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

Field strength-dependent sensitivity of chemical exchange saturation transfer: a methodological comparison of 3 Tesla and 7 Tesla in a clinical cohort.

Brain communications·2026

Same author

Comprehensive Performance Testing and External Validation of an AI Algorithm to Detect and Segment Brain Metastases.

Neuro-oncology·2026

Same author

Promising performance of locally deployed large language models for postoperative orthopaedic patient questions: An In Silico analysis.

Journal of experimental orthopaedics·2026

Same author

Fast surface reconstruction of human brain MRI: benchmarking deep-learning based morphometry tools.

Scientific reports·2026

Same author

Mitigating the Impact of MR Sequence Parameters: Increasing the Robustness of DL-Based Cortical Thickness Estimates.

Human brain mapping·2026

Same author

Mind the gap: challenges and future directions for content-based image retrieval in clinical radiology.

Frontiers in radiology·2026

Search research articles

Related Experiment Video

Updated: Jan 3, 2026

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

Divide and Conquer: Stratifying Training Data by Tumor Grade Improves Deep Learning-Based Brain Tumor Segmentation.

Michael Rebsamen^1,2, Urspeter Knecht³, Mauricio Reyes⁴

¹Support Center for Advanced Neuroimaging (SCAN), University Institute of Diagnostic and Interventional Neuroradiology, Inselspital, Bern University Hospital, University of Bern, Bern, Switzerland.

Frontiers in Neuroscience

|November 22, 2019

Summary

Stratifying brain tumor segmentation models by tumor type (high-grade glioma and low-grade glioma) significantly improves performance on individual cases, not just overall scores. This suggests treating different tumor types as separate tasks enhances segmentation accuracy.

Keywords:

automatic segmentation brain tumors data stratification deep learning magnetic resonance imaging training strategy

More Related Videos

Deep Learning-Based Segmentation of Cryo-Electron Tomograms

Deep Learning-Based Segmentation of Cryo-Electron Tomograms

Published on: November 11, 2022

Quantifying the Brain Metastatic Tumor Micro-Environment using an Organ-On-A Chip 3D Model, Machine Learning, and Confocal Tomography

Quantifying the Brain Metastatic Tumor Micro-Environment using an Organ-On-A Chip 3D Model, Machine Learning, and Confocal Tomography

Published on: August 16, 2020

Related Experiment Videos

Last Updated: Jan 3, 2026

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

Deep Learning-Based Segmentation of Cryo-Electron Tomograms

Deep Learning-Based Segmentation of Cryo-Electron Tomograms

Published on: November 11, 2022

Quantifying the Brain Metastatic Tumor Micro-Environment using an Organ-On-A Chip 3D Model, Machine Learning, and Confocal Tomography

Quantifying the Brain Metastatic Tumor Micro-Environment using an Organ-On-A Chip 3D Model, Machine Learning, and Confocal Tomography

Published on: August 16, 2020

Area of Science:

Medical image computing
Deep learning for medical image analysis
Brain tumor segmentation

Background:

Deep learning models typically improve with more data, generalizing across heterogeneous inputs.
Brain tumor segmentation is well-studied, largely due to the Multimodal Brain Tumor Segmentation (BraTS) challenge dataset.
Existing research often prioritizes model architecture over data properties.

Purpose of the Study:

To investigate the impact of tumor type on brain tumor segmentation performance.
To examine if stratifying the training dataset by tumor type (high-grade glioma [HGG] vs. low-grade glioma [LGG]) improves segmentation.
To analyze the influence of latent variables like image quality and tumor subtypes on segmentation accuracy.

Main Methods:

Utilized the BraTS 2018 dataset and the third-ranked segmentation method.
Stratified the training data into HGG and LGG subjects.
Trained two separate segmentation models, one for HGG and one for LGG, and compared to a baseline model trained on combined data.

Main Results:

Stratification yielded statistically significant improvements in case-wise rankings for tumor core segmentation (64.9% of cases, p < 0.0001).
Overall mean Dice scores showed only minor gains, but individual subject performance improved notably.
Cases with poor image quality or challenging subtypes (e.g., IDH1-mutant tumors) were identified as benefiting less from stratification.

Conclusions:

Stratifying brain tumor segmentation models by tumor type significantly increases performance, particularly evident in individual case rankings.
Treating brain tumor segmentation as separate tasks for different tumor types may be more effective than a single universal tool.
Considering tumor type information for test data could lead to a more clinically relevant BraTS competition.