Original investigationComputer-aided Classification of Interstitial Lung Diseases Via MDCT: 3D Adaptive Multiple Feature Method (3D AMFM)
Section snippets
Data
Subjects were selected from scans gathered via MDCT images through our National Institutes of Health–funded Biomedical Research Partnership (BRP) (HL-064368) seeking to establish an atlas of the normal human lung based on structural and functional measures derived from MDCT imaging. We are gathering cohorts of volunteers categorized as normal smokers, patients with emphysema, and patients with other ILD for comparison against the normal lung atlas. Studies were approved by the University of
Results
The optimal three parameters—1) the kernel type, 2) SVM type, and 3) ν—were linear, ν-SVC, and 0.14, respectively. Table 2, Table 3 show the confusion tables generated by the Bayesian and SVM methods for the 10-fold cross-validation method. In the confusion tables, each row represents a disease pattern; each column represents a category classified by the computer. For example, in Table 2, among all 287 NN samples classified by experts, 7 samples were classified as EMPH, 3 were classified as GG,
Discussion
The results of this study demonstrate that SVM and Bayesian methods are comparable in the classification of patterns associated with interstitial lung disease and both methods produce highly sensitive and specific discrimination amongst common radiologically defined tissue types. Sensitivity and specificity in detecting the difference between normal lung from “normal” nonsmokers and “normal” lung form smokers as shown in Figure 8, Figure 9 are lower than other tissue types, but encouragingly
Conclusion
We conclude that volumetric features including statistical, histogram, and model-based features can successfully differentiate between a simultaneous mixture of textures represented by emphysema, multiple interstitial lung disease patterns along with normal nonsmoker and normal smoker’s lungs. Additionally, the SVM and Bayesian methods are comparable in the classification of ILD parenchymal patterns. Our system is highly sensitive and specific in detecting early abnormalities both via a
References (32)
- et al.
Characterization of the interstitial lung diseases via density-based and texture-based analysis of computed tomography images of lung structure and function
Acad Radiol
(2003) - et al.
Analysis of x-ray radiographs is a more reliable descriptor of none loss than mineral content in a rat model of localized disuse induced by the clostridium botulinum toxin
Bone
(2001) - et al.
Artificial neural networks in chest radiographyapplication to the differential diagnosis of interstitial lung disease
Acad Radiol
(1999) - et al.
Asserta physician-in-the-loop content-based retrieval system for HRCT image databases
Computer Vision Image Understanding
(1999) - et al.
Computer-aided diagnosis and evaluation of lung disease
J Thorac Imaging
(2004) - et al.
Smoking-related interstitial lung diseasesa concise review
Eur Resp J
(2001) - et al.
Interstitial lung diseasesan epidemiological overview
Eur Resp J
(2001) - et al.
Computer recognition of regional lung disease patterns
Am J Resp Crit Care Med
(1999) - et al.
Quantification of pulmonary emphysema from lung computed tomography images
Am J Resp Crit Care Med
(1997) - et al.
Sensitivity and specificity for detection of 3D texture analysis of the lung parenchyma is better than 2D for discrimination of lung pathology in stage 0 COPD
Proc SPIE Med Imaging Physiol Function Structure Med Images
(2005)
A tutorial on v-support vector machine
Appl Stochastic Models Business Industry
Libsvma library for support vector machines
Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease. NHLBI/WHO Global Initiative for Chronic Obstructive Lung Disease (GOLD) workshop summary
Am J Resp Crit Care Med
Texture analysis. The handbook of pattern recognition and computer vision
Cited by (139)
Artificial Intelligence in the Imaging of Diffuse Lung Disease
2022, Radiologic Clinics of North AmericaAdvances in the characterisation of COPD using quantitative imaging
2024, ERS MonographQuantitative Imaging Metrics for the Assessment of Pulmonary Pathophysiology An Official American Thoracic Society and Fleischner Society Joint Workshop Report
2023, Annals of the American Thoracic SocietyA Review on Texture Feature Analysis of Chest Computed Tomography Images for Detection and Classification of Pulmonary Diseases
2023, Lecture Notes in Networks and Systems