A recent study was accepted and will be presented at the upcoming American Association of Physicists in Medicine (AAPM) conference, demonstrating utilization of MIM Maestro/ProtégéAI+ software to automatically detect the VeraForm marker and provide auto-contouring of tumor bed volume for radiation planning/treatment following lumpectomy. The study involved 5 radiation oncologists from different institutions, who manually outlined tumor bed volumes and compared it to the ones generated by the Maestro/ProtégéAI+ auto contour software.
‘We compared the outlines using the Dice Coefficient (DC), which measures how much the outlines overlap, and the Conformity Index (CI), which indicates how well the outlines match in shape and size, commented Brian Lawenda M.D. (GenesisCare), a radiation oncologist involved in the study. With outlines from 21 patients, we found no significant difference between the doctor's and software-generated outlines. This suggests the software can reliably standardize VeraForm demarcated TBV outlining, potentially making the contouring process more consistent and time efficient’, continued Dr. Lawenda.
‘With the VeraForm Marker and MIM Maestro/Protégé AI+, it may be possible to automatically contour the breast tumor bed following lumpectomy. This results in a greater standardization of the contour of planning target volumes (PTV) and reduces possible discrepancies in contouring by different radiation oncologists ‘continued Frank Rafie PhD., Clinical Associate Professor of Medical Physics at University of Washington, Dept of Radiation Oncology.
Standardization of radiation therapy planning and treatment for breast cancer patients has long been a goal for radiation oncologists. One of the key challenges, however, has been the reliable identification of the volume of the tumor bed. However, the introduction of VeraForm for tumor bed marking is helping to overcome that issue. With a unique imaging signature, different from bone, soft tissue and metal, VeraForm enables auto-detection and auto-contouring, thus minimizing the guess work during the treatment planning phase. As this study summarizes, the utilization of auto-contour software with VeraForm is on par with the physician-drawn approach and could pave the way towards better treatment standardization with less variability.
Novel Implementation of Target Auto-Contour to Standardize Breast Radiotherapy Treatment Planning
Dominic P. Rafie1, David C. Beyer1, Scott B. Schneider2, Brian D. Lawenda3, Chirag Shah4, Stephanie Rice5, Frank C. Rafie5
1Cancer Centers of Northern Arizona Healthcare – Sedona, AZ, 2Compass Oncology, Vancouver, WA, 3GenesisCare, Kennewick, WA, 4Taussig Cancer Institute, Cleveland Clinic, OH, 5Cancer Centers of Northern Arizona Healthcare – Flagstaff, AZ
Purpose: Contouring tumor bed volumes (TBV) is crucial for breast-conserving radiotherapy to ensure accurate target delineation. Surgical clip marker (CM) delineation lacks standardization, with high interobserver variability among physicians. Continuous 3D radiopaque filament markers (FM) woven into TBV outperform CM for delineation accuracy. It is possible to generate an FM contour using software-based autocontour workflow (AW) that may yield TBV contours comparable to physicians. This study examines whether there is a significant difference between contours of physicians and AW.
Methods: Five physicians were blinded to patient clinical history and independently contoured TBV on de-identified CTs with FDA-approved 3D continuous radiopaque FM. AW identified the “breast” contour and evaluated regions that matched the increased density of FM, separating those breast contour values to create FM contour. An average expansion value of 0.62 cm was applied to FM contour based on average TBV contoured by all physicians. A two-tailed paired sample t-test compared the average Dice Coefficient (DC) and Conformity Index (CI) between Physicians versus Physicians and AW versus Physicians. A P-value of < .05 indicated statistical significance.
Results: With 21 patients (122 contours), the average DC and CI for Physicians versus Physicians were 0.61 and 0.43 respectively. AW versus Physicians had an average DC and CI of 0.60 and 0.41 respectively. There was no statistically significant difference between analyzed groups for DC (P = .59) and CI (P = .32). All manual target contours by physicians and autocontours by AW resulted in acceptable treatment plans.
Conclusion: Findings demonstrate no statistically significant difference between TBV drawn by physicians and AW. Therefore, AW could be a tool for physicians that increases standardization of TBV contours. As some patients had both CM and FM, future goals are to examine patients only with FM and further develop AW.
Comments