DOI: 10.1148/radiol.2372041555
Computer-aided Detection of Peripheral Lung Cancers Missed at CT: ROC Analyses without and with Localization1
Feng Li, MD, PhD,
Hidetaka Arimura, PhD,
Kenji Suzuki, PhD,
Junji Shiraishi, PhD,
Qiang Li, PhD,
Hiroyuki Abe, MD, PhD,
Roger Engelmann, MS,
Shusuke Sone, MD, PhD,
Heber MacMahon, MD and
Kunio Doi, PhD
1 From the Kurt Rossmann Laboratories for Radiologic Image Research, Department of Radiology, MC-2026, University of Chicago, 5841 S Maryland Ave, Chicago, IL 60637 (F.L., K.S., J.S., Q.L., H.A., R.E., H.M., K.D.); Department of Health Sciences, Faculty of Medicine, Kyushu University, Fukuoka, Japan (H.A.); and J. A. Azumi General Hospital, Nagano, Japan (S.S.). From the 2003 RSNA Annual Meeting. Received September 8, 2004; revision requested October 29; revision received December 27; accepted January 21, 2005. Supported in part by U.S. Public Health Service grants CA62625 and CA98119
Address correspondence to F.L. (e-mail: feng{at}uchicago.edu).
View larger version (44K):
[in a new window]
|
Figure 1. (a) Example of the multiformat display used by the six radiologists in group 1. From the top to the bottom of the entire lung for each patient, 27 consecutive transverse CT sections were displayed in a multiformat (3 x 3) mode on three high-spatial-resolution monitors. (b) Stacked cine-mode display used by the eight radiologists in group 2. Magnified and stacked transverse CT sections were displayed on one monitor.
|
|
View larger version (38K):
[in a new window]
|
Figure 2. Graph shows ROC curves for detecting cancers missed at CT without and with use of the CAD scheme and for the two display modes. With the CAD scheme, the average Az value improved significantly from 0.757 to 0.862 for group 1 (P = .04) and from 0.768 to 0.848 for group 2 (P = .01).
|
|
View larger version (30K):
[in a new window]
|
Figure 3. Bar graph shows the number of radiologists who correctly detected cancer in each of 17 patients with lung cancer with and without the use of the CAD scheme. In eight patients, the CAD scheme had a beneficial effect for one to seven radiologists. In two patients, the use of CAD had a detrimental effect for two radiologists. (See details in Discussion.)
|
|
View larger version (41K):
[in a new window]
|
Figure 4. Graph shows LROC curves in the detection of cancers missed at CT for radiologists with and without use of the CAD scheme and with two display modes. The LROC curve was improved for groups 1 and 2 with use of the CAD scheme.
|
|
View larger version (95K):
[in a new window]
|
Figure 5a. Images obtained in a 69-year-old woman in whom CAD was helpful. Transverse CT scans show (a) a missed lung cancer (arrow) with pure GGO in the right upper lobe, (b) the cancer, and (c) a false-positive finding. Circles in b and c indicate the computer detections. In this patient, 10 radiologists did not detect the cancer without CAD, whereas CAD helped seven radiologists find the cancer.
|
|
View larger version (94K):
[in a new window]
|
Figure 5b. Images obtained in a 69-year-old woman in whom CAD was helpful. Transverse CT scans show (a) a missed lung cancer (arrow) with pure GGO in the right upper lobe, (b) the cancer, and (c) a false-positive finding. Circles in b and c indicate the computer detections. In this patient, 10 radiologists did not detect the cancer without CAD, whereas CAD helped seven radiologists find the cancer.
|
|
View larger version (91K):
[in a new window]
|
Figure 5c. Images obtained in a 69-year-old woman in whom CAD was helpful. Transverse CT scans show (a) a missed lung cancer (arrow) with pure GGO in the right upper lobe, (b) the cancer, and (c) a false-positive finding. Circles in b and c indicate the computer detections. In this patient, 10 radiologists did not detect the cancer without CAD, whereas CAD helped seven radiologists find the cancer.
|
|
Copyright © 2005 by the Radiological Society of North America.
