SHORT COMMUNICATION

Utility of an open-source DICOM viewer software (OsiriX)to assess pulmonary fibrosis in systemic sclerosis: preliminaryresults

Alarico Ariani • Marina Carotti • Marwin Gutierrez •

Elisabetta Bichisecchi • Walter Grassi •

Gian Marco Giuseppetti • Fausto Salaffi

Received: 2 August 2012 / Accepted: 31 July 2013

� Springer-Verlag Berlin Heidelberg 2013

Abstract To investigate the utility of an open-source

Digital Imaging and Communication in Medicine viewer

software—OsiriX—to assess pulmonary fibrosis (PF) in

patients with systemic sclerosis (SSc). Chest high-resolu-

tion computed tomography (HRCT) examinations obtained

from 10 patients with diagnosis of SSc were analysed by

two radiologists adopting a standard semiquantitative

scoring for PF. Pulmonary involvement was evaluated in

three sections (superior, middle and inferior). For the

assessment of the extension of PF, the adopted semiquan-

titative HRCT score ranged from 0 to 3 (0 = absence of

PF; 1 = 1–20 % of lung section involvement;

2 = 21–40 % of lung section involvement; 3 = 41–100 %

of lung section involvement). Further, a quantitative

assessment (i.e. parameters of distribution of lung attenu-

ation such as kurtosis and mean lung attenuation) of PF

was independently performed on the same sections by a

rheumatologist, independently and blinded to radiologists’

scoring, using OsiriX. The results obtained were compared

with those of HRCT semiquantitative analysis. Intra-reader

reliability of HRCT findings and feasibility of OsiriX

quantitative segmentation was recorded. A significant

association between the median values of kurtosis by both

the quantitative OsiriX assessment and the HRCT semi-

quantitative analysis was found (p \ 0.0001). Moreover,

kurtosis correlated significantly with the mean lung

attenuation (Spearman’s rho = 0.885; p = 0.0001). An

excellent intra-reader reliability of HRCT findings among

both readers was obtained. A significant difference

between the mean time spent on the OsiriX quantitative

analysis (mean 1.85 ± SD 1.3 min) and the mean time

spent by the radiologist for the HRCT semiquantitative

assessment (mean 8.5 ± SD 4.5 min, p \ 0.00001) was

noted. The study provides the new working hypothesis that

OsiriX may be a useful and feasible tool to achieve a

quantitative evaluation of PF in SSc patients.

Keywords Pulmonary fibrosis � Systemic sclerosis �DICOM � HRCT � kurtosis � OsiriX

Introduction

Systemic sclerosis (SSc) is a systemic autoimmune disease

characterized by a micro- and macro-vascular damage [1,

2]. Pulmonary fibrosis (PF) is a frequent manifestation in

SSc. Its severity may vary considerably depending on the

underlying disease, and frequently it can be the cause of

death of these patients [3–5].

Currently, chest high-resolution computed tomography

(HRCT) is considered the most accurate not invasive

imaging method for PF assessment. Both severity and

extent of PF are usually estimated by the adoption of

semiquantitative scoring methods [6, 7]. However, the

correct evaluation of the semiquantitative scores could

represent a problem for the inexperienced physicians since

there is a wide interobserver variability even among expert

radiologists [8].

Recently, special softwares providing an automatic lung

parenchyma identification and a quantitative assessment of

PF have been developed [9, 10]. In particular, they allow to

A. Ariani � M. Gutierrez (&) � W. Grassi � F. Salaffi

Clinica Reumatologica, Dipartimento di Scienze Cliniche e

Molecolari, Universita Politecnica delle Marche, Ospedale ‘‘A.

