PenoMeter: a machine learning and algorithmic tool to advance Peyronie's disease assessment

Area of Science:

• Medical Imaging
• Computational Biology
• Urology

Background:

• Objective and reproducible quantification of Peyronie's disease (PD) curvature is limited.
• Accurate PD assessment is crucial for patient management and treatment evaluation.

Purpose of the Study:

• To develop an automated computational tool, PenoMeter, for objective penile curvature measurement from 2D images.
• To accurately and reproducibly assess the degree of penile angulation using artificial intelligence.

Main Methods:

• PenoMeter utilizes instance segmentation for penile anatomical identification and key point detection for shaft corners.
• Geometric calculations determine the point of maximal curvature and measure angulation.
• The model was trained on datasets and validated using independent digital penile images.

Main Results:

• PenoMeter demonstrated no intra-observer variability (0°), outperforming urologists (3.8°-7.8° variability).
• The tool achieved 86% agreement within the variability range of expert urologists.
• PenoMeter provides objective, AI-powered assessment for healthcare practitioners aiding in PD evaluation.

Conclusions:

• PenoMeter offers objective, accurate, and reproducible penile curvature assessment from digital images.
• The tool has potential for assisting initial PD assessments and tracking treatment outcomes.
• PenoMeter is an assistive tool and does not replace in-office gold-standard assessments.