Prostate cancer (PCa) is one of the most commonly occurring forms of cancer, accounting for 21% of all cancer in men. Multi-parametric magnetic resonance imaging (mpMRI) has led to improved capabilities for detecting, localizing, and staging PCa. Combined with image-guided prostate biopsy, mpMRI has helped to improve diagnosis of clinically significant PCa, which helps to reduce mortality as well as unnecessary biopsies or treatments. Until recently, the diagnostic capabilities of mpMRI were limited by lack of standardization in imaging, interpretation, and reporting methods, which are all subject to high inter- and intra-observer variability. To address these problems, the Prostate Imaging-Reporting and Data System (PI-RADS) was designed to standardize the reporting of PCa. PI-RADS aims to standardize imaging acquisition parameters for mpMRI, simplify radiological reporting, and develop assessment categories to stratify levels of PCa. A recent meta- analysis reported the diagnostic performance of PI-RADS to have a high pooled sensitivity of 89% and specificity of 73%; unfortunately, there is still high variability in these results. The current clinical practice for interpreting mpMRI has limitations that may contribute to this variation. Currently, no image registration exists between the images, and radiologists rely on mental alignment of the images while reading a set of mpMR images, which introduces a potential source of variability into PCa diagnosis. Localization and reporting of PCa is specified with respect to the PI-RADS sector atlas, and this is another source of operator variation. An explicit manual delineation of the prostate into its constituent PI-RADS sectors would reduce variation, but this is time-consuming and infeasible in the clinical setting. The overarching goal of this proposal is to reduce the inter- and intra-observer variability while interpreting mpMRI images using the PI-RADS protocol to improve consistency and accuracy of PCa diagnosis. The primary innovation is creation of a population of PI-RADS sector atlases and their application to automatically segment anatomical prostate images with respect to this atlas label protocol. This project is significant in that it has the potential to reduce the variability in PCa interpretation and reporting by providing automated image analysis tools. While radiological results are currently communicated in a non-standardized format, the proposed work will facilitate development of automated electronic report generation capabilities to foster data sharing and collaborations. Ultimately, enhancements from this project will create a novel feature for Eigen?s (the applicant company?s) FDA 510(k)- cleared imaging product, ProFuse, that should improve the diagnosis of PCa.
In Aim 1 of this project, a tool to co-register and visualize multi-parametric prostate MR imaging will be developed.
In Aim 2, an image segmentation method to automatically localize the anatomical PI-RADS sector map standard within the prostate will be developed.
Both aims will utilize a database of existing mpMRI images to develop and validate the algorithms and validate their accuracy.

Public Health Relevance

Prostate cancer accounts for 21% of all cancer in men. The Prostate Imaging-Reporting and Data System (PI- RADS) was designed to standardize the reporting of prostate cancer based on medical imaging, but there is still high variability involved due to inter- and intra-observer variability. This project proposes to develop image analysis tools to automate and standardize the interpretation and reporting of radiological prostate cancer diagnosis. This system will be developed by automating components of the PI-RADS protocol and integrating these features into an already effective prostate cancer software system that is used for fusion-guided biopsies.

National Institute of Health (NIH)
National Cancer Institute (NCI)
Small Business Technology Transfer (STTR) Grants - Phase I (R41)
Project #
Application #
Study Section
Special Emphasis Panel (ZRG1)
Program Officer
Zhao, Ming
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Zmk Medical Technologies, D/B/A Eigen
Grass Valley
United States
Zip Code