Related Experiment Video
Updated: Jun 2, 2026
09:06
Whole-brain Segmentation and Change-point Analysis of Anatomical Brain MRI—Application in Premanifest Huntington's Disease
Published on: June 9, 2018
12.2K
A Data-Centric Approach to Deep Learning for Brain Metastasis Analysis at MRI
Laurens Topff1,2, Liliana Petrychenko1,2, Neeraj Jain1,2,3
1Department of Radiology, Netherlands Cancer Institute, Plesmanlaan 121, 1066 CX Amsterdam, the Netherlands.
Radiology
|June 24, 2025
Summary
This study developed a generalizable AI system for detecting and segmenting brain metastases (BMs) on MRI scans, achieving high accuracy for lesions of all sizes. The AI model shows promise for improved brain metastases detection and tracking.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Brain metastases (BMs) incidence is rising, necessitating advanced detection and analysis tools.
- Current artificial intelligence (AI) models for MRI-based BMs detection have limitations in identifying small lesions and generalizability.
- Developing robust AI systems is crucial for accurate volumetric analysis and longitudinal tracking of BMs.
Purpose of the Study:
- To create a generalizable deep learning system for detecting, segmenting, and longitudinally tracking brain metastases (BMs) of any size on MRI.
- To overcome limitations of existing AI models in identifying small BMs and ensuring broad applicability across different datasets and scanners.
Main Methods:
- A data-centric approach using a modified nnU-Net framework was employed for deep learning model development.
- A multicenter dataset of pre- and/or posttreatment MRI scans from patients with and without BMs was collected and curated.
- Iterative radiologist annotation with quality control and robust data preprocessing/augmentation were utilized.
Main Results:
- The deep learning system achieved high sensitivity (98.0% internal, 97.4% external testing) in detecting BMs across various sizes, including those <3 mm.
- Excellent segmentation performance was observed, with median Dice similarity coefficients of 0.89 (internal) and 0.90 (external).
- The model demonstrated strong generalizability, maintaining high performance on external test datasets from different scanners and patient cohorts.
Conclusions:
- The developed deep learning system exhibits high performance and generalizability for detecting and segmenting brain metastases (BMs) of all sizes on MRI.
- This AI system has the potential to significantly aid in the clinical management of patients with brain metastases.
- The findings support the use of advanced AI in improving the accuracy and efficiency of brain metastases analysis in clinical practice.
More Related Videos
Related Concept Videos
01:24Magnetic Resonance Imaging
Magnetic resonance imaging (MRI) is a noninvasive medical imaging technique based on a phenomenon of nuclear physics discovered in the 1930s, in which matter exposed to magnetic fields and radio waves was found to emit radio signals. In 1970, a physician and researcher named Raymond Damadian noticed that malignant (cancerous) tissue gave off different signals than normal body tissue. He applied for a patent for the first MRI scanning device in clinical use by the early 1980s. The early MRI...
01:14Brain Imaging
Brain imaging technologies provide critical insights into both the structure and function of the human brain, enabling medical professionals and researchers to diagnose, study, and treat neurological disorders or psychiatric disorders more effectively.
These technologies include computerized axial tomography (CAT or CT scans), positron-emission tomography (PET scans), magnetic resonance imaging (MRI), functional magnetic resonance imaging (fMRI), and Transcranial Magnetic Stimulation (TMS).
These technologies include computerized axial tomography (CAT or CT scans), positron-emission tomography (PET scans), magnetic resonance imaging (MRI), functional magnetic resonance imaging (fMRI), and Transcranial Magnetic Stimulation (TMS).