Saved in:
Bibliographic Details
Main Authors: Dayag, Elisha, Tran, Nhat Thanh, Xin, Jack
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.11131
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910209841037312
author Dayag, Elisha
Tran, Nhat Thanh
Xin, Jack
author_facet Dayag, Elisha
Tran, Nhat Thanh
Xin, Jack
contents Accurate medical image segmentation is an integral part of the medical image analysis pipeline that requires the ability to merge local and global information. While vision transformers are able to capture global interactions using vanilla self-attention, their quadratic computational complexity in the input size remains a struggle for medical image segmentation tasks. Motivated by the dispersion property of vanilla self-attention and recent development of Mamba form of attention, Scalable and Efficient Mamba like Attention (SEMA) utilizes token localization via local window attention to avoid dispersion and maintain focusing, complemented by theoretically consistent arithmetic averaging to capture global aspect of attention. In this work, we present USEMA, a hybrid UNet architecture that merges the local feature extraction ability of convolutional neural networks (CNNs) with SEMA attention. We conduct experiments with USEMA across a variety of modalities and image sizes, demonstrating improved computational efficiency compared to transformer based models using full self-attention, and superior segmentation performance relative to purely convolution and Mamba-based models.
format Preprint
id arxiv_https___arxiv_org_abs_2605_11131
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle USEMA: a Scalable Efficient Mamba Like Attention for Medical Image Segmentation
Dayag, Elisha
Tran, Nhat Thanh
Xin, Jack
Computer Vision and Pattern Recognition
Accurate medical image segmentation is an integral part of the medical image analysis pipeline that requires the ability to merge local and global information. While vision transformers are able to capture global interactions using vanilla self-attention, their quadratic computational complexity in the input size remains a struggle for medical image segmentation tasks. Motivated by the dispersion property of vanilla self-attention and recent development of Mamba form of attention, Scalable and Efficient Mamba like Attention (SEMA) utilizes token localization via local window attention to avoid dispersion and maintain focusing, complemented by theoretically consistent arithmetic averaging to capture global aspect of attention. In this work, we present USEMA, a hybrid UNet architecture that merges the local feature extraction ability of convolutional neural networks (CNNs) with SEMA attention. We conduct experiments with USEMA across a variety of modalities and image sizes, demonstrating improved computational efficiency compared to transformer based models using full self-attention, and superior segmentation performance relative to purely convolution and Mamba-based models.
title USEMA: a Scalable Efficient Mamba Like Attention for Medical Image Segmentation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2605.11131