Artificial Intelligence in Medicine: AI Boosts Accuracy in Biparametric MRI

The least expensive and most efficient test used to screen men for prostate cancer (PCa) is the PSA blood test. However, the test is not specific for prostate cancer, so a suspicious result (abnormally high or rising) requires further information.

Until fairly recently, the conventional next step was a TRUS-guided needle biopsy. Unfortunately, this method tended to underdiagnose clinically significant PCa (csPCa) and to overdiagnose insignificant disease. Inaccurate diagnosis led to aggressive whole gland treatments for patients who would otherwise have been candidates for focal treatment or even Active Surveillance (AS). On the other hand, correctly identifying and characterizing csPCa would allow appropriate treatment planning with greater probability of successful cancer control while lowering treatment side effect risks.

Therefore, the most desirable pathway is one that identifies patients with csPCa before biopsy—and multiparametric MRI (mpMRI) has superior performance in this task. It offers two key advantages. First, it reduces the number of men who need a biopsy (on MRI no PCa is present, or insignificant PCa is present that qualifies for AS while monitoring for disease progression). Second, when csPCa is visible on mpMRI, the imaging allows a biopsy targeted to the suspicious lesion, greatly reducing the number of biopsy needles while increasing the accuracy of diagnosis.

Biparametric MRI and AI

Now there is a trend toward the use of biparametric MRI (bpMRI), a scan that eliminates the use of dynamic contrast enhancement. This shortens scan time, making prostate MRI more available to more patients—and especially useful in cases where use of a contrast agent is contraindicated. However, contrast adds clues that csPCa is present. Thus, there is a concern that without contrast agent, some csPCa may be missed.

This is where Artificial Intelligence (AI) comes in. A March 2025 article in Academic Radiology reported a summary of 19 published studies involving the use of various AI programs to interpret bpMRI patient scans.^[i] The analysis generated three key findings:

1. High diagnostic performance – compared with radiology readers who were unassisted by AI, the performance of bpMRI + AI had high sensitivity (85-88%) and specificity (79-83%) for detecting csPCa.
2. AI enhanced image interpretation – this included classifying tumors better the reducing discrepancies from one reader to another, therefore more reliable imaging findings.
3. Better treatment planning – by picking up more detailed diagnostic features, doctors are better equipped to develop appropriate treatment strategies.

The authors concluded that integrating AI into bpMRI scans overcomes the limitations of no contrast use and inconsistent subjective interpretations by radiologists, emphasizing “… that deep learning and machine learning technologies can enhance bpMRI capabilities in characterizing csPCa.” Both doctors and patients alike benefit from the addition of AI programs into today’s detection and diagnosis of csPCa.

NOTE: This content is solely for purposes of information and does not substitute for diagnostic or medical advice. Talk to your doctor if you are experiencing pelvic pain, or have any other health concerns or questions of a personal medical nature.

References

[i] Yan G, Wang Y, Chen L. Diagnostic Performance of Artificial Intelligence Based on Biparametric MRI for Clinically Significant Prostate Cancer: A Systematic Review and Meta-analysis. Acad Radiol. 2025 Mar 8:S1076- 6332(25)00188-6.