Real-Time Quality Control of RF Coils Using Clinical MRI Data
T McKeown1*, S Robertson2*, E Samei2, (1) Duke University, Durham, NC, (2) Duke University Health System, Durham, NC
Presentations
PO-GePV-I-25 (Sunday, 7/10/2022) [Eastern Time (GMT-4)]
ePoster Forums
Purpose: To develop an automated method for quantifying RF coil performance using raw clinical MRI data.
Methods: For one month, raw MRI data was collected from every localizer series acquired on three 3T and two 1.5T Siemens MRI scanners. All data was transferred to a network drive using the Yarra Raw Data Storage Service. All processing was performed by custom MATLAB software. Each dataset was reconstructed into three image sets: individual coil channel images, a sum-of-squares image, and individual coil sensitivities. Using the joint information contained in these three image sets, an automatic segmentation identified regions representing signal and noise. For each coil channel, SNR was quantified as the ratio of the highest local signal mean to the noise standard deviation. The SNR was multiplied by a scaling factor to account for differences between localizer acquisition parameters. The mean signal, noise standard deviation, and SNR of each coil channel were stored in a database. Trends of individual coil performance were used to identify faulty elements.
Results: Our automated quality control program has successfully processed 750 patient localizers across 29 coils on 5 different MRI machines. The software robustly reconstructed imaging data with various acquisition strategies including partial Fourier and parallel imaging. Our segmentation remained accurate across a diverse patient population and a large variety of anatomical regions. In the first month of usage, this QC program detected faulty coil elements on three separate RF coils. The faulty elements were validated using vendor-provided testing software.
Conclusion: This work demonstrates that RF coil performance can be quantified directly from raw clinical data. Avoiding additional phantom testing saved valuable staff and scanner time, and the continuous stream of clinical data allowed RF coil issues to be immediately identified rather than waiting for periodic physics testing.
Keywords
Taxonomy
IM- MRI : Quality Control and Image Quality Assessment
Contact Email