Share Email Print

Proceedings Paper

Effectiveness of computer aided detection for solitary pulmonary nodules
Author(s): Jiayong Yan; Wenjie Li; Xiangying Du; Huihai Lu; Jianrong Xu; Mantao Xu; Dongdong Rong
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

This study is to investigate the incremental effect of using a high performance computer-aided detection (CAD) system in detection of solitary pulmonary nodules in chest radiographs. The Kodak Chest CAD system was evaluated by a panel of six radiologists at different levels of experience. The observer study consisted of two independent phases: readings without CAD and readings with assistance of CAD. The study was conducted over a set of chest radiographs comprising 150 cancer cases and 150 cancer-free cases. The actual sensitivity of the CAD system is 72% with 3.7 false positives per case. Receiver operating characteristic (ROC) analysis was used to assess the overall observer performance. The AUZ (area under ROC curve) showed a significantly improvement (P=0.0001) from 0.844 to 0.884 after using CAD. The ROC analysis was also applied for observer performances on nodules in different sizes and visibilities. The average AUZs are improved from 0.798 to 0.835 (P=0.0003) for 5-10mm nodules, 0.853 to 0.907 (P=0.001) for 10-15mm nodules, 0.864 to 0.897 (P=0.051) for 15-20 mm nodules and 0.859 to 0.896 (P=0.0342) for 20-30mm nodules, respectively. For different visibilities, the average AUZs are improved from 0.886 to 0.915 (P=0.0337), 0.803 to 0.840 (P=0.063), 0.830 to 0.893 (P=0.0001), and 0.813 to 0.847 (P=0.152), for nodules clearly visible, hidden by ribs, partially overlap with ribs, and overlap with other structures, respectively. These results showed that observer performance could be greatly improved when the CAD system is employed as a second reader, especially for small nodules and nodules occluded by ribs.

Paper Details

Date Published: 12 March 2009
PDF: 8 pages
Proc. SPIE 7263, Medical Imaging 2009: Image Perception, Observer Performance, and Technology Assessment, 72631P (12 March 2009); doi: 10.1117/12.804716
Show Author Affiliations
Jiayong Yan, Carestream Health Global R&D (China)
Wenjie Li, Carestream Health Global R&D (China)
Shanghai Jiaotong Univ. (China)
Xiangying Du, Xuanwu Hospital of Capital Medical Univ. (China)
Huihai Lu, Carestream Health Global R&D (China)
Jianrong Xu, Renji Hospital of Shanghai Jiaotong Univ. (China)
Mantao Xu, Carestream Health Global R&D (China)
Dongdong Rong, Xuanwu Hospital of Capital Medical Univ. (China)

Published in SPIE Proceedings Vol. 7263:
Medical Imaging 2009: Image Perception, Observer Performance, and Technology Assessment
Berkman Sahiner; David J. Manning, Editor(s)

© SPIE. Terms of Use
Back to Top
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?