Aytekin Oto MD, Presenter: Consultant, General Electric Company Research grant, Koninklijke Philips Electronics NV Research grant, Bayer AG Research grant, Visualase Inc Research grant, General Electric Company

Maryellen L. Giger PhD, Abstract Co-Author: Stockholder, Hologic, Inc Royalties, Hologic, Inc Royalties, General Electric Company Royalties, Median Technologies Royalties, Riverain Medical Royalties, Mitsubishi Corporation Royalties, Toshiba Corporation

Li Lan MS, Abstract Co-Author: Nothing to Disclose

Arda Kayhan MD, Abstract Co-Author: Nothing to Disclose

Garima Agrawal MD, Abstract Co-Author: Nothing to Disclose

Maria Tretiakova PhD, Abstract Co-Author: Nothing to Disclose

Tatjana Antic, Abstract Co-Author: Nothing to Disclose

Scott Eggener, Abstract Co-Author: Nothing to Disclose

To develop a computer aided diagnosis method for analysis of multi-parametric MR data to assess prostate cancer aggressiveness.

Our database include 56 patients who underwent pre-operative endorectal 1.5 Tesla MRI (including T2-weighted, dynamic contrast enhanced MRI (DCE-MRI) and diffusion weighted imaging (DWI)) and radical prostatectomy. Region of interests were drawn jointly by a radiologist and pathologist over the index tumor in each case based on correlation with histopathology. Gleason score (GS) of index tumor was recorded for each case. Apparent diffusion coefficient (ADC) was calculated from these regions using commercially available workstations. Within each region, kinetic and texture analysis were performed to yield dynamic and morphological features, respectively. These features along with the ADC were then merged using a Bayesian Neural Network (BANN). The receiver operating characteristic (ROC) curve was constructed and the area under ROC curve (AUC) was used to determine the performance of the individual features, as well as the BANN, in terms of their ability to distinguish between Gleason scores.

42/56 (75%) cancer foci had GS of 6 or 7 and 14/56 (25%) cancer foci had GS of 8 or 9. Our initial round-robin validation yielded AUC values of 0.80 ± 0.07 for the combined computer output in the task of distinguishing between cases with Gleason Scores 6, 7 and cases with Gleason Scores 8,9, as compared to AUC values of 0.62, 0.66, and 0.65 obtained for individual ADC, kinetic, and texture feature analysis, respectively.

Combining multiple quantitative parameters obtained using endorectal MRI improves performance of any single parameter in predicting Gleason score. Computerized analysis of multiparametric MR data has potential for non-invasive assessment of tumor aggressiveness in prostate cancer.

Computerized analysis of prostate MRI, as a “virtual biopsy”, can potentially integrate multi-protocol MR image information and aid in clinical decision making.

Oto, A, Giger, M, Lan, L, Kayhan, A, Agrawal, G, Tretiakova, M, Antic, T, Eggener, S, Computer-aided Diagnosis of Prostate Cancer Aggressiveness Using Multiparametric MRI Data.  Radiological Society of North America 2010 Scientific Assembly and Annual Meeting, November 28 - December 3, 2010 ,Chicago IL. http://archive.rsna.org/2010/9009820.html