aiqbench analysis
LightMedSeg-ISLES demonstrates that highly efficient medical imaging models can achieve performance nearly identical to massive nnU-Net architectures while drastically reducing parameter count and computational cost. This advancement addresses critical deployment barriers in resource-constrained clinical environments by proving that optimization techniques like test-time augmentation can compensate for reduced model size without sacrificing diagnostic accuracy.
LightMedSeg-ISLES: Stroke Lesion Segmentation with 81x Fewer Parameters than nnU-Net
arXiv:2609.09634v1 Announce Type: cross Abstract: Large networks and ensembles often lead medical image segmentation challenges, but their storage and inference demands complicate deployment. We present LightMedSeg-ISLES, a 1.26-million-parameter pipeline for T1-weighted stroke lesion segmentation in ISLES'26. On a 146-case held-out cohort, flip test-time augmentation produces 0.618 mean Dice and 0.599 lesion-wise F1. A 102.35-million-parameter nnU-Net ResEnc-L produces 0.634 Dice and 0.544 lesion-wise F1 after size filtering. LightMedSeg therefore retains 97.5\% of nnU-Net's Dice with 81.4$\times$ fewer parameters while improving lesion-wise F1 by 0.055. Its four-pass TTA operating point requires 4.7$\times$ fewer FLOPs per standardized patch than nnU-Net. It also slightly exceeds filtered UNETR++ and nnFormer. Longer training and stronger augmentation add 0.0358 Dice without increasing capacity, establishing a strong single-checkpoint alternative to much larger models.
Abstract:Large networks and ensembles often lead medical image segmentation challenges, but their storage and inference demands complicate deployment. We present LightMedSeg-ISLES, a 1.26-million-parameter pipeline for T1-weighted stroke lesion segmentation in ISLES'26. On a 146-case held-out cohort, flip test-time augmentation produces 0.618 mean Dice and 0.599 lesion-wise F1. A 102.35-million-parameter nnU-Net ResEnc-L produces 0.634 Dice and 0.544 lesion-wise F1 after size filtering. LightMedSeg therefore retains 97.5\% of nnU-Net's Dice with 81.4$\times$ fewer parameters while improving lesion-wise F1 by 0.055. Its four-pass TTA operating point requires 4.7$\times$ fewer FLOPs per standardized patch than nnU-Net. It also slightly exceeds filtered UNETR++ and nnFormer. Longer training and stronger augmentation add 0.0358 Dice without increasing capacity, establishing a strong single-checkpoint alternative to much larger models.
8 pages, 3 figures. Submitted to ISLES 2026 challenge. To be published in Nature Lecture Notes in Computer Science (LNCS)
Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
arXiv:2609.09634 [cs.CV]
(or arXiv:2609.09634v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2609.09634 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 240 words
Related coverage
Comparative evaluation of training strategies using partially labelled datasets for segmentation of white matter hyperintensities and stroke lesions in FLAIR MRI
arXiv · Jun 9, 2026
Comparative evaluation of training strategies using partially labelled datasets for segmentation of white matter hyperintensities and stroke lesions in FLAIR MRI
arXiv · Jul 23, 2026
Adaptive Post-Processing Drives Instance-Level Detection in Stroke Lesion Segmentation
arXiv · Aug 18, 2026
[AINews] not much happened today
Latent Space · Oct 3, 2026