Takatoshi Aoki MD, Presenter: Nothing to Disclose

Seiichi Murakami, Abstract Co-Author: Nothing to Disclose

Masami Fujii MD, Abstract Co-Author: Nothing to Disclose

Hyoungseop Kim, Abstract Co-Author: Nothing to Disclose

Shigehiko Katsuragawa PhD, Abstract Co-Author: Shareholder, Hologic, Inc

Yukunori Korogi MD, Abstract Co-Author: Nothing to Disclose

Noriaki Miyake, Abstract Co-Author: Nothing to Disclose

Junji Shiraishi, Abstract Co-Author: Nothing to Disclose

We heve been developing a new temporal subtraction (TS) method which could reduce misregistration artifacts on the subtraction images in successive thoracic CT scans. In this study, the usefulness of this computerized method on the radiologists’ performance for lung nodule detection on thin-section CT was evaluated with Jacknife free-response receiver operating characteristics (JAFROC) observer study.

Fifty pairs of standard-dose 2mm-thick CT and their TS images were used for an observer performance study. All 50 datasets were interpreted by two thoracic radiologists to identify the reference standard of actionable nodules (non-calcified nodules larger than 4 mm). Eight radiologists (four attending radiologists and four radiology residents) participated in this observer study. PC based reading tool for viewing stack images with and without TS images were originally developed. Ratings and locations of “lesions” obtained with this tool were utilized for assessing the statistical significance of differences between radiologists' performances without and with the TS images in JAFROC analysis.

A total of 30 nodules, ranging in size from 5 to 19 mm, were identified as the reference standard. The average figure-of-merit (FOM) values for all radiologists increased to a statistically significant degree, from 0.838 without the TS images to 0.894 with the images (P=.033). The average sensitivity for detecting actionable nodules was improved from 73.4 % to 83.4 % at a false-positive rate of 0.15 per case by use of the TS images. The reading time with TS images was not significantly different from that without.

The novel TS method for CT images would increase the observer performance for lung nodule detection on thin-section CT without extending the reading time.

A novel TS method can sufficiently assist the radiologists’ interpretation for lung nodule detection by reducing the artifacts on TS images.

Aoki, T, Murakami, S, Fujii, M, Kim, H, Katsuragawa, S, Korogi, Y, Miyake, N, Shiraishi, J, A New Temporal Subtraction Method for Lung Nodule Detection in Successive Diagnostic Thoracic CT Scans: JAFROC Observer Study.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11010628.html