Murri’’, Via dei Colli, 52 Jesi, 60035 Ancona, Italy

e-mail: dr.gmarwin@gmail.com

M. Carotti � E. Bichisecchi � G. M. Giuseppetti

S. O. D. Radiologia Clinica, Dipartimento di Scienze

Radiologiche, Ospedali Riuniti, Ancona, Italy

123

Rheumatol Int

DOI 10.1007/s00296-013-2845-6

obtain a more detailed information in regard to the severity

of PF since they permit an accurate evaluation of lung

attenuation distribution parameters such as kurtosis, sym-

metry and mean lung attenuation (MLA) [11]. Unfortu-

nately, the lack of standardization and the relative high cost

of the licences limit their use in clinical practice.

The growing physicians’ demand about visualization

and management of medical digital images [i.e. the Digital

Imaging and Communication in Medicine (DICOM) files]

has urged to the development of open-source (and free of

charge) softwares which allow to view DICOMs and to

perform very complex imaging analysis (‘‘post-process-

ing’’). OsiriX is one of the most popular DICOM viewers

created for Mac OSX platform [12]. Some OsiriX tools,

such as the possibility to perform multimodality and mul-

tidimensional post-processing analysis and reworking on

3D data obtained from chest HRCT, open up a new

research area to attain PF assessment. The main aim of the

present study was to investigate the utility of OsiriX for the

rheumatologist to assess pulmonary fibrosis in SSc. The

study was also planned to compare the PF assessment

carried out by OsiriX with respect to a standard semi-

quantitative method based on lung findings performed by a

radiologist.

Patients and methods

Patients

Ten patients with diagnosis of SSc (1 male and 9 females)

were included in the present study. The diagnosis was

made according to the American College of Rheumatology

(ACR) classification criteria [13]. Mean ± SD age was

56 ± 7 years (range 45–61), and the mean ± SD disease

duration was 10 ± 2 years (range 7–11).

Inclusion criteria included diagnosis of SSc with docu-

mented PF, [ 18 years, chest HRCT performed no longer

than 6 months prior to the beginning of this study. Patients

who refer a history of pulmonary neoplasia or other causes

of interstitial fluid such as hearth failure, diastolic dys-

function, asthma or pulmonary oedema were excluded of

the study.

All patients were attending the outpatient and inpatient

clinics of the Rheumatology Department of the Universita

Politecnica delle Marche (Ancona, Italy).

Study design

Chest HRCT and OsiriX analysis were carried out at the

Radiology and Rheumatology Departments of the Univer-

sita Politecnica delle Marche, Ancona, Italy, respectively.

All HRCT examinations were reviewed by a radiologist

expert on the HRCT interstitial lung disease (MC), blinded

to the clinical data. A second experienced thoracic radiol-

ogist (EB), blinded with respect to the first one and HRCT

findings, scored the PF in order to assess the inter-reader

agreement. Prior to the study, the investigators reached a

consensus on the PF HRCT interpretation.

Afterwards, on the same pulmonary sections, a quanti-

tative assessment of PF was performed independently and

blinded to radiologists’ scoring, by a rheumatologist (AA),

using OsiriX. The results obtained were compared with

those of HRCT semiquantitative analysis.

In order to determine the feasibility, the time spent for

both semiquantitative HRCT analysis and quantitative

segmentation by OsiriX was recorded.

The study was conducted according to the Declaration

of Helsinki and local regulations. The institutional review

board approved the study and informed consent was

obtained from all patients.

Chest HRCT assessment

All HRCT examinations were performed by standard pro-

tocol using a CT 64 GE light Speed VCT power scanner

with a rotation tube scanning time of 0.65 s. Scans were

obtained at full inspiration from the apex to the lung base

with the patients in the supine position, at 120 kV and

300 mAs, with a slice thickness of 1.25 mm and slice

spacing of 7 mm. HRCT assessment do not include the use

of contrast media agents.

HRCT pulmonary involvement was evaluated in three

pulmonary sections for each patient (obtaining 30 pul-

monary segmentations) according to Sverzellati et al. [9]

[8]. The sections considered were superior (origin of the

large vessels), middle (carina of trachea) and inferior (right

inferior pulmonary vein).

