Share Email Print
cover

Proceedings Paper

Computer-aided detection of bladder mass within non-contrast-enhanced region of CT Urography (CTU)
Author(s): Kenny H. Cha; Lubomir M. Hadjiiski; Heang-Ping Chan; Elaine M. Caoili; Richard H. Cohan; Alon Weizer; Chuan Zhou
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

We are developing a computer-aided detection system for bladder cancer in CT urography (CTU). We have previously developed methods for detection of bladder masses within the contrast-enhanced region of the bladder. In this study, we investigated methods for detection of bladder masses within the non-contrast enhanced region. The bladder was first segmented using a newly developed deep-learning convolutional neural network in combination with level sets. The non-contrast-enhanced region was separated from the contrast-enhanced region with a maximum-intensityprojection- based method. The non-contrast region was smoothed and a gray level threshold was employed to segment the bladder wall and potential masses. The bladder wall was transformed into a straightened thickness profile, which was analyzed to identify lesion candidates as a prescreening step. The lesion candidates were segmented using our autoinitialized cascaded level set (AI-CALS) segmentation method, and 27 morphological features were extracted for each candidate. Stepwise feature selection with simplex optimization and leave-one-case-out resampling were used for training and validation of a false positive (FP) classifier. In each leave-one-case-out cycle, features were selected from the training cases and a linear discriminant analysis (LDA) classifier was designed to merge the selected features into a single score for classification of the left-out test case. A data set of 33 cases with 42 biopsy-proven lesions in the noncontrast enhanced region was collected. During prescreening, the system obtained 83.3% sensitivity at an average of 2.4 FPs/case. After feature extraction and FP reduction by LDA, the system achieved 81.0% sensitivity at 2.0 FPs/case, and 73.8% sensitivity at 1.5 FPs/case.

Paper Details

Date Published: 24 March 2016
PDF: 6 pages
Proc. SPIE 9785, Medical Imaging 2016: Computer-Aided Diagnosis, 97853W (24 March 2016); doi: 10.1117/12.2216289
Show Author Affiliations
Kenny H. Cha, Univ. of Michigan Health System (United States)
Lubomir M. Hadjiiski, Univ. of Michigan Health System (United States)
Heang-Ping Chan, Univ. of Michigan Health System (United States)
Elaine M. Caoili, Univ. of Michigan Health System (United States)
Richard H. Cohan, Univ. of Michigan Health System (United States)
Alon Weizer, Univ. of Michigan Health System (United States)
Chuan Zhou, Univ. of Michigan Health System (United States)


Published in SPIE Proceedings Vol. 9785:
Medical Imaging 2016: Computer-Aided Diagnosis
Georgia D. Tourassi; Samuel G. Armato, Editor(s)

© SPIE. Terms of Use
Back to Top