Kenji Suzuki PhD, Presenter: Nothing to Disclose

Feng Li MD, PhD, Abstract Co-Author: Nothing to Disclose

Junji Shiraishi PhD, Abstract Co-Author: Nothing to Disclose

Qiang Li PhD, Abstract Co-Author: Nothing to Disclose

Heber MacMahon MD, Abstract Co-Author: Nothing to Disclose

Kunio Doi PhD, Abstract Co-Author: Nothing to Disclose

We carried out an observer study for evaluating radiologists’ performance without and with our computer-aided diagnostic (CAD) scheme for distinction between malignant and benign nodules on high-resolution CT (HRCT). Our purpose was to analyze radiologists’ responses with the CAD scheme in their classification task of nodules on HRCT.

Our observer study was carried out with 16 radiologists for 28 primary cancers and 28 benign nodules from a CT screening program. Results showed that radiologists’ performance was improved with CAD by a statistically significant level, and their performance with CAD (Az value of 0.85) was greater than either radiologists alone (0.79) or computer alone (0.83). To investigate radiologists’ responses with CAD, we analyzed the relationship among radiologists’ initial ratings (i.e., ratings without CAD), computer outputs, and radiologists’ ratings with CAD.

The average change in radiologists’ ratings from without to with CAD was strongly related to a) the likelihood of malignancy (computer output) and also b) the difference between their initial ratings and computer outputs, where the correlation coefficient was 0.93 and 0.90, respectively. Our analysis showed that radiologists changed the majority of their ratings in agreement with CAD results (55.7% of ratings), and did not change their decisions for most cases (39.4%). The majority of the changes contributed to the improvement in their performance (79.8% of the changes), i.e., a beneficial effect of CAD. In addition, they were able to maintain some of their correct ratings despite incorrect CAD results (7.2%). This indicated that radiologists were able to disregard some of "obvious" incorrect CAD results. Thus, the improvement in radiologists’ performance above CAD results was produced by the synergistic effect of radiologists’ decision making and CAD outputs.

Radiologists’ decision making was influenced by computer outputs, and they changed their ratings by taking into account the difference between computer outputs and their initial ratings. Radiologists’ performance could be improved above the CAD performance by maintaining their correct judgments despite incorrect CAD results.

K.D.,H.M.: K.D., H.M. are shareholders in R2 Technology, Inc., Sunnyvale, CA, and K.D. is a shareholder in Deus Technologies, Inc., Rockville, MD.

Suzuki, K, Li, F, Shiraishi, J, Li, Q, MacMahon, H, Doi, K, Analysis of Radiologists’ Responses with CAD in the Distinction between Malignant and Benign Pulmonary Nodules on High-Resolution CT.  Radiological Society of North America 2004 Scientific Assembly and Annual Meeting, November 28 - December 3, 2004 ,Chicago IL. http://archive.rsna.org/2004/4412115.html