Fully automated method for lung emphysema
quantification for Multidetector CT images
Irene Mayorga-Ruiz1,David García-Juan1, Ángel Alberich-Bayarri1,2, Fabio
García-Castro1, Luis Martí-Bonmatí1,2
1 QUIBIM S.L., Valencia, Spain
2 La Fe Health Research Institute,Valencia, Spain
• Introduction
• Purpose
• Materials and Methods
• Results
• Conclusions
Outline
Introduction
Lung emphysema is defined as an abnormal and persistent increment of the distal
aerial spaces to the terminal bronchiole, it could be accompanied with lung
parenchyma walls destruction and without fibrosis.
Purpose
Lung emphysema is considered with chronic bronchitis part of the Pulmonary Obstructive
Chronic Diseases (COPD).
Change of the paradigm between the qualitative and the quantitative information of
the disease.
Development of zero-click automatic algorithm able to segment, quantify and
characterized lung emphysema areas, lung parenchyma, blood vessels and lung
parenchyma densities.
Materials and Methods
Image Modality: 39 standard MDCT scans
- 22 male and 17 female
Acquisition Characteristics
- Voltage:120 kVp
- Current: 250 mA
- Slice thickness ≤ 2mm
- Pixel Size ≤ 1mm
Materials and Methods
Materials and Methods
Lung Emphysema Quantification
Fixed thresholding (-950 HU)
Adaptive thresholding (QUIBIM Patent)
Materials and Methods
• Quantification
1. Percentage:
Ratio between the total number of voxels of the ROI and the total number of the mask [%]
2. Volume Quantification:
Product between the number of voxels of the ROI and the dimensions of a slice [ml]
Results
Mean
VolumeMale Female
Right Lung 3209,78 ml 2515,07 ml
Left Lung 2979,18 ml 2136,17 ml
Total 6188,96 ml 5187,50 ml
• Algorithm execution takes 30-45 min on average depending study size
• Right lung volume is 8,7% and 9,5% greater than left lung volume for male
and female respectively
• Lung emphysema quantification by adaptive thresholding is 50% smaller on
average than lung emphysema quantification by fixed Thresholding
-10
0
10
20
30
40
50
60
0 5 10 15 20 25 30 35 40 45
TotalPercentage[%]
Pa entID
EmphysemaQuan fica onMethodsCompara ve
FixedThresholding
Adap veThresholding
Results
• 29 patients were diagnosed of
emphysema caused by external
agents. Emphysema was mainly
detected in the lung apex.
• 2 patients present alfa-1-antitripsine
deficient being the emphysema
detected mainly on lungs base.
• 8 patients were unspecific due to
percentage of emphysema below 5 %.
http://www.quibim.com
Emphysema
Emphysema
QUIBIM PATENTED Adaptive Low Density Thresholding*
Legend
TA Patient Sex
Birthdate
Patient ID
Study Date
Modality CT
M
Imaging Center Patient Name
Study Description
26/09/2013
Left Lung
1233
1368
44.85 754
Whole Lung
Right Lung
903140 3609
1019
83455.07
0.490.39
17.32
15.11 20.30 99
9.69 1542
Upper Third
Middle Third
Lower Third
VolumesLeft Lung Right Lung
Whole Lung
Upper Third
Middle Third
Lower Third
VesselsAutomatic Vessel Segmentation
*Patent 201.331.295. Segmentation method by adaptive thresholding for the obtention of air reference values slice-by-slice in CT imaging studies.
Total Emphysema Percentage: Total Lung Volume:17.88 % 6749 mL
Emphysema
Percentage(%)
Volume (mL)Percentage
(%)Volume (mL)
4
132
338
474
6
267
459
732
VesselVolume (mL)
Volume (mL) Volume (mL)Vessel
Volume (mL)
12
33
45
12
36
51
Healthy Percentage < 5%
Data from this quantification report should be considered as the results of research with an evidence level 2 (Centre for Evidence-based Medicine) in phase of clinical approval.
QUIBIM S.L. - Quantitative Imaging Biomarkers in Medicine. Avenida Fernando Abril Martorell 106, Torre A, Biopolo La Fe, Valencia (SPAIN)
Conclusions
• Relation between the smoker status and the presence of emphysema
exists
• Adaptive thresholding is able to perform a better characterization of
emphysema due to image-threshold specificity
• Lung emphysema etiology could be inferred due to emphysema
quantification by lung thirds.
Automated lung emphysema quantification can be used for the diagnosis
and follow up of COPD.
Luis Martí Bonmatí – MD, PhD. GIBI PI and QUIBIMFounder
Ángel Alberich-Bayarri – PhD. GIBI Director andQUIBIMCEO
QUIBIM StaffFabio García Castro - M.ScRafa Hernández Navarro - B.ScDavid García - M.ScEncarna Sánchez - M.ScKatherineWilisch Ramírez - M.ScIrene Mayorga Ruiz - M.ScBelén Fos Guarinos - Internship StudentAna Jiménez Pastor - Internship Student
GIBI230 StaffEnrique Ruiz Martínez –M.ScAmadeoTen Esteve –M.ScAna Penadés - Adm.AlfredoTorregrosa - Internship StudentCarlos Moya - Internship Student
Team
Chief Scientific Officer Chief Technology Officer Back-End Development of Imaging Biomarkers Chief Marketing Officer
Coordinator and CEO Support
MS Biomedical Engineering
Imaging Study Coordinator
Chief FinancialOfficer
Image AnalysisScientist Artificial Intelligence Artificial Intelligence
Imaging Biomarker Developer
Imaging Biomarker Developer