The following elementary lesions were evaluated in

each section: ground glass opacity, interface irregularities

between peripheral pleura and lung parenchyma, septal

lines (due to interlobular septa and subpleural thickening),

honeycombing and thin-walled subpleural cysts.

For the assessment of the extent of PF, a semiquantita-

tive HRCT, score ranged from 0 to 3 (0 = absence of PF;

1 = 1–20 % of lung section involvement; 2 = 21–40 % of

lung section involvement; 3 = 41–100 % of lung section

involvement), was adopted.

OsiriX assessment

HRCT images were reconstructed and analysed by OsiriX,

a DICOM viewer software (OsiriX version 3.9; Apple

Computer) on a Mac Mini (2.4 GHz Intel Core 2 Duo

Rheumatol Int

123

Desktop Computer, 4 GB random-access memory; Apple

Computer, Cupertino, CA, USA) running Mac Operating

System X 10.5.

After inserting the CD-ROM or DVD containing HRCT

scan data in the drive, the DICOM data were automatically

extracted from the disc by OsiriX. The DICOM data were

stored in the OsiriX using the ‘‘Copy linked files to Data-

base folder’’ under ‘‘file’’ in the OsiriX dropdown menu.

The same three pulmonary sections evaluated previously

by the radiologist were assessed. For each section, a

semiautomatic lung parenchymal segmentation was per-

formed in order to obtain the corresponding histograms of

attenuation; then, descriptive parameters of the distribution

such as MLA, kurtosis and symmetry were calculated.

Figure 1 shows the representative sequences of the OsiriX

segmentation process.

Statistical analysis

Statistical analysis was performed using MedCalc (version

12.0 for Windows XP, Belgium).

Segmentation data did not follow a Gaussian distribution,

and therefore, we did not use parametric tests. Thus, the

kurtosis, symmetry and MLA values were represented as

median with the relative interquartile. Spearman’s rank order

test was used to evaluate the correlation between kurtosis and

MLA. Additionally, we determined the relationship between

kurtosis and MLA, and the different scores of HRTC, using

Kruskall–Wallis and Wilcoxon tests to assess the level of

significance of the different severity categories. The intra-

reader HRCT agreement between the two radiologists has

been calculated by weighted kappa statistic. A kappa value of

0–0.20 was considered poor, 0.21–0.40 fair, 0.41–0.60

moderate, 0.61–0.80 good and 0.81–1.00 excellent. Feasi-

bility of OsiriX was estimated by comparing the time

spent in quantitative analysis with respect to HRCT semi-

quantitative analysis by the independent samples t test. A

p value\0.05 was considered statistically significant.

Results

Figure 2 shows representative examples of semiquantita-

tive analysis with the relative distribution diagrams

(including kurtosis, symmetry and MLA) obtained using

OsiriX.

Fig. 1 Segmentation algorithm. The basic steps for the segmentation

of lung parenchyma of the section shown in a are as follows: selection

of the command to generate the Region of Interest (b), insertion of

parameters for the detection of lung parenchyma (c), applying the

command of ‘‘Brush ROI’’ to obtain a complete segmentation (d)

Rheumatol Int

123

On categorizing patients into different grades of sever-

ity, with respect to HRCT score, we found a significant

association between the median values of kurtosis of each

lung section assessed by OsiriX and HRCT semiquantita-

tive score (p \ 0.0001). In particular, kurtosis was signif-

icantly different in patients with low HRCT score (0 or 1)

compared with patients with higher scores (2 and 3)

(p \ 0.01) (Fig. 3a).

The kurtosis median value, measured by computer

analysis, was significantly higher in the apical lung sec-

tions, compared with the lower ones (p \ 0.05) (Fig. 3b).

Similarly, low score patients (0 or 1) and high score

patients (2 and 3) have significantly different

(p \ 0.01) MLA median values, measured over the three

lung sections (Fig. 3c).

A significant correlation between kurtosis and MLA was

found (rho = 0.885 p = 0.0001) (Fig. 3d).

The global kappa value for the intra-reader reliability of

HRCT findings reached by both radiologists was 0.92.

