Developing an Ultrasound and Deep Learning System for Accurate Bladder Tumor Diagnosis A Multicenter Study on Automatic Tumor Detection and Grading

What this Trial Is About

Researchers are developing a deep learning system based on ultrasound images to improve the diagnosis of bladder tumors. This system aims to automatically segment tumors, determine their stage T-staging, and predict pathological grading. The goal is to make bladder cancer diagnosis more objective, accurate, and efficient by reducing reliance on physician experience and supporting precision medicine and resource optimization. The study focuses on creating and testing this diagnostic model using ultrasound images from patients suspected of having bladder tumors. It is an observational study conducted across multiple centers, starting in May 2025 and continuing through May 2027, during which the overall diagnostic accuracy of the deep learning system will be evaluated. Participants will be patients aged 18 to 85 years who have a suspected bladder mass detected by abdominal ultrasound and are scheduled for surgical treatment of bladder tumors. Researchers will collect ultrasound data and surgical pathology results to assess the models performance. The primary outcome measure is the overall diagnostic accuracy of the system, with safety and other assessments conducted as part of routine care. Participation will last as long as needed to complete the diagnostic evaluation and surgery within the study period.

Construction of a Deep Learning-Based Precise Diagnostic Framework for Bladder Tumors Using Ultrasound: A Multicenter, Ambispective Cohort Study