Scaling Artificial Intelligence for Multi-Tumor Early Detection with More Reports, Fewer Masks

2025-10-16

AI Summary

Early multi-tumor detection in CT scans is hindered by reliance on labor-intensive pixel-level annotations and poor sensitivity for small lesions. Method: We propose R-Super, the first weakly supervised learning framework for tumor segmentation that requires only off-the-shelf clinical radiology reports—no pixel-level masks. It leverages natural language processing to extract anatomical and pathological descriptions from reports, generating spatially grounded cross-modal alignment cues between text and image. Results: Trained on 100,000 real-world reports, R-Super matches the performance of a fully supervised baseline trained on 723 meticulously annotated cases. With minimal additional mask supervision (≤50 samples), sensitivity and specificity improve by 13% and 8%, respectively. The framework generalizes successfully to six organs—including spleen, gallbladder, and prostate—with no prior public segmentation models. Across seven tumor types, it outperforms radiologists on five, substantially reducing annotation burden and enabling scalable, cost-effective multi-tumor early screening.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Semi-Supervised Learning
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Early tumor detection save lives. Each year, more than 300 million computed tomography (CT) scans are performed worldwide, offering a vast opportunity for effective cancer screening. However, detecting small or early-stage tumors on these CT scans remains challenging, even for experts. Artificial intelligence (AI) models can assist by highlighting suspicious regions, but training such models typically requires extensive tumor masks--detailed, voxel-wise outlines of tumors manually drawn by radiologists. Drawing these masks is costly, requiring years of effort and millions of dollars. In contrast, nearly every CT scan in clinical practice is already accompanied by medical reports describing the tumor's size, number, appearance, and sometimes, pathology results--information that is rich, abundant, and often underutilized for AI training. We introduce R-Super, which trains AI to segment tumors that match their descriptions in medical reports. This approach scales AI training with large collections of readily available medical reports, substantially reducing the need for manually drawn tumor masks. When trained on 101,654 reports, AI models achieved performance comparable to those trained on 723 masks. Combining reports and masks further improved sensitivity by +13% and specificity by +8%, surpassing radiologists in detecting five of the seven tumor types. Notably, R-Super enabled segmentation of tumors in the spleen, gallbladder, prostate, bladder, uterus, and esophagus, for which no public masks or AI models previously existed. This study challenges the long-held belief that large-scale, labor-intensive tumor mask creation is indispensable, establishing a scalable and accessible path toward early detection across diverse tumor types. We plan to release our trained models, code, and dataset at https://github.com/MrGiovanni/R-Super

Problem

• Reducing reliance on costly manual tumor masks for AI training
• Leveraging abundant medical reports to scale tumor detection AI
• Enabling multi-tumor segmentation where no prior models existed

Innovation

• Uses medical reports instead of manual tumor masks
• Trains AI to segment tumors matching report descriptions
• Enables scalable multi-tumor detection with fewer masks