the spineVisibility of small deVisibility of linear an

TG18-KN Degree of difficultyOverall contrastOverall sharpnessReproduction of trabReproduction of bon

TG18-MM1 and TG18-MM2 Degree of difficultyOverall contrast andOverall sharpnesssnoSharp appearance oStructure of the clipAppearance and visVisibility of structure

1srd–sudg. These correspond to a chest radiograph, a kne


t

de

-

l

:

radiograph, and two digital mammograms. Clinical critfor evaluating these images are given in Table VIII.images may be scored according to these criteria corresing to the different image features. The radiologist wwishes to evaluate his/her display should independentlythe image features according to the criteria in Table Vthen compare their ratings to those obtained with a hquality transilluminated film print of the patterns. Significdiscrepancies need to be brought to the attention of thsponsible medical physicist or service engineer.

IV. CONCLUSIONS

Electronic display is a key component of medical imagsystems as it serves as the final element of the imaging cDue to hardware variability and degradation over time,important to assure that a medical display system is apriate for the intended medical application and that itsformance is stable over time. Acceptance testing and qucontrol testing of medical display devices are essentiaquirements for high-quality medical practice. The guidelestablished by the AAPM Task Group 18 delineate spetesting procedures and acceptance criteria for that pu

xam

n of the thorax, as shown by the central position of a spinous procesof the clavicles

apulaele rib cage above the diaphragmon of the vascular pattern of the lungs, particularly the peripheral vese trachea and proximal bronchie borders of the heart and the aortae diaphragmac lung and the mediastinumatic featuresugh the heart shadow

in the whole lung, including the retrocardiac areasicular details out to the lung periphery

xam

ar detaild soft tissue

xamhtnessdoper’s ligamentsthe presence of the gap at its apexsTG18-MM1 onlydof subtle microcalcificationssTG18-MM1 onlyd

the margins of the breastsTG18-MM1 onlyd

for e

uctioendse scwho

ductiof thof thof thcardiragmthro

tailsd ret

for e

eculy an

for ebrigblurf Coandibilitys at

ethat can be readily implemented in a clinical setting.

yan,ia deven-

agid,logy

en,ncyphysology

iol-

RT

soft-

chis,”CA,

phorplays

and

mu-

n of

tain

ea-

e,

cinelec-uitetp://

pur-ocu-

http://

, K.n, M.f, J.dicalts inn,

y,ston,

s on

lity

g-s

.ects

e oning

du-

-to-by a

RT

4tter-

nce

eou,scale


adElectronic mail: [email protected]. J. Ackerman, J. N. Gitlin, R. W. Gayler, C. D. Flagle, and R. N. Br“Receiver operating characteristic analysis of fracture and pneumontection: comparison of laser-digitized workstation images and contional analog radiographs,” Radiology186, 263–268s1993d.

2W. W. Scott, J. E. Rosenbaum, S. J. Ackerman, R. L. Reichle, D. MJ. C. Weller, and J. N. Gitlin, “Subtle orthopedic fractures: Teleradioworkstation versus film interpretation,” Radiology187, 811–815s1993d.

3W. W. Scott, D. A. Bluemke, W. K. Mysko, G. E. R. Weller, G. D. KelR. L. Reichle, J. C. Weller, and J. N. Gitlin, “Interpretation of emergedepartment radiographs by radiologists and emergency medicinecians: teleradiology workstation versus radiograph readings,” Radi195, 223–229s1995d.

4“AAPM, The role of the clinical medical physicist in diagnostic radogy,” Report of the AAPM Task Group No. 2sAIP, New York, 1994d.

5E. Muka, H. Blume, and S. Daly, “Display of medical images on Csoft-copy displays: a tutorial,” Proc. SPIE2431, 341–359s1995d.

6E. Senol and E. Muka, “Spatial frequency characteristics of CRTcopy displays,” Proc. SPIE2431, 302–315s1995d.

7E. F. Kelley, G. R. Jones, P. A. Boynton, M. D. Grote, and D. J. Be“A survey of the components of display measurement standards,SIDDigest sSociety for Information Display Headquarters, San Jose,1995d, pp. 637–640.

8E. Muka, T. Mertelmeier, R. M. Slone, and E. Senol, “Impact of phosluminescence noise on the specification of high-resolution CRT disfor medical imaging,” Proc. SPIE,3031, 210–221s1997d.

9H. Roehrig, H. Blume, T. L. Ji, and M. Browne, “Performance testsquality control of cathode ray tube displays,” J. Digit Imaging3, 134–145 s1990d.

