Program Official

Principal Investigator

Gregory John
Awardee Organization

University Of Minnesota
United States

Fiscal Year
Activity Code
Early Stage Investigator Grants (ESI)
Not Applicable
Project End Date

Computer Aided Diagnostic System for Prostate Cancer Detection Using Quantitative Multiparametric MRI

Despite the prevalence of prostate cancer, the current tools available to manage the disease continues to leave physicians and their patients in a position to overdiagnose and overtreat. The confidence to pursue more conservative approaches like active surveillance are limited, as biopsy is known to underestimate the grade and extent of disease, both of which are important for risk stratification. Targeted biopsies, by means of MRIguidance, are becoming the preferred way to ensure the most aggressive appearing lesions are sampled in the hopes of avoiding some of the issues with standard biopsy approaches. These targeted biopsies make use of multi-parametric MRI (mpMRI) which includes both anatomical and functional information that are complimentary and together increase the sensitivity and specificity for cancer detection. However, the ability to effectively use mpMRI requires specialized training while the standards for properly using the multiple MRI datasets are still being developed. To address this issue, we have developed an alternative method that would provide a quantitative, user-independent, summary of the mpMRI data (qMRI) to visually “map” disease and assess its aggressiveness. Using quantitative MRI, a Composite Biomarker Score (CBS) map is generated, with a demonstrated increase in sensitivity and specificity for tumor detection compared to any single qMRI parameter. Our primary goal is to integrate this predictive qMRI model into a computer-aided diagnostic (CAD) system (referred to as CBS-CAD) to improve the use of mpMRI in PCa management. Employing quantitative MRI (qMRI) can address the issues of a qualitative image analysis if the major roadblocks to its adoption can be overcome. To address the roadblocks and implement the CBS-CAD system we will pursue the following specific aims: 1) develop an analysis pipeline to evaluate qMRI performance and translate CBS-CAD methods, 2) perform a multi-vendor, multi-site quantitative imaging technical performance evaluation and 3) perform a multi-center clinical validation study assessing CBS-CAD performance. Our expected outcome of this academic-industry partnership will be the integration of several novel technologies into a comprehensive CAD system consisting of a phantom and automated software for 1) qMRI system validation and 2) clinical translation of novel models for detecting cancer and assessing aggressiveness.


  • Jin J, Zhang L, Leng E, Metzger GJ, Koopmeiners JS. Bayesian spatial models for voxel-wise prostate cancer classification using multi-parametric magnetic resonance imaging data. Statistics in medicine. 2022 Feb 10;41(3):483-499. Epub 2021 Nov 7. PMID: 34747059
  • Masotti M, Zhang L, Leng E, Metzger GJ, Koopmeiners JS. A novel Bayesian functional spatial partitioning method with application to prostate cancer lesion detection using MRI. Biometrics. 2023 Jun;79(2):604-615. Epub 2021 Dec 8. PMID: 34806765
  • Jin J, Zhang L, Leng E, Metzger GJ, Koopmeiners JS. Multi-resolution super learner for voxel-wise classification of prostate cancer using multi-parametric MRI. Journal of applied statistics. 2021 Dec 17;50(3):805-826. doi: 10.1080/02664763.2021.2017411. eCollection 2023. PMID: 36819087