Improving Early Prostate Cancer Detection Through Artificial Intelligence: Evidence from a Systematic Review

Area of Science:

• Oncology
• Medical Imaging
• Artificial Intelligence

Background:

• Prostate cancer is a leading cause of cancer mortality in men.
• Traditional diagnostic methods (PSA, DRE, biopsy) have limitations in accuracy and consistency.
• Multiparametric MRI improves detection but relies on expert interpretation; AI offers potential for enhanced accuracy and efficiency.

Purpose of the Study:

• To evaluate the diagnostic accuracy of AI technologies for early prostate cancer detection.
• To assess the timeliness of reporting with AI-based diagnostic tools.
• To compare AI performance against conventional diagnostic methods.

Main Methods:

• Systematic literature search (PubMed, Scopus, Web of Science, Cochrane) from Jan 2015-Apr 2025, adhering to PRISMA 2020 guidelines.
• Inclusion of RCTs, cohort, case-control, and pilot studies using AI for prostate cancer diagnosis.
• Narrative synthesis of AUC-ROC, sensitivity, specificity, predictive values, DOR, and reporting time; QUADAS-AI tool for bias assessment.

Main Results:

• Twenty-three studies with 23,270 patients were analyzed.
• AI achieved a median AUC-ROC of 0.88, with median sensitivity of 0.86 and specificity of 0.83.
• AI or AI-assisted readings matched or improved diagnostic accuracy, reduced variability, and decreased reporting time by up to 56% compared to radiologists.

Conclusions:

• AI demonstrates significant diagnostic performance in early prostate cancer detection.
• Methodological heterogeneity and lack of standardization currently limit the generalizability of AI findings.
• Large-scale prospective trials are essential to validate the clinical integration of AI in prostate cancer diagnostics.