AI enhanced diagnosis of Peyronies disease a novel approach using Computer Vision

TL;DR

This paper introduces an AI-based computer vision tool that accurately and non-invasively diagnoses Peyronie's Disease by measuring penile curvature with high sensitivity and specificity, improving traditional diagnostic methods.

Contribution

The study develops a novel AI-driven, non-invasive diagnostic approach using key point detection for Peyronie's Disease, validated against conventional methods.

Findings

Achieved 96.7% sensitivity in PD detection

Attained 100% specificity in distinguishing normal from diseased cases

Demonstrated high accuracy in landmark identification

Abstract

This study presents an innovative AI-driven tool for diagnosing Peyronie's Disease (PD), a condition that affects between 0.3% and 13.1% of men worldwide. Our method uses key point detection on both images and videos to measure penile curvature angles, utilizing advanced computer vision techniques. This tool has demonstrated high accuracy in identifying anatomical landmarks, validated against conventional goniometer measurements. Traditional PD diagnosis often involves subjective and invasive methods, which can lead to patient discomfort and inaccuracies. Our approach offers a precise, reliable, and non-invasive diagnostic tool to address these drawbacks. The model distinguishes between PD and normal anatomical changes with a sensitivity of 96.7% and a specificity of 100%. This advancement represents a significant improvement in urological diagnostics, greatly enhancing the efficacy and convenience of PD assessment for healthcare providers and patients.