TY - JOUR TI - Linking and characterizing biologic scales of imaging data DO - https://doi.org/doi:10.7282/T3MC8X4G PY - 2014 AB - Prostate cancer is the second most commonly diagnosed cancer of men, an estimated 192,000 men are diagnosed each year in the United States (source: American Cancer Society). The current gold standard for prostate cancer diagnosis is pathologist inspection of prostate needle biopsy samples obtained using transrectal ultrasound (TRUS). TRUS-guided biopsy is routine because TRUS is widely available and acquires real-time imagery. However, TRUS-guided biopsy has a low sensitivity, and initial biopsy misses approximately half of all men with prostate cancer. Multi-parametric Magnetic Resonance Imaging (MRI) has shown promise in detecting, localizing, and grading prostate cancer. MRI-TRUS fusion, whereby MRI is acquired pre-operatively then aligned to TRUS during biopsy, allows for both modalities to be leveraged. MRI-TRUS fusion will enable the construction of joint classifiers, which leverage imaging characteristics on both MRI and TRUS, to detect, localize, and grade prostate cancer. In order to train and validate these classifiers, ground truth spatial extent and aggressiveness of prostate cancer must be obtained. Manual pathologist inspection provides the ultimate definitive diagnosis of prostate cancer, with the Gleason grading system providing a measure of prostate cancer aggressiveness. Therefore whole mount histopathology (WMH) is aligned to fused MRI-TRUS imagery to provide ground truth of cancer location and aggressiveness. A drawback to this approach is that Gleason grade is subject to inter- and intra-observer variability. Hence there is a need for reproducible, computer assisted grading of pathology which can serve as a surrogate for ground truth prostate cancer aggressiveness. In Aim 1 we develop a novel registration algorithm, multi-attribute probabilistic elastic registration (MAPPER), to align MRI and TRUS prostate imagery. In Aim 2 we align WMH with fused MRI-TRUS imagery (Aim 1). In Aim 3 we develop novel morphologic features to distinguish between aggressive and non-aggressive prostate cancer on histopathology. This will enable WMH to serve as ground truth for prostate cancer aggressiveness in order to train a MRI-TRUS classifier. Future work will leverage the tools developed to combine signatures of prostate cancer appearance across MRI, TRUS, and WMH and enable the development of tools to target biopsy to aggressive prostate cancer. KW - Biomedical Engineering KW - Prostate--Cancer--Early detection KW - Prostate--Magnetic resonance imaging KW - Cancer--Ultrasonic imaging LA - eng ER -