With respect to the feasibility, a significant difference

between the time spent by the rheumatologist for the

OsiriX quantitative analysis, including three pulmonary

sections, and the mean time spent by the radiologist for the

HRCT semiquantitative assessment was found (mean

1.85 ± SD 1.3 min vs mean 8.5 ± SD 4.5 min,

p \ 0.00001, respectively).

Discussion

To the best of our knowledge, this is the first study pro-

viding evidence in favour of the utility of this ‘‘novel’’

open-source DICOM viewer software (OsiriX) as an

adjunct method to assess PF in patients with SSc.

PF plays an unfavourable prognostic factor of life in

SSc patients. Chest HRCT allows an accurate assessment

of both severity and extension of PF; however, it requires

both adequate knowledge of semiquantitative methods and

wide experience to score PF. Moreover, the reported con-

troversial data about its reproducibility makes its system-

atic application difficult in clinical trials [14, 15]. In order

to overcome this barrier, sophisticated automatic algo-

rithms to identify accurately the PF areas (segmentation)

have been developed. These segmentations permit to obtain

histograms of distributions of the lung attenuation which

facilitate the identification and quantification of PF through

the analysis of the descriptive parameters (mode, mean

value, kurtosis and skewness) [16, 17]. Additionally, these

descriptive parameters demonstrated to be higher repro-

ducible than the PF semiquantitative assessment, usually

performed by the radiologist [18]. Although these soft-

wares appear to be a really promising tool for the assess-

ment of PF, their application in clinical practice still

remains limited, due to the lack of standardization and the

relatively high cost of licence [19].

In our study we used an open-source DICOM viewer

software, OsiriX, to quantify PF in SSc patients. This

opens up an interesting window of research focused on

its use as a useful additional tool to study pulmonary

changes in patients with SSc. OsiriX offers peculiar

characteristics for the pulmonary assessment: it is a free

of charge software and no intensive computer training

is mandatory to become able to perform lung segmen-

tation and to obtain the corresponding distribution his-

togram (essential for kurtosis, MLA and skewness

calculation).

From an analysis of our results, the following consid-

erations can be formulated. First, the quantitative OsiriX

analysis showed a high agreement with the assessment of

PF compared with the semiquantitative HRCT analysis

performed by experienced radiologists in lung fibrosis. The

sections with the lowest kurtosis and highest MLA had a

semiquantitative score indicative of an extensive pulmon-

ary involvement.

Fig. 2 Examples of segmentation of lung and its distribution

histogram. On the left the lower section of the lungs of three patients

with different semiquantitative scoring on a Likert scale (LS)

indicated in the upper right. The green area corresponds to the

region that the OsiriX algorithm recognizes as lung parenchyma. To

the right of each section, there is the corresponding distribution

histogram of lung attenuation values and the consequent descriptive

parameters: kurtosis (K), symmetry (S), mean lung attenuation

(MLA) expressed in Hounsfield unit. Lung attenuation values assume

approximately a bimodal distribution. Note that the ‘‘highest peak’’

(mode’s main distribution) tends to be progressively closer to the HU

value 0 (to the right along the horizontal axis) and the base tends to

expand making the ‘‘minor peak’’ hardly intelligible

Rheumatol Int

123

Second, OsiriX demonstrated to be reliable in the

assessment of PF since, as shown by histological studies,

it confirms that PF tends to increase progressively with an

apical–basal gradient in SSc patients [7]. In fact, we

observed that the basal lung areas showed a lower degree

of kurtosis than the apical ones. Third, the values of MLA

obtained are in line with those reported previously in other

studies [8, 11, 16]. Fourth, the mean time spent to perform

an OsiriX quantitative analysis for each patient was much

less than the semiquantitative assessment of HRCT per-

formed by the radiologist’s lecture.

Taking these observations into account, the results of

our study seem to be encouraging since they induce to

consider OsiriX a useful tool for the assessment of PF.

