Live intelligence

Latest AI updates

3 updates
Research

TotalSegmentator Aorta Report Automates Aortic Diameter Measurement on CT

Researchers introduced TotalSegmentator Aorta Report, an open-source tool that automatically measures aortic diameters at 11 standardized landmarks on CT images. In a retrospective study of 239 CT angiography examinations, the tool achieved a mean absolute error of 1.06 mm, with 86.6% of measurements within 2 mm of a radiologist-reviewed reference standard. Agreement was lower at the sinuses of Valsalva and in examinations with mural thrombus, while automatic measurement failed for 1.2% of landmarks.

TotalSegmentatoraortic diameter measurementCT angiography
Research

Radiology Agentic Framework Tests Limits of FDA-Cleared AI Tool Catalogs

The paper introduces a radiology agentic framework that standardizes how an orchestrator calls modular tools to produce structured reports. It simulates report generation on a multi-modality corpus using either a catalog consolidated from FDA-cleared radiology AI devices or an unlimited catalog, with tool outputs simulated from reference reports. With the cleared catalog and clinical indication alone, the orchestrator recovers under a third of clinically actionable findings; recovery improves with the unlimited catalog and further when planning from the target report. Most missed findings are classes no cleared device offers for the study's modality, while adding frequently missing tools closes most of the gap to the unlimited catalog.

radiology AIagentic systemsclinical reporting
Research Editor selected

GDGraph Compiles Radiology Guidelines into Executable Decision Graphs, Lifting LLM Accuracy to 90.8%

Researchers present Guideline-derived Decision Graph (GDGraph), a framework that decouples guideline compilation from case evaluation for radiological risk stratification and management planning. During compilation, a large language model partitions guidelines into section-based chunks, extracts local decision subgraphs with linked definitions, and aggregates related branches into an executable decision graph. At inference, a fast System One model answers localized clinical questions while deterministic traversal resolves Boolean logic, yielding a traceable decision path without task-specific training. Across 347 cases spanning renal cyst, adnexal mass, and thyroid nodule guidelines, GDGraph raises System One accuracy from 57.3% to 90.8%, outperforming retrieval-augmented GPT-4o (80.1%) and graph-guided GPT-4o with chain-of-thought (82.7%) at substantially lower latency and cost.

Why it matters: It shows how scattered guideline prose, tables and cross-referenced definitions can be compiled into an auditable decision path that beats GPT-4o baselines while costing less.

GDGraphClinical Decision SupportRadiology AI