AI in High-Frequency Micro-Ultrasound: Advancing Prostate Imaging from Segmentation to Cancer Detection

TL;DR

This review summarizes how AI is being used with high-frequency prostate ultrasound to detect cancer and improve imaging accuracy, but more work is needed for real-world use.

Contribution

The paper provides the first comprehensive review of AI applications in 29 MHz micro-ultrasound for prostate cancer, highlighting technical approaches and current limitations.

Findings

AI models for cancer detection achieved AUROC values of 0.76–0.81 for core-level analysis.

Segmentation models achieved high accuracy with a Dice similarity coefficient of approximately 0.94.

A single study demonstrated precise 3D registration with histopathology (Dice 0.97 and landmark error < 3 mm).

Abstract

High-frequency micro-ultrasound is an emerging imaging technique for prostate cancer that allows doctors to visualize prostate tissue in real time during biopsy. In recent years, artificial intelligence has been applied to micro-ultrasound images to help identify suspicious areas, outline the prostate gland, and improve biopsy targeting. However, the rapid growth of this field has made it difficult for clinicians to understand what these technologies can currently do and what their real limitations are. In this review, we summarize and critically assess all published studies that have applied artificial intelligence to 29 MHz ExactVu micro-ultrasound of the prostate. We describe how these systems are used for cancer detection, prostate segmentation, and image alignment, and we highlight the main technical and clinical challenges that still need to be addressed. This work provides a practical overview for clinicians and researchers and helps guide future development of artificial intelligence in prostate imaging.

Background/Objective: High-frequency micro-ultrasound (micro-US) offers real-time, high-resolution imaging for prostate cancer. Although artificial intelligence (AI) has shown potential in enhancing micro-US interpretation, a comprehensive review of this emerging field is currently missing. This review synthesizes current evidence on AI applied to ExactVu 29 MHz micro-US for prostate cancer. Methods: PubMed/MEDLINE, Embase, Scopus, Web of Science and the Cochrane Library were searched up to December 2025. Studies were included if they applied machine learning or deep learning directly to 29 MHz micro-US data and reported quantitative performance metrics. Results: Ten studies met the inclusion criteria: six on prostate cancer detection, three on prostate segmentation and one on micro-US–histopathology registration. Detection models ranged from classical quantitative ultrasound machine learning to deep architectures using self-supervision, transformers, multiple-instance learning, ensemble calibration and 3D segmentation-based pipelines. Among core-level models for clinically significant cancer, area under the receiver operating characteristic curve (AUROC) values clustered around 0.76–0.81; one lesion-level framework reported an AUROC of 0.92, though at a non-comparable analytical unit. Segmentation studies achieved accurate prostate delineation (Dice similarity coefficient ≈ 0.94), and a single study demonstrated high-precision 3D registration to whole-mount histopathology (Dice similarity coefficient 0.97 and landmark error < 3 mm). All studies evaluated AI on previously acquired data, without real-time clinical implementation. Conclusions: AI for micro-US shows promising and reproducible early results across detection, segmentation and registration, but evidence is still limited. In view of the potential of AI to optimize micro-US utilization and its related advantages, additional efforts are warranted to achieve clinical adoption.