What question did this study set out to answer?

July 24, 2026Open Access

DynU-Net: Dynamic Uncertainty-Aware Multi-task U-Net for Joint Lesion Segmentation and Classification in Medical Imaging

NTNgoc Ly TranChonnam National University TNThi Thu Thuy NguyenChonnam National University BNBa Hung NgoChonnam National University

Key Points

The study aims to improve the efficacy of joint lesion segmentation and classification in medical imaging using a novel framework.
Developed DynU-Net, a dynamic uncertainty-aware network for multi-task learning.
Conducted experiments on three public datasets: Brain Stroke CT, Brain Tumor MRI, and BUSI (ultrasound).
Evaluated performance using Dice scores and macro-F1 scores, focusing on computational efficiency.
DynU-Net achieved a Dice score of 88.90% ± 0.33% on the Brain Stroke CT dataset.
Obtained a macro-F1 score of 96.73% ± 0.74% on the same dataset.
Demonstrated improved performance compared to single-task and existing multi-task baselines.

Abstract

Abstract Segmentation and classification of lesion is essential for computer-aided diagnosis (CAD) in grayscale medical imaging, enabling precise localization and reliable categorization of various pathological conditions. However, conventional multi-task learning (MTL) frameworks often suffer from suboptimal task interactions and require manual loss weight tuning, limiting their effectiveness on challenging modalities such as ultrasound, MRI, and CT. To address these limitations, we propose a Dynamic Uncertainty-aware Network (DynU-Net), a multi-task framework that adaptively balances segmentation and classification through learnable per-task uncertainty parameters. This mechanism eliminates the need for manual hyperparameter tuning while mitigating gradient imbalance during optimization. Extensive experiments on three public datasets, including Brain Stroke CT, Brain Tumor MRI, and BUSI (ultrasound), demonstrate that DynU-Net consistently outperforms both single-task and existing multi-task baselines. In particular, it achieves a Dice score of 88.90% ± 0.33% and a macro-F1 score of 96.73% ± 0.74% on the Brain Stroke CT dataset while maintaining favorable computational efficiency compared to competing architectures. These results highlight the effectiveness of dynamic uncertainty-aware optimization in enabling robust and balanced multi-task learning for joint lesion segmentation and classification on public medical imaging benchmarks.

Ask AI

Helpful

Bookmark

View Full Paper