LightMIS: Ultra-Lightweight Medical Image Segmentation Without a Stage-Wise Decoder

Hugging Face Daily Papers Papers
medical-imaging image-segmentation lightweight-model convolutional-networks open-source arxiv efficient-ai

Summary

LightMIS presents a scalable family of ultra-lightweight convolutional networks for 2D binary medical image segmentation, achieving high accuracy with reduced parameters and GFLOPs, suitable for on-device execution.

We present LightMIS, a scalable family of ultra-lightweight convolutional networks for 2D binary medical image segmentation without a learned stage-wise decoder. LightMIS aligns the outputs of a five-level encoder to a common resolution using Scale-Aligned Projection blocks, aggregates them once, and refines the fused representation with an Adaptive Fusion Cascade. The cascade combines Adaptive Kernel Fusion with the proposed Progressive Receptive Fusion module, which uses temporary channel expansion, complementary depthwise receptive fields, and progressive cross-branch information transfer. We evaluate LightMIS-T, LightMIS-S, and LightMIS using five-fold cross-validation under a common nnU-Net v2.3.1 protocol on DRIVE, Kvasir-SEG, DSB18, BUSI, ISIC-2017, and ISIC-2018. Full LightMIS contains 0.131 M parameters and requires 0.575 GFLOPs for a 3times256times256 input, achieving modality-macro Dice and IoU scores of 86.71% and 78.99%, respectively. Mobile U-ViT obtains 86.75% Dice and 79.07% IoU, so the observed differences are 0.04 and 0.08 percentage points. Relative to Mobile U-ViT, nnWNet, and nnU-Net, LightMIS reduces parameter count by 90.58-99.61% and GFLOPs by 82.54-96.14%. On an Arm Mali-G52 MC2 GPU, all LightMIS variants achieve full GPU delegation, with median delegated latency ranging from 53.31 ms for LightMIS-T to 138.31 ms for LightMIS. These results demonstrate a favorable accuracy-complexity trade-off and on-device execution feasibility for the evaluated tasks. The code is publicly available at https://github.com/AndreiiArhire/LightMIS.
Original Article
View Cached Full Text

Cached at: 09/28/26, 04:05 PM

Paper page - LightMIS: Ultra-Lightweight Medical Image Segmentation Without a Stage-Wise Decoder

Source: https://huggingface.co/papers/2609.28327

Abstract

WepresentLightMIS,ascalablefamilyofultra-lightweightconvolutionalnetworksfor2Dbinarymedicalimagesegmentationwithoutalearnedstage-wisedecoder.LightMISalignstheoutputsofafive-levelencodertoacommonresolutionusingScale-AlignedProjectionblocks,aggregatesthemonce,andrefinesthefusedrepresentationwithanAdaptiveFusionCascade.ThecascadecombinesAdaptiveKernelFusionwiththeproposedProgressiveReceptiveFusionmodule,whichusestemporarychannelexpansion,complementarydepthwisereceptivefields,andprogressivecross-branchinformationtransfer.WeevaluateLightMIS-T,LightMIS-S,andLightMISusingfive-foldcross-validationunderacommonnnU-Netv2.3.1protocolonDRIVE,Kvasir-SEG,DSB18,BUSI,ISIC-2017,andISIC-2018.FullLightMIScontains0.131Mparametersandrequires0.575GFLOPsfora3times256times256input,achievingmodality-macroDiceandIoUscoresof86.71%and78.99%,respectively.MobileU-ViTobtains86.75%Diceand79.07%IoU,sotheobserveddifferencesare0.04and0.08percentagepoints.RelativetoMobileU-ViT,nnWNet,andnnU-Net,LightMISreducesparametercountby90.58-99.61%andGFLOPsby82.54-96.14%.OnanArmMali-G52MC2GPU,allLightMISvariantsachievefullGPUdelegation,withmediandelegatedlatencyrangingfrom53.31msforLightMIS-Tto138.31msforLightMIS.Theseresultsdemonstrateafavorableaccuracy-complexitytrade-offandon-deviceexecutionfeasibilityfortheevaluatedtasks.Thecodeispubliclyavailableathttps://github.com/AndreiiArhire/LightMIS.

View arXiv pageView PDFGitHub3Add to collection

Get this paper in your agent:

hf papers read 2609\.28327

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2609.28327 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2609.28327 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2609.28327 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

M^{2}SNet: Multi-scale in Multi-scale Subtraction Network for Medical Image Segmentation

Papers with Code Trending

This paper proposes M2SNet, a multi-scale multi-scale subtraction network for medical image segmentation that uses subtraction operations to capture difference features between encoder levels, improving localization and edge sharpness. It achieves state-of-the-art performance on eleven datasets across four medical imaging modalities.

MiLSD: A Micro Line-Segment Detector for Resource-Constrained Devices

arXiv cs.AI

This paper presents MiLSD, a micro line-segment detector designed for resource-constrained devices like microcontrollers. It systematically compares output representations within a compact fully-convolutional backbone and shows that 8-bit quantization preserves full-precision performance, while 4-bit quantization causes degradation, achieving improved accuracy within a 1 MB activation budget.

@helloiamleonie: OpenMed Liquid AI

X AI KOLs Timeline

OpenMed demonstrated using Liquid AI's LFM2.5-VL-3B model to analyze a generated skin image, mapping regions and measuring dimensions locally on a Mac Studio for visual review.