JANUS: Anatomy-Conditioned Gating for Robust CT Triage Under Distribution Shift
2026-05-13 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors developed JANUS, a computer model that helps doctors quickly identify health problems in CT scans by combining detailed visual information with important physical measurements from the images. JANUS was tested on large sets of CT scans and performed better than other models, especially when spotting issues related to size and density. It also worked well on data from different hospitals, showing it can stay accurate even when conditions change. The authors also measured how often the model avoids confident mistakes, finding that JANUS reduces false alarms effectively. Overall, their approach seems to make diagnosing from CT scans both more accurate and reliable.
CT triageVision Transformersmacro-radiomicsAnatomically Guided GatingAUROCAUPRCdomain shiftcalibrationfalse positivesphysiological veto rate
Authors
Lavsen Dahal, Yubraj Bhandari, Geoffrey Rubin, Joseph Y. Lo
Abstract
Automated CT triage requires models that are simultaneously accurate across diverse pathologies and reliable under institutional shift. While Vision Transformers provide strong visual representations, many clinically significant findings are defined by quantitative imaging biomarkers rather than appearance alone. We introduce JANUS, a physiology-guided dual-stream architecture that conditions visual embeddings on macro-radiomic priors via Anatomically Guided Gating. On the MERLIN test set (N=5082), JANUS attains macro-AUROC 0.88 and AUPRC 0.74, outperforming all reproduced baselines. It generalizes to an external dataset N=2000; AUROC 0.87), with the largest gains on findings defined by size and attenuation as well as improved calibration on both datasets. We further quantify prediction suppression using the Physiological Veto Rate (PVR), showing that under domain shift JANUS reduces high-confidence false positives substantially more often than true positives. Together, these results are consistent with physically grounded conditioning that improves both discrimination and reliability in CT triage. Code is made publicly available at github repository https://github.com/lavsendahal/janus and model weights are at https://huggingface.co/lavsendahal/janus.