Risk-Sensitive Conformal Prediction for Catheter Placement Detection in Chest X-rays

TL;DR

This paper introduces a multi-task learning model combined with risk-sensitive conformal prediction for detecting catheter positions in chest X-rays, providing reliable uncertainty quantification and zero high-risk errors for clinical use.

Contribution

It presents a novel integration of multi-task learning with risk-sensitive conformal prediction to improve reliability and safety in medical image analysis.

Findings

Achieved 90.68% overall coverage in predictions.

Attained 99.29% coverage for critical conditions.

Zero high-risk mispredictions in clinical scenarios.

Abstract

This paper presents a novel approach to catheter and line position detection in chest X-rays, combining multi-task learning with risk-sensitive conformal prediction to address critical clinical requirements. Our model simultaneously performs classification, segmentation, and landmark detection, leveraging the synergistic relationship between these tasks to improve overall performance. We further enhance clinical reliability through risk-sensitive conformal prediction, which provides statistically guaranteed prediction sets with higher reliability for clinically critical findings. Experimental results demonstrate excellent performance with 90.68\% overall empirical coverage and 99.29\% coverage for critical conditions, while maintaining remarkable precision in prediction sets. Most importantly, our risk-sensitive approach achieves zero high-risk mispredictions (cases where the system dangerously declares problematic tubes as confidently normal), making the system particularly suitable for clinical deployment. This work offers both accurate predictions and reliably quantified uncertainty -- essential features for life-critical medical applications.