Deep learning model TREX predicts rectal cancer regrowth from endoscopy

Researchers have developed a deep learning model called TREX to predict rectal cancer regrowth from longitudinal endoscopy images. TREX utilizes a siamese network with Swin Transformers and dual cross-attention to analyze pairs of images taken at different times, distinguishing between continued response and local regrowth. The model demonstrated high accuracy in detecting regrowth and showed promise in early detection months before clinical confirmation, even matching attending-level accuracy in a surgeon survey. AI

IMPACT Introduces a novel deep learning approach for early detection of rectal cancer regrowth, potentially improving patient surveillance and outcomes.

RANK_REASON Academic paper detailing a novel deep learning approach for medical image analysis. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• Rectal cancer
• Swin Transformers

• paper
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →