Related Experiment Video
Updated: May 29, 2025
09:11
Automation of the Micronucleus Assay Using Imaging Flow Cytometry and Artificial Intelligence
Published on: January 27, 2023
2.0K
Artificial Intelligence Apps for Medical Image Analysis using pyCERR and Cancer Genomics Cloud
Biorxiv : the Preprint Server for Biology
|February 3, 2025
Summary
This study presents a cloud-based software framework for AI-driven medical image analysis. It simplifies deploying AI models for radiological research, enabling reproducible end-to-end analysis without specialized hardware.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Radiological Research
Background:
- Medical image analysis for AI modeling requires robust software and hardware infrastructure.
- Current workflows often involve complex installations and dependencies, hindering accessibility.
- Integrating diverse data types and AI models presents significant challenges.
Purpose of the Study:
- To introduce a user-friendly, cloud-based software framework for AI analyses of medical images.
- To enable researchers to deploy AI-based workflows by customizing software and hardware dependencies.
- To facilitate reproducible, end-to-end radiological image analysis.
Main Methods:
- Development of a framework integrating the Computational Environment for Radiological Research (pyCERR) and Cancer Genomics Cloud (CGC).
- Porting pyCERR to Python for enhanced data organization, access, and transformation of multi-modal datasets.
- Implementation of analysis modules for image segmentation, radiomics, DCE MRI, and radiotherapy models.
- Provision of image processing utilities for training and inferring convolutional neural network models.
- Enabling round-trip analysis via APIs for seamless data processing and result retrieval.
Main Results:
- A cloud-compatible framework facilitating AI-based radiological image analysis.
- Extensible data structures and multi-modal visualization tools within pyCERR.
- Simplified deployment and execution of AI models on cloud infrastructure.
- Round-trip analysis capability, eliminating the need for local specialized software or GPU hardware.
- APIs for accessing deployed AI models in various programming languages.
Conclusions:
- The presented framework streamlines end-to-end radiological image analysis and promotes reproducible research.
- It simplifies data management, visualization, and access to image metadata for AI modeling.
- The framework enhances accessibility to advanced AI tools for the research community.