Despite these aspects, we are aware that there are limita-

tions in our study. First, a correct analysis by OsiriX

depends directly on a correct HRCT data acquisition.

Second, the low number of enrolled patients does not

permit an accurate evaluation in terms of sensitivity and

specificity which could more strongly support these data.

Third, OsiriX segmentation algorithm does not consider

completely the PF at the basal peripheral regions (usually

in late stage disease patients). This flaw probably could be

explained by the fact that the PF is quite difficult to dis-

tinguish (in terms of density) from surrounding tissues

[20]. However, the operator can manually complete the

segmentation, indicating the neglected areas. Fourth,

OsiriX makes a quantitative assessment of pulmonary

fibrosis but does not allow the automatic identification of

elementary lesions. Moreover, we have evaluated only

three pulmonary sections, not the entire lung. Note, how-

ever, that some authors [21] have demonstrated the corre-

lation between the extent of pulmonary fibrosis in three

sections and the PF width in the entire lung. Finally,

although OsiriX is a free of charge software, it needs to be

installed on a Mac OsX platform.

Despite the above limitations, we believe that OsiriX

can be used as an adjunct method in the assessment of PF.

Besides, it can play a relevant role for screening purposes

aiming towards the early identification of SSc patients that

require a chest HRCT.

In conclusion, our pilot study provides the new working

hypothesis that OsiriX DICOM viewer software may be a

useful post-processing tool to achieve a quantitative eval-

uation of PF in SSc patients. Nevertheless, additional

investigations on larger series of cohorts about the whole

lung, studying sensitivity and specificity, may be useful to

more strongly support these data.

Conflict of interest The authors declare that they have no conflict

of interest.

Glossary

Open source Computer software available in

source code form: the source

code and certain other rights

normally reserved for copyright

holders are provided under a free

Fig. 3 a MLA stratified

according to the

semiquantitative evaluation.

b Kurtosis stratified according

to pulmonary section. c Kurtosis

stratified according to the

semiquantitative evaluation.

d Correlation between kurtosis

and MLA. MLA mean lung

attenuation

Rheumatol Int

123

software licence that permits

users to study, change, improve

and at times also to distribute the

software

DICOM Digital Imaging and COmmu-

nications in Medicine is a standard

for handling, storing, printing and

transmitting information in medical

imaging. It includes a reproduction

of the original uncompressed image

and a series of additional data

(called metadata) that are useful

for medical interpretation (such as

patient demographics, spatial

resolution, distance between the

planes of acquisition, radiation

absorbed)

Attenuation Value expressed in units Hounsfield

(HU) indicative of the X-rays

absorption degree that pass

through a single spatial unit (pixel)

of the lung examined with the CT

Distribution histogram Histogram showing the

frequency with a given value of

lung attenuation occurs in a given

set of CT image pixels

MLA Mean lung attenuation: arithmetic

mean of all lung attenuation values

of a single section

Kurtosis Distribution parameter which

gives an account of the

similarity between the histogram

distribution ‘‘shape’’ with the

Gaussian curve pattern. High

values indicate a departure from

normal distribution

Skewness Distribution parameter which gives

an account of how the histogram

distribution is symmetrical about a

central value. High values indicate

an asymmetry of the distribution

References

1. Steen VD, Conte C, Owens GR, Medsger TA (1994) Severe

restrictive lung disease in systemic sclerosis. Arthritis Rheum

37:1283–1289

2. Karassa FB, Ioannidis JPA (2008) Mortality in systemic sclerosis.

Clin Exp Rheumatol 26:S85–S93

3. Goh NSL, Desai SR, Veeraraghavan S et al (2008) Interstitial lung

disease in systemic sclerosis: a simple staging system. Am J Respir

Crit Care Med 177:1248–1254. doi:10.1164/rccm.200706-877OC

4. Wells AU (2008) High-resolution computed tomography and

scleroderma lung disease. Rheumatology (Oxford) 47(Suppl

5):v59–v61. doi:10.1093/rheumatology/ken271

5. Goldin JG, Lynch DA, Strollo DC et al (2008) High-resolution

CT scan findings in patients with symptomatic scleroderma-

related interstitial lung disease. Chest 134:358–367. doi:10.1378/

chest.07-2444

6. Warrick JH, Bhalla M, Schabel SI, Silver RM (1991) High res-

olution computed tomography in early scleroderma lung disease.