10J. E. Gray, “Use of SMPTE test pattern in picture archiving and comnication systems,” J. Digit Imaging5, 54–58s1992d.

11R. D. Nawfel, K. H. Chan, D. J. Wagenaar, and P. F. Judy, “Evaluatiovideo gray-scale display,” Med. Phys.19, 561–567s1992d.

12D. A. Reimann, M. J. Flynn, and J. J. Ciarelli, “A system to mainperceptually linear networked display devices,” Proc. SPIE2431, 316–326 s1995d.

13M. P. Eckert and D. P. Chakraborty, “Video display quality control msurements for PACS,” Proc. SPIE2431, 328–340s1995d.

14H. Kato, “Hard- and soft-copy image quality,”RSNA Categorical Coursin PhysicssRadiological Society of North AmericasRSNAd PublicationCategorical Course Syllabus, Oak Brook, IL, 1995d, pp. 243–258.

15NEMA PS 3.14-2000, “Digital imaging and communications in medisDICOMd Part 14: Grayscale display standard function,” National Etrical Manufacturers Association, NEMA, 1300 North 17th Street, S1847, Rosslyn, VA 22209, http://www.nema.org/ Draft viewable at ht


-

i-

medical.nema.org/dicom/2000/draft/00_14DR.PDF, Available forchase in whole or in parts from Global Sales, Global Engineering Dments, 15 Inverness Way, East Englewood, Colorado, 80112;global.ihs.com/

16E. Samei, A. Badano, D. Chakraborty, K. Compton, C. CorneliusCorrigan, M. J. Flynn, B. Hemminger, N. Hangiandreou, J. JohnsoMoxley, W. Pavlicek, H. Roehrig, L. Rutz, J. Shepard, R. UzenofWang, and C. Willis, “Assessment of display performance for meimaging systems,” Report of the American Association of PhysicisMedicinesAAPMd Task Group 18sMedical Physics Publishing, MadisoWI, 2005d, AAPM On-line Report No. 03.

17American College of RadiologysACRd, ACR Standard for Teleradiolog1999. American College of Radiology, 1891 Preston White Drive, ReVA 20191-4397; http://www.acr.org/

18P. G. J. Barten,Contrast Sensitivity of the Human Eye and Its EffectImage QualitysSPIE Press, Bellingham, WA, 1999d.

19M. J. Flynn, J. Kanicki, A. Badano, and W. R. Eyler, “High fideelectronic display of digital radiographs,” Radiographics19, 1653–1669s1999d.

20E. Samei, in “Digital mammography displays,”Advances in Breast Imaing: Physics, Technology, andClinical Applications, edited by A. Karellaand M. L. GigersRadiological Society of North AmericasRSNAd Publi-cation, Categorical Course Syllabus, Oak Brook, IL, 2004d, pp. 135–143

21H. Blume, A. M. K. Ho, F. Stevens, and P. M. Steven, “Practical aspof gray-scale calibration of display systems,” Proc. SPIE,4323, 28–41s2001d.

22E. Samei and S. L. Wright, “The effect of viewing angle responsDICOM compliance of LCD displays,” Proc. SPIE Medical Imag5371, 170–177s2004d.

23E. Samei and M. J. Flynn, “A method for in-field evaluation of the molation transfer function of electronic display devices,” Proc. SPIE4319,599–607s2001d.

24H. Roehrig, T. L. Ji, M. Browne, W. J. Dallas, and H. Blume, “Signalnoise ratio and maximum information content of images displayedCRT,” Proc. SPIE1232, 115–133s1990d.

25H. Roehrig, H. Blume, T. L. Ji, and M. K. Sundareshan, “Noise of Cdisplay systems,” Proc. SPIE1897, 232–245s1993d.

26A. Rose,Vision, Human and ElectronicsPlenum Press, New York, 197d.27M. J. Flynn and A. Badano, “Image quality degradation by light sca

ing in display devices,” J. Digit Imaging12, 50–59s1999d.28A. Badano, M. J. Flynn, and J. Kanicki, “Accurate small-spot lumina

measurements in display devices,” Displays23, 177–182s2002d.29K. A. Fetterly, S. N. Bernatz, D. S. Groth, and N. J. Hangiandr

“Quantitative color measurement for the characterization of grayPACS CRTs,” Radiology209(P), 320 s1998d.

Date post:	19-May-2020
Category:	Documents
Upload:	others
View:	6 times
Download:	0 times

Assessment of display performance for medical imaging...

Documents