Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Related Concept Videos

You might also read

Related Articles

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

Sort by

Same author

A Region-Aware Structured Framework Improves Prediction of Gene Expression from DNA Methylation.

Computational and structural biotechnology journal·2026

Same author

Sequential glioblastoma segmentation via topological data analysis and spatial adjacency.

Biomedical physics & engineering express·2026

Same author

Unveiling EMG semantics: a prototype-learning approach to generalizable gesture classification.

Journal of neural engineering·2024

See all related articles

Search research articles

Related Experiment Video

Updated: Jun 4, 2025

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

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

Published on: November 30, 2022

HeatGSNs: Integrating Eigenfilters and Low-Pass Graph Heat Kernels into Graph Spectral Convolutional Networks for

Jihun Bae¹, Hunmin Lee², Jinglu Hu³

¹Waseda University Graduate School of Information Production and Systems, Hibikino 2-7, Wakamatsu-ku, Kitakyushu, 808-0135, JAPAN.

Biomedical Physics & Engineering Express

|December 20, 2024

Summary

Novel HeatGSNs effectively address class imbalance in brain tumor MRI datasets using graph spectral convolutional networks. These models capture tumor geometric similarities, achieving high accuracy with fewer parameters than existing methods.

Keywords:

Brain tumor segmentation Eigenfilters Graph Heat Kernels Graph Signal Processing Graph Spectral Convolutional Networks Tumor instance classification

More Related Videos

Brain Infarct Segmentation and Registration on MRI or CT for Lesion-symptom Mapping

Brain Infarct Segmentation and Registration on MRI or CT for Lesion-symptom Mapping

Published on: September 25, 2019

Automated Segmentation of Cortical Grey Matter from T1-Weighted MRI Images

Automated Segmentation of Cortical Grey Matter from T1-Weighted MRI Images

Published on: January 7, 2019

Related Experiment Videos

Last Updated: Jun 4, 2025

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

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

Published on: November 30, 2022

Brain Infarct Segmentation and Registration on MRI or CT for Lesion-symptom Mapping

Brain Infarct Segmentation and Registration on MRI or CT for Lesion-symptom Mapping

Published on: September 25, 2019

Automated Segmentation of Cortical Grey Matter from T1-Weighted MRI Images

Automated Segmentation of Cortical Grey Matter from T1-Weighted MRI Images

Published on: January 7, 2019

Area of Science:

Medical imaging analysis
Graph representation learning
Artificial intelligence in oncology

Background:

Brain tumor segmentation faces challenges with imbalanced MRI datasets.
Deep learning models like CNNs and Transformers struggle with graph-based networks due to over-smoothing and convergence issues.
Existing graph networks are not ideal for capturing complex tumor geometric features.

Purpose of the Study:

To propose HeatGSNs, a novel graph spectral convolutional network for brain tumor segmentation.
To address class imbalance and overcome limitations of existing graph networks.
To improve the accuracy and efficiency of brain tumor learning tasks.

Main Methods:

Developed HeatGSNs incorporating eigenfilters and learnable low-pass graph heat kernels.
Implemented a continuous feature propagation mechanism using graph heat kernels.
Approximated the mechanism with cosine form for shift-scaled Chebyshev polynomial and modified Bessel functions.

Main Results:

Achieved a best average Dice score of 90% on the BRATS2021 dataset.
Obtained an average Hausdorff Distance (95%) of 5.45mm and an average accuracy of 80.11%.
Demonstrated significantly fewer parameters (1.79M) compared to existing methods.

Conclusions:

HeatGSNs effectively capture geometric similarities within tumor classes.
The proposed method offers an efficient and effective solution for brain tumor segmentation.
HeatGSNs show promise for improving deep learning in medical imaging tasks.