Lesion-Aware Reconstruction with Principal Network: Enhancing Pseudo-Label Reliability in Semi-Supervised Clinical Lesion Detection

Shiwan Di, Jupeng Li, Yuxuan Yang, Qian JIN, Guorui AN, Jingwen Yang, Wang Yue, Yong GUO, Xinyue ZHANG, Ruohan MA, Gang LI

Keywords: Semi-supervised learning, Lesion detection, Pseudo-label calibration, Lesion-aware reconstruction, Teacher-Principal-Student framework

Abstract:

Purpose In lesion detection tasks, labeled medical data are often scarce, limiting the performance of fully supervised models. Teacher-Student frameworks based on semi-supervised learning (SSL) have emerged as effective solutions to leverage unlabeled data. However, the inherent high-confidence bias of teacher networks frequently leads to erroneous pseudo-label propagation, degrading the generalization ability of student networks. To address this critical issue, we propose a novel Teacher-Principal-Student (TPS) framework.
Methods The core innovation lies in introducing a Principal network, which integrates health-aware reconstruction to filter low-quality pseudo-labels generated by the teacher network. Specifically, the Principal network leverages anatomical prior knowledge and reconstruction consistency constraints to assess the reliability of teacher-generated pseudo-labels, ensuring only high-fidelity pseudo-labeled data are used for student network training. This design fundamentally mitigates the adverse effects of the teacher's prediction bias and error propagation.
Results Extensive experiments on jaw lesion detection datasets demonstrate the superiority of our approach. With the same annotation ratio, our SSL strategy achieves 81.5% mAP@0.5, outperforming mainstream semi-supervised methods by 3% while narrowing the performance gap with fully supervised learning to only 3.3%.
Conclusion Our proposed TPS framework outperforms state-of-the-art semi-supervised approaches in jaw lesion detection. It not only achieves competitive performance comparable to fully supervised models but also significantly reduces reliance on labeled clinical data, providing a reliable technical solution to promote the clinical translation of automated jaw lesion detection systems.

Primary Subject Area: Detection and Diagnosis

Secondary Subject Area: Image Synthesis

Registration Requirement: Yes

Visa & Travel: Yes

Read CFP & Author Instructions: Yes

Originality Policy: Yes

Single-blind & Not Under Review Elsewhere: Yes

LLM Policy: Yes

Submission Number: 193