J Rheumatol 18:1520–1528

7. Cozzi F, Chiesura Corona M, Rizzi M et al (2001) Lung fibrosis

quantified by HRCT in scleroderma patients with different dis-

ease forms and ANA specificities. Reumatismo 53:55–62

8. Zavaletta VA, Bartholmai BJ, Robb RA (2007) High resolution

multidetector CT-aided tissue analysis and quantification of lung

fibrosis. Acad Radiol 14:772–787. doi:10.1016/j.acra.2007.03.

009

9. Sverzellati N, Zompatori M, De Luca G et al (2005) Evaluation

of quantitative CT indexes in idiopathic interstitial pneumonitis

using a low-dose technique. Eur J Radiol 56:370–375. doi:10.

1016/j.ejrad.2005.05.012

10. Sumikawa H, Johkoh T, Yamamoto S et al (2006) Quantitative

analysis for computed tomography findings of various diffuse

lung diseases using volume histogram analysis. J Comput Assist

Tomogr 30:244–249

11. Best AC, Lynch AM, Bozic CM et al (2003) Quantitative CT

indexes in idiopathic pulmonary fibrosis: relationship with

physiologic impairment. Radiology 228:407–414. doi:10.1148/

radiol.2282020274

12. Rosset A, Spadola L, Ratib O (2004) OsiriX: an open-source

software for navigating in multidimensional DICOM images.

J Digit Imaging 17:205–216. doi:10.1007/s10278-004-1014-6

13. LeRoy EC, Black CM, Fleischmajer R et al (1988) Scleroderma

(systemic sclerosis): classification, subsets and pathogenesis.

J Rheumatol 15:202–205

14. Strange C, Seibold JR (2008) Scleroderma lung disease: ‘‘if you

don’t know where you are going, any road will take you there’’.

Am J Respir Crit Care Med 177:1178–1179. doi:10.1164/rccm.

200802-304ED

15. Collins CD, Wells AU, Hansell DM et al (1994) Observer vari-

ation in pattern type and extent of disease in fibrosing alveolitis

on thin section computed tomography and chest radiography.

Clin Radiol 49:236–240

16. Lynch DA (2007) Quantitative CT of fibrotic interstitial lung

disease. Chest 131:643–644. doi:10.1378/chest.06-2955

17. Sumikawa H, Johkoh T, Yamamoto S et al (2009) Computed

tomography values calculation and volume histogram analysis for

various computed tomographic patterns of diffuse lung diseases.

J Comput Assist Tomogr 33:731–738. doi:10.1097/RCT.

0b013e31818da65c

18. Camiciottoli G, Orlandi I, Bartolucci M et al (2007) Lung CT

densitometry in systemic sclerosis: correlation with lung func-

tion, exercise testing, and quality of life. Chest 131:672–681.

doi:10.1378/chest.06-1401

19. Kim HG, Tashkin DP, Clements PJ et al (2010) A computer-

aided diagnosis system for quantitative scoring of extent of lung

fibrosis in scleroderma patients. Clin Exp Rheumatol 28:S26–S35

20. Hoffman EA, Reinhardt JM, Sonka M et al (2003) Character-

ization of the interstitial lung diseases via density-based and

texture-based analysis of computed tomography images of lung

structure and function. Acad Radiol 10:1104–1118

21. Kazerooni EA, Martinez FJ, Flint A et al (1997) Thin-section CT

obtained at 10-mm increments versus limited three-level thin-

section CT for idiopathic pulmonary fibrosis: correlation with

pathologic scoring. AJR Am J Roentgenol 169:977–983

Rheumatol Int

123