Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Video

Updated: Jul 6, 2026

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

Selective state space U net for linear complexity globally coherent high resolution image segmentation.

Adi Alhudhaif1

  • 1Department of Computer Science, College of Computer Engineering and Sciences in Al-kharj, Prince Sattam bin Abdulaziz University, P.O. Box 151, Al-kharj, 11942, Saudi Arabia. A.alhudhaif@psau.edu.sa.

Scientific Reports
|July 4, 2026
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

State Space Representation01:27

State Space Representation

The frequency-domain technique, commonly used in analyzing and designing feedback control systems, is effective for linear, time-invariant systems. However, it falls short when dealing with nonlinear, time-varying, and multiple-input multiple-output systems. The time-domain or state-space approach addresses these limitations by utilizing state variables to construct simultaneous, first-order differential equations, known as state equations, for an nth-order system.
Consider an RLC circuit, a...

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Distance-based temporal similarity metrics for adaptive channel selection in multi-modal EEG-fNIRS BCI frameworks.

Scientific reports·2026
Same author

Pain Level Classification from Speech Using GRU-Mixer Architecture with Log-Mel Spectrogram Features.

Diagnostics (Basel, Switzerland)·2025
Same author

Non-invasive enhanced hypertension detection through ballistocardiograph signals with Mamba model.

PeerJ. Computer science·2025
Same author

An innovative approach to detecting the freshness of fruits and vegetables through the integration of convolutional neural networks and bidirectional long short-term memory network.

Current research in food science·2024
Same author

Common-Unique Decomposition Driven Diffusion Model for Contrast-Enhanced Liver MR Images Multi-Phase Interconversion.

IEEE journal of biomedical and health informatics·2024
Same author

Spatio-temporal characterisation and compensation method based on CNN and LSTM for residential travel data.

PeerJ. Computer science·2024

Selective State-Space U-Net (SSS-U-Net) integrates selective state-space models for high-resolution image segmentation. This framework offers efficient global context modeling with linear complexity, outperforming existing methods on benchmarks.

Area of Science:

  • Computer Vision
  • Deep Learning
  • Image Segmentation

Background:

  • Accurate high-resolution image segmentation needs both fine spatial details and long-range context.
  • Convolutional Neural Networks (CNNs) excel at local features but have limited receptive fields.
  • Transformer models capture global context but suffer from quadratic computational costs.
  • Selective State-Space Models (SSMs) offer linear complexity for global dependencies, but vision applications lack standardized designs.

Purpose of the Study:

  • Introduce Selective State-Space U-Net (SSS-U-Net), a unified framework for integrating SSMs into encoder-decoder segmentation networks.
  • Formalize SSM integration with a taxonomy of topologies (sequential, parallel, bottleneck) and multi-directional spatial scanning.
  • Provide a standardized protocol for mapping continuous-time SSMs to high-resolution vision tasks with linear complexity.
Keywords:
Biomedical image analysisGlobal context modelingHigh-resolution image segmentationIndustrial anomaly detectionLinear-complexity deep learningMamba architectureSelective state-space models (SSMs)

Related Experiment Videos

Last Updated: Jul 6, 2026

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

Main Methods:

  • Developed SSS-U-Net, a novel architectural framework for dense prediction tasks.
  • Implemented three macro-integration topologies: sequential, parallel, and bottleneck state-space routing.
  • Defined multi-directional spatial scanning mechanisms for preserving 2D coherence with 1D SSMs.
  • Ensured linear complexity scaling for high-resolution image analysis.

Main Results:

  • Achieved 94.2% mean Dice score on Kvasir-SEG (polyp segmentation) and 98.6% pixel-level AUROC on MVTec AD (anomaly detection).
  • Demonstrated superior predictive performance compared to attention-based methods.
  • Showcased significantly lower memory consumption and computational overhead.
  • Ablation studies confirmed the benefits of selective state-space gating, multi-directional scanning, and hybrid CNN-Mamba fusion.

Conclusions:

  • SSS-U-Net provides a reproducible architectural blueprint for integrating SSMs into segmentation systems.
  • The framework offers practical design guidelines for scalable high-resolution image analysis.
  • SSS-U-Net is effective across biomedical and industrial imaging, including resource-constrained applications.