Room: AAPM ePoster Library
A software application has been developed to manage the datasets and automate data analysis associated with physicist’s tasks within the ACR MRI Accreditation Program (MRAP).
The application contains four major modules: a database repository for record-keeping and historical data tracking, an automated image analysis kernel, an automated report-generating tool, and a front-end graphical user interface. It is designed to handle the major physicist’s tasks within the ACR MRI accreditation program (ACR MRAP): acceptance testing, annual system evaluation, and phantom testing for accreditation or renewal. All measurement processes, parameter calculations, decision criteria, and reporting formats rigorously follow the ACR specifications, described in the website package https://accreditationsupport.acr.org/support/solutions/articles/11000063276-complete-accreditation-information-mri. In the analysis module, the images are identified with a convolutional neural network (CNN) to address MRI scanner platform-specific dependencies (such as acquisition order and labeling). The measurements in images are processed using advanced image processing algorithms (e.g. shape features and ROIs). Parameters can be retrieved from the database and analyzed for trends. The software was developed on Matlab platform and compiled as a standalone program for PC and Mac systems.
By using this software tool, the physicist’s workflow becomes more convenient and better organized. Automated evaluation of ROI measurements is more consistent than manual one, mainly due to better consistency of ROI selection. The accuracy of measurements for Image-Intensity-Uniformity, Percent-Signal-Ghosting and Signal-to-Noise-Ratio is significantly improved. Reports generated using the software are standardized for all physicists and all MRI scanners within the entire medical system enterprise.
The software represents a significant improvement on data management and processing for routine physicist’s tasks within the scope of the ACR MRAP. When the stable version becomes available, we plan to make it available under open license to the medical physics community.