TriGAN-SiaMT: A triple-segmentor adversarial network with bounding box priors for semi-supervised brain lesion segmentation

Abstract

Accurate brain lesion segmentation in MRI is critical for clinical decision-making, but pixel-wise annotations remain costly and time-consuming. We propose TriGAN-SiaMT, a novel semi-supervised segmentation framework that combines adversarial learning, consistency regularization, and bounding box priors. Our architecture comprises three segmentors (S0, S1, S2) and two discriminators (D0, D1). It includes: (1) a supervised branch (S0↔D0) trained on a small labeled subset; (2) a Siamese branch (S1↔D1) with an identical architecture to S0↔D0, but trained on unlabeled data; and (3) a teacher branch (S2) updated via exponential moving average (EMA) from S1, following the Mean Teacher (MT) paradigm. The teacher S2 generates pseudo-labels to supervise S1. It also provides soft segmentations to guide D1, which does not see any labeled data. The model enforces consistency at multiple levels: between S0 and S1 (Siamese consistency), and between S1 and S2 (EMA consistency). Bounding box priors are incorporated as weak supervision for both labeled and unlabeled images, improving lesion localization. Evaluated on the ISLES 2022 and BraTS 2019 datasets, TriGAN-SiaMT achieves DSC scores of 84.80 % and 86.32 %, respectively, using only 5 % labeled data. These results demonstrate strong performance under limited supervision and robust generalization across brain lesions. Keywords Brain lesion segmentation, Deep learning, Semi-supervised learning, Siamese, Mean-Teacher