Multicentric thrombus segmentation using an attention-based recurrent network with gradual modality dropout

TL;DR

This paper presents an attention-based recurrent network with a progressive modality dropout training schedule for robust 3D brain lesion segmentation across multiple centers and modalities.

Contribution

It introduces a novel method combining attention, recurrent units, and gradual modality dropout to improve multi-center small-lesion detection in medical imaging.

Findings

Achieves over 90% detection rate with a Dice score of 0.65 on monocentric data.

Maintains 80% detection rate with a Dice score of 0.35 in multi-center, missing modality scenarios.

Demonstrates transferability to other small-lesion 3D medical imaging tasks.

Abstract

Detecting and delineating tiny targets in 3D brain scans is a central yet under-addressed challenge in medical imaging.In ischemic stroke, for instance, the culprit thrombus is small, low-contrast, and variably expressed across modalities(e.g., susceptibility-weighted T2 blooming, diffusion restriction on DWI/ADC), while real-world multi-center dataintroduce domain shifts, anisotropy, and frequent missing sequences. We introduce a methodology that couples an attention-based recurrent segmentation network (UpAttLLSTM), a training schedule that progressively increases the difficulty of hetero-modal learning, with gradual modality dropout, UpAttLLSTM aggregates context across slices via recurrent units (2.5D) and uses attention gates to fuse complementary cues across available sequences, making it robust to anisotropy and class imbalance. Gradual modality dropout systematically simulates site heterogeneity,noise, and missing modalities during training, acting as both augmentation and regularization to improve multi-center generalization. On a monocentric cohort, our approach detects thrombi in >90% of cases with a Dice score of 0.65. In a multi-center setting with missing modalities, it achieves-80% detection with a Dice score around 0.35. Beyond stroke, the proposed methodology directly transfers to other small-lesion tasks in 3D medical imaging where targets are scarce, subtle, and modality-dependent