Atlas based AAM and SVM model for fully automatic MRI prostate segmentation.

Automatic prostate segmentation in MR images is a challenging task due to inter-patient prostate shape and texture variability, and the lack of a clear prostate boundary. We propose a supervised learning framework that combines the atlas based AAM and SVM model to achieve a relatively high segmentation result of the prostate boundary. The performance of the segmentation is evaluated with cross validation on 40 MR image datasets, yielding an average segmentation accuracy near 90%.

Is apparent diffusion coefficient associated with clinical risk scores for prostate cancers that are visible on 3-T MR images?

PURPOSE: To investigate whether apparent diffusion coefficients (ADCs) derived from diffusion-weighted (DW) magnetic resonance (MR) imaging at 3 T correlate with the clinical risk of prostate cancer in patients with tumors that are visible on MR images, with MR imaging/transrectal ultrasonography (US) fusion-guided biopsy as a reference. MATERIALS AND METHODS: Forty-eight consecutive patients (median age, 60 years; median serum prostate-specific antigen value, 6.3 ng/mL) who underwent DW imaging during 3-T MR imaging with an endorectal coil were included in this retrospective institutional review board-approved study, and informed consent was obtained from each patient. Patients underwent targeted MR imaging/transrectal US fusion-guided prostate biopsy. Mean ADCs of cancerous target tumors were correlated with Gleason and D'Amico clinical risk scores. The true risk group rate and predictive value of the mean ADC for classifying a tumor by its D'Amico clinical risk score was determined by using linear discriminant and receiver operating characteristic analyses. RESULTS: A significant negative correlation was found between mean ADCs of tumors in the peripheral zone and their Gleason scores (P = .003; Spearman ρ = -0.60) and D'Amico clinical risk scores (P < .0001; Spearman ρ = -0.69). ADC was found to distinguish tumors in the peripheral zone with intermediate to high clinical risk from those with low clinical risk with a correct classification rate of 0.73. CONCLUSION: There is a significant negative correlation between ADCs and Gleason and D'Amico clinical risk scores. ADCs may therefore be useful in predicting the aggressiveness of prostate cancer. SUPPLEMENTAL MATERIAL: http://radiology.rsna.org/lookup/suppl/doi:10.1148/radiol.10100667/-/DC1.