A Ratiometric Method for Assessing Multi-Modal Breast MRI Image Quality
B Byrd1,2*, V Krishnaswamy2, J Gui3, M Fox2,4, K Paulsen1,2, R Barth2,4, T Rooney2,5, (1) Thayer School of Engineering at Dartmouth College, Hanover, NH, (2) CairnSurgical, LLC, Lebanon, NH, (3) Geisel School Of Medicine, Hanover, NH (4) Dartmouth-Hitchcock Medical Center, Lebanon, NH (5) University of Virginia Health, Charlottesville, VA
Presentations
SU-E-207-4 (Sunday, 7/10/2022) 1:00 PM - 2:00 PM [Eastern Time (GMT-4)]
Room 207
Purpose: To quantitatively compare image contrast between different contrast-enhanced MR sequences and patient orientations for the specific purpose of tumor localization during breast conserving surgery (BCS).
Methods: Participating patients were entered into one of two prospective experimental studies approved by the Dartmouth Committee for the Protection of Human Subjects. The control patient cohort (n = 17) received the standard-of-care, T1-weighted prone and supine imaging on separate days (ClinicalTrials.gov #NCT03573661). The prone-to-supine (P2S) patient cohort (n = 61) received T1-weighted prone MRI imaging, followed immediately by delayed-supine MR imaging in the same session (single bolus injection) (ClinicalTrials.gov #NCT03573804).Tumor regions were segmented by a radiologist and normal boundary tissue of equal volume was defined around the segmented tumor region. The boundary region was further classified into adipose and glandular regions. Through using normalized contrast metric measurements for regional tumor contrast (i.e. tumor-to-boundary, tumor-to-adipose, and tumor-to-fibrous tissue contrast), image contrast metrics were compared for statistical differences/non-inferiority across three cohorts consisting of control prone imaging (n = 78), control supine imaging (n = 16), and P2S delayed-supine imaging (n = 61).
Results: By comparing normalized contrast measurements, control supine image contrast was found to be non-inferior to control prone in all contrast metrics examined with a non-inferiority margin of 0.1 normalized contrast units. The major difference in the P2S delayed-supine cohort was the tumor-to-fibrous tissue contrast when compared to control prone (Two-tailed t-test p < 0.001, PHP>0.99) and control supine datasets (Two-tailed t-test p=0.014, PHP = 0.76).
Conclusion: Ratiometric contrast measurements enable quantitative comparisons of regional tumor contrast which are system- and modality-independent. In this study, quantitative contrast comparisons reveal that the delayed-supine imaging cohort had inferior tumor-to-fibrous tissue contrast, while the control supine cohort possessed non-inferior image contrast metrics, when compared to the gold-standard, prone MRI cohort.
Funding Support, Disclosures, and Conflict of Interest: V. Krisnaswamy, K. Paulsen, and R. Barth are co-founders of CairnSurgical LLC. and have owndership in the company. B. Byrd, M. Fix, and T. Rooney are employees of CairnSurgical.
Keywords
Breast, MRI, Image-guided Surgery
Taxonomy
IM- MRI : Quality Control and Image Quality Assessment
Contact Email