Related Experiment Video
Updated: Jul 31, 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
12.1K
Cloud-based large-scale curation of medical imaging data using AI segmentation
Vamsi Krishna Thiriveedhi1, Deepa Krishnaswamy1, David Clunie2
1Brigham and Women's Hospital, Boston, MA.
Research Square
|May 15, 2024
Summary
Cloud computing enables efficient, large-scale medical image analysis. This study used cloud resources to analyze over 126,000 lung screening CT scans, completing the task in hours at a low cost.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Cloud Computing
Background:
- Medical imaging AI requires significant computational power, often exceeding on-premises capabilities.
- Cloud computing offers scalable and economical solutions for demanding AI tasks.
- Few studies evaluate cloud price/performance for medical image analysis.
Approach:
- Evaluated NCI CRDC Cloud Resources (Terra and Seven Bridges) for AI-based curation of National Lung Screening Trial (NLST) CT images.
- Performed automatic image segmentation and radiomics feature extraction on >126,000 CT volumes using TotalSegmentator and pyradiomics.
- Utilized over 21,000 Virtual Machines (VMs) for rapid analysis.
Key Points:
- Analysis completed in under 9 hours using cloud resources, compared to an estimated 522 days on a single workstation.
- Total cost for large-scale analysis was $1,011.05.
- Generated 9,565,554 segmentations and radiomics features for the NLST dataset.
Conclusions:
- Cloud computing provides a cost-effective and scalable solution for large-scale medical image analysis.
- Developed CloudSegmentator, an open-source workflow for reproducible medical image computing.
- Offers practical recommendations for optimizing cloud resource utilization in medical imaging research.
