Purpose: Breast conserving surgery has a high positive margin rate leading to 15% of patients requiring a second surgery. Clinically accepted tools such as standard specimen x-ray have severe limitations. This study aims to show that micro-CT can improve malignancy detection across the entire surface of the resected specimen without significantly changing clinical workflow.
Methods: Specimens from 33 patients undergoing breast conserving surgery were imaged as a part of a clinical trial at the time of surgery. Micro-CT scans were prospectively analyzed by two radiologists to determine margin status and tumor-to-edge distances for all six anatomic margins as well as the primary tumor diameter with a time-constraint of five minutes to mimic clinical conditions.
Results: Micro-CT imaging of breast cancer specimens was done in 3-4 minutes and scans were analyzed in under five minutes. Sensitivity, specificity, and accuracy compared to pathology were 28.6-57.1%, 94.2%-90.1%, and 91.9-88.9% respectively. This can be compared to standard specimen x-ray imaging on the same data which gives values of 29%, 98%, and 95%. The primary tumor diameter obtained from the micro-CT scans compared to measurements in Pathology showed a Pearson Correlation Coefficient of 0.61 or 0.43 depending on the radiologist.
Conclusion: This is the first study comparing micro-CT scans taken at the time of surgery to standard-of-care intraoperative tools. This study was designed to mimic clinical conditions, with imaging and analysis taking less than 10 minutes. This is comparable to the timing of current tools. The results are comparable to standard 2D specimen x-ray but with the ability to determine margin status across the entire tissue surface. There is inter radiologist variation in the analysis of these micro-CT scans. This could be due to a lack of training data for this novel modality.