Part I: prostate cancer detection, artificial intelligence for prostate cancer and how we measure diagnostic performance: a comprehensive review

This paper aims to address this need by first outlining the different types of AI used for the detection and diagnosis of prostate cancer, next deciphering how data collection methods, statistical analysis metrics (such as ROC and FROC analysis) and end points/outcomes (lesion detection vs. case diagnosis) affect the performance and limit the ability to compare between studies.

Part II: Effect of different evaluation methods to the application of a computer-aided prostate MRI detection/diagnosis (CADe/CADx) device on reader performance

Learn more about how AI operates and how it is evaluated from the experts, users and developers of leading Prostate AI software, ProstatIDTM.

Improving standard of Prostate Cancer Diagnosis

Learn more about Improving standard of Prostate Cancer Diagnosis and Care while increasing MRI revenues.

Peer Reviewed Articles


1. Assessment of the Diagnostic Accuracy of Biparametric Magnetic Resonance Imaging for Prostate Cancer in Biopsy-Naive Men, The Biparametric MRI for Detection of Prostate Cancer (BIDOC) StudyMore bpMRI, Lars Boesen, MD,PhD1Nis Nørgaard, MD1Vibeke Løgager, MD2; et al, JAMA Netw Open. 2018;1(2):e180219. doi:10.1001/jamanetworkopen.2018.0219

Conclusions and Relevance:  Low-suspicion bpMRI has a high NPV {97%} in ruling out

significant prostate cancer in biopsy-naive men. Using a simple and rapid bpMRI method as a triage test seems to improve risk stratification and may be used to exclude aggressive disease and avoid unnecessary biopsies with its inherent risks. Future studies are needed to fully explore its role in clinical prostate cancer management.


2. Clinically Significant Prostate Cancer Detection With Biparametric MRI: A Systematic Review and Meta-Analysis, Renato Cuocolo, MD1, Francesco Verde, MD1, Andrea

Ponsiglione, MD1, Valeria Romeo, MD, PhD1, Mario Petretta, MD2, Massimo Imbriaco, MD1 and Arnaldo Stanzione, MD1, American Journal of Roentgenology. 2021;216: 608-621. 10.2214/AJR.20.23219

Conclusion: Results confirm the feasibility of bpMRI for the detection of csPCa and for reducing acquisition time, patient discomfort, and costs. Nevertheless, the available studies proved to be heterogeneous, indicating a need for a more robust validation of this imaging protocol and a standardization of prostate bpMRI acquisition and reporting.


3. Hamdy, Donovan, Lane, Metcalfe, et. al., Fifteen-year Outcomes after Monitoring, Surgery, or Radiotherapy for Prostate Cancer, NEJM, March 11, 2023.  DOI: 10.1056/NEJMoa2214122

Concluded after a 15-year study of 1600 prostate patients that delaying aggressive care and performing active surveillance can often better serve the patient.


4. NIH – ProstatID’s beginnings: Detection of prostate cancer in multiparametric

MRI using random forest with instance weighting, Nathan Lay,a Yohannes Tsehay,a Matthew D. Greer,b Baris Turkbey,b Jin Tae Kwak,c Peter L. Choyke,b Peter Pinto,Bradford J. Wood,c and Ronald M. Summers,a, Journal of Medical Imaging 4(2), 024506 (Apr–Jun 2017). *a: National Institutes of Health, Clinical Center, Imaging Biomarkers and Computer Aided Diagnosis Laboratory, Bethesda, Maryland, United States; b: National Institutes of Health, National Cancer Institute, Urologic Oncology Branch and Molecular Imaging Program, Bethesda, Maryland, United States; c, National Institutes of Health, Clinical Center, Center for Interventional Oncology, Bethesda, Maryland, United States.

Conclusion: The random forest, features, sampling strategy, and instance-level weighting improve prostate cancer detection performance [area under the curve (AUC) 0.93] in comparison to SVM {earlier CAD method} (AUC 0.86) on the same test data.


5. Comparison of machine learning methods for detection of prostate cancer using bpMRI radiomics features, Ethan J Ulrich1 , Jasser Dhaouadi1, Robben Schat2, Benjamin Spilseth2, and Randall Jones1, Book of Abstracts, ISMRM Annual Meeting, London, UK, May 2022.

Conclusion Regarding BotImage’s ProstatID:  While all models demonstrated similar performance when evaluating at the lesion-level (ROC analysis), bpRF outperforms other models on FROC analysis. This indicates that the bpRF model produced fewer false positive detections at equal sensitivity.  This comparison was done with traditional Neural Net and other algorithms published recently.   The ProstatID detection engine is superior.


6. Improving Prostate Cancer Detection with MRI: A Multi-Reader, Multi-Case Study Using Computer-Aided Detection (CAD), Mark A. Anderson, MD*, Sarah Mercaldo, PhD*, Ryan Chung, MD, Ethan Ulrich, BS, Randall W. Jones, PhD, MBA, Mukesh Harisinghani, MD*, Academic Radiology 2022,

* Dept. of Radiology, Mass General Hospital

Conclusion Regarding BotImage’s ProstatID as the authors utilized ProstatID for the paper: Addition of a random forest method-based {ProstatID}, CAD-generated MRI image series improved inter-reader agreement and diagnostic performance for detection of clinically significant prostate cancer, particularly in the transition zone.