Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Model Agnostic Preference Optimization for Medical Image Segmentation

About

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.

Yunseong Nam, Jiwon Jang, Dongkyu Won, Sang Hyun Park, Soopil Kim• 2025

Related benchmarks

TaskDatasetResultRank
Medical Image SegmentationISIC
DICE85.96
64
Medical Image SegmentationKvasir
mDice89.04
29
2D Medical Image SegmentationEBHI
ASD1.18
14
2D Medical Image SegmentationFIVES
ASD5.55
14
2D Medical Image SegmentationCOVID-QU-Ex
ASD0.69
14
2D Medical Image SegmentationCOVID-QU-Ex
Dice Score96.53
14
2D Medical Image SegmentationKvasir
ASD11.48
14
Lung SegmentationLUNA16
Dice97.02
13
3D Medical Image SegmentationParse 2022
Dice Coefficient74.2
8
3D Medical Image SegmentationSLIVER 07
Dice Coefficient0.7493
8
Showing 10 of 10 rows

Other info

Follow for update