arXiv subjects

Md. Mehedi Hassan

Publications and source records attributed to Md. Mehedi Hassan.

2 recordsLinked to original sources

arXiv2026

SYNAPSE-Net: A Unified Framework with Lesion-Aware Hierarchical Gating for Robust Segmentation of Heterogeneous Brain Lesions

Automatic segmentation of diverse heterogeneous brain lesions using multi-modal MRI is a challenging problem in clinical neuroimaging, mainly because of the lack of generalizability and high prediction variance of pathology-specific deep learning models. In this work, we propose a unified and adaptive multi-stream framework called SYNAPSE-Net to perform robust multi-pathology segmentation with reduced performance variance. The framework is based on multi-stream convolutional encoders with global context modeling and a cross-modal attention fusion strategy to ensure stable and effective multi-modal feature integration. It also employs a variance-aware training strategy to enhance the robustness of the network across diverse tasks. The framework is extensively validated using three public challenge datasets: WMH MICCAI 2017, ISLES 2022, and BraTS 2020. The results show consistent improvements in boundary accuracy, delineation quality, and stability across diverse pathologies. This proposed framework achieved a high Dice similarity coefficient (DSC) of 0.831 and a low Hausdorff distance at the 95th percentile (HD95) of 3.03 on the WMH MICCAI 2017 dataset. It also achieved the lowest HD95 of 9.69 on the ISLES 2022 dataset and the highest tumor core DSC of 0.8651 on the BraTS 2020 dataset. These results validate the robustness of the proposed framework in providing a clinically relevant computer-aided solution for automated brain lesion segmentation. Source code and pretrained models are publicly available at https://github.com/mubid-01/SYNAPSE-Net-pre.

cs.CV↗

arXiv2026

Context-Aware Asymmetric Ensembling for Interpretable Retinopathy of Prematurity Screening via Active Query and Vascular Attention

Retinopathy of Prematurity (ROP) is among the major causes of preventable childhood blindness. Automated screening remains challenging, primarily due to limited data availability and the complex condition involving both structural staging and microvascular abnormalities. Current deep learning models depend heavily on large private datasets and passive multimodal fusion, which commonly fail to generalize on small, imbalanced public cohorts. We thus propose the Context-Aware Asymmetric Ensemble Model (CAA Ensemble) that simulates clinical reasoning through two specialized streams. First, the Multi-Scale Active Query Network (MS-AQNet) serves as a structure specialist, utilizing clinical contexts as dynamic query vectors to spatially control visual feature extraction for localization of the fibrovascular ridge. Secondly, VascuMIL encodes Vascular Topology Maps (VMAP) within a gated Multiple Instance Learning (MIL) network to precisely identify vascular tortuosity. A synergistic meta-learner ensembles these orthogonal signals to resolve diagnostic discordance across multiple objectives. Tested on a highly imbalanced cohort of 188 infants (6,004 images), the framework attained State-of-the-Art performance on two distinct clinical tasks: achieving a Macro F1-Score of 0.93 for Broad ROP staging and an AUC of 0.996 for Plus Disease detection. Crucially, the system features `Glass Box' transparency through counterfactual attention heatmaps and vascular threat maps, proving that clinical metadata dictates the model's visual search. Additionally, this study demonstrates that architectural inductive bias can serve as an effective bridge for the medical AI data gap.

eess.IV↗