Model Agnostic Preference Optimization for Medical Image Segmentation
This work addresses the need for scalable and flexible supervision in medical image segmentation, offering a model-agnostic solution that improves segmentation quality for healthcare applications, though it appears incremental as it builds on existing preference optimization methods.
The paper tackles the problem of model-specific and low-diversity prediction sampling in preference optimization for medical image segmentation by proposing MAPO, a model-agnostic framework that uses Dropout-driven stochastic hypotheses to construct preference-consistent gradients without ground-truth supervision, resulting in enhanced boundary adherence, reduced overfitting, and more stable optimization dynamics across diverse datasets.
Preference optimization offers a scalable supervision paradigm based on relative preference signals, yet prior attempts in medical image segmentation remain model-specific and rely on low-diversity prediction sampling. In this paper, we propose MAPO (Model-Agnostic Preference Optimization), a training framework that utilizes Dropout-driven stochastic segmentation hypotheses to construct preference-consistent gradients without direct ground-truth supervision. MAPO is fully architecture- and dimensionality-agnostic, supporting 2D/3D CNN and Transformer-based segmentation pipelines. Comprehensive evaluations across diverse medical datasets reveal that MAPO consistently enhances boundary adherence, reduces overfitting, and yields more stable optimization dynamics compared to conventional supervised training.