Automated Review of MR-Guided Stereotactic Body Radiation Therapy Treatments Using MR-Linac Delivery Log Files
B Hunt1,2*, G Gill2, M Rose2, P Schaner1,2, G Russo1,2, B Zaki1,2, D Gladstone1, R Zhang1,2, (1) Dartmouth College, Hanover, NH, (2) Dartmouth-Hitchcock Medical Center, Lebanon, NH
Presentations
PO-GePV-T-424 (Sunday, 7/10/2022) [Eastern Time (GMT-4)]
ePoster Forums
Purpose: Magnetic resonance guided stereotactic body radiotherapy (MRgSBRT) is an emerging method of treating mobile tumors. We hypothesized that automated analysis of MRgSBRT treatments using MR-guided linear accelerator delivery log files could provide valuable metrics for assessing operational efficiency of MRgSBRT services.
Methods: We conducted a retrospective review of delivery log files and videos from patients undergoing MRgSBRT in the first 20 months of our institutional implementation. Custom Python scripts were written to parse text log files and extract: date and time of treatment, site, fraction, couch position, adaptive plan utilization, time per beam segment delivered, and delivery completion status. Descriptive statistics were assessed for the study cohort.
Results: We analyzed log files over a span of 20.4 months which included 392 MRgSBRT treatment fractions from 83 patients. Over this span