Radiation-Preserving Selective Imaging for Pediatric Hip Dysplasia: A Cross-Modal Ultrasound-Xray Policy with Limited Labels

2025-11-23

AI Summary

This study addresses excessive ionizing radiation exposure in pediatric developmental dysplasia of the hip (DDH) diagnosis by proposing an “ultrasound-first, radiation-sparing” personalized imaging decision strategy. Methodologically, we develop a self-supervised, dual-modality encoder (SimSiam coupled with modality-specific ResNet-18), freeze its backbone, and attach a lightweight regression head to predict key radiographic metrics—e.g., alpha angle—with mean absolute error ≈9.7°. We integrate unilateral compliance–based deferred decision-making with decision curve analysis to yield interpretable, coverage-controllable imaging recommendations. Our key contribution is the first integration of cross-modal learning under limited labeling with statistically calibrated deferred decision-making, enabling clinically adjustable ultrasound-only pathways. Experiments demonstrate that X-ray metric prediction meets clinical accuracy requirements; moreover, threshold tuning allows flexible adaptation from conservative to permissive strategies, achieving non-zero ultrasound-only pathway rates while maintaining reliable diagnostic coverage.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Search and Optimization: Sampling/Simulation-based Search
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

We study an ultrasound-first, radiation-preserving policy for developmental dysplasia of the hip (DDH) that requests a radiograph only when needed. We (i) pretrain modality-specific encoders (ResNet-18) with SimSiam on a large unlabelled registry (37186 ultrasound; 19546 radiographs), (ii) freeze the backbones and fit small, measurement-faithful heads on DDH relevant landmarks and measurements (iii) calibrate a one sided conformal deferral rule on ultrasound predictions that provides finite sample coverage guarantees under exchangeability, using a held-out calibration set. Ultrasound heads predict Graf alpha, beta, and femoral head coverage; X-ray heads predict acetabular index (AI), center-edge (CE) angle and IHDI grade. On our held out labeled evaluation set, ultrasound measurement error is modest (e.g., alpha MAE ~= 9.7 degrees, coverage MAE ~= 14.0%), while radiographic probes achieve AI and CE MAEs of ~= 7.6 degrees and ~= 8.9 degrees, respectively. The calibrated US-only policy is explored across rule families (alpha-only; alpha OR coverage; alpha AND coverage), uncertainty inflation factors, and per-utility trade-offs using decision-curve analysis. Conservative settings yield high coverage with near-zero US-only rates; permissive settings (e.g., alpha OR coverage at larger deltas) achieve non-zero US-only throughput with expected coverage tradeoffs. The result is a simple, reproducible pipeline that turns limited labels into interpretable measurements and tunable selective imaging curves suitable for clinical handoff and future external validation.

Problem

• Reducing radiation exposure in pediatric hip dysplasia diagnosis
• Developing selective imaging policy using ultrasound before X-rays
• Creating calibrated deferral rules with limited labeled data

Innovation

• Pretrained modality-specific encoders using SimSiam
• Frozen backbones with measurement-faithful heads
• Calibrated conformal deferral rule for ultrasound predictions