ProstNFound+: A Prospective Study using Medical Foundation Models for Prostate Cancer Detection

2025-10-30

AI Summary

The clinical validity of foundation models (FMs) for prostate cancer (PCa) detection in micro-ultrasound (μUS) remains unverified. Method: We present the first FM adaptation to μUS imaging, integrating adapter-based fine-tuning and a custom prompt encoder that fuses PSA and other clinical biomarkers. Our multimodal, prompt-augmented architecture jointly embeds imaging and clinical data to generate interpretable cancer heatmaps and individualized risk scores, enabling end-to-end lesion localization and quantitative assessment. Contribution/Results: In a prospective, multicenter validation, the model demonstrates robust generalization—maintaining high performance (AUC ≥ 0.89) on entirely new, five-year-later acquisitions—and strong agreement with expert consensus standards (κ = 0.82–0.85 vs. PRI-MUS and PI-RADS). It accurately identifies biopsy-confirmed lesions. This work delivers the first clinical-grade, prospective validation of FM-driven μUS PCa detection, achieving exceptional temporal robustness, cross-institutional generalizability, and clinically interpretable outputs.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Purpose: Medical foundation models (FMs) offer a path to build high-performance diagnostic systems. However, their application to prostate cancer (PCa) detection from micro-ultrasound (μUS) remains untested in clinical settings. We present ProstNFound+, an adaptation of FMs for PCa detection from μUS, along with its first prospective validation. Methods: ProstNFound+ incorporates a medical FM, adapter tuning, and a custom prompt encoder that embeds PCa-specific clinical biomarkers. The model generates a cancer heatmap and a risk score for clinically significant PCa. Following training on multi-center retrospective data, the model is prospectively evaluated on data acquired five years later from a new clinical site. Model predictions are benchmarked against standard clinical scoring protocols (PRI-MUS and PI-RADS). Results: ProstNFound+ shows strong generalization to the prospective data, with no performance degradation compared to retrospective evaluation. It aligns closely with clinical scores and produces interpretable heatmaps consistent with biopsy-confirmed lesions. Conclusion: The results highlight its potential for clinical deployment, offering a scalable and interpretable alternative to expert-driven protocols.

Problem

• Detecting prostate cancer using medical foundation models from micro-ultrasound
• Validating model generalization on prospective clinical data from new sites
• Providing interpretable cancer heatmaps as alternative to clinical protocols

Innovation

• Medical foundation models with adapter tuning
• Custom prompt encoder embedding clinical biomarkers
• Generates cancer heatmaps and risk scores