Segmentation

Uncertainty-Aware Retinal Vessel Segmentation via Ensemble Distillation
Distills an ensemble of segmentation models into a single model to estimate uncertainty in retinal vessel segmentation, matching ensemble-level calibration at a fraction of the compute.
PDF Source Document
Deep Ensemble Approach for Enhancing Brain Tumor Segmentation in Resource-Limited Settings
An ensemble of UNet3D, V-Net, and MSA-VNet fine-tuned on the BraTS-SSA dataset for glioma segmentation, aimed at resource-limited clinical settings in Sub-Saharan Africa.
PDF Source Document
MAPS-Glioma: Modality-Specific Augmentation and Tissue-Adaptive Postprocessing for Robust Glioma Segmentation in Resource-Limited Settings
A 3D U-Net framework with modality-specific augmentation and tissue-adaptive postprocessing for glioma segmentation, trained and validated on African MRI data from the BraTS-Africa challenge.
Source Document
Benchmarking Retinal Blood Vessel Segmentation Models for Cross-Dataset and Cross-Disease Generalization
A rigorous benchmark of five published retinal vessel segmentation architectures on the FIVES dataset, evaluating robustness to image quality and disease-induced domain shifts.
PDF Source Document
Efficiently Correcting Patch-Based Segmentation Errors to Control Image-Level Performance in Retinal Images
A quality-control and error-correction framework that selects the retinal-image patches most worth manually reviewing, using ensemble uncertainty to estimate the resulting Dice Score improvement.
PDF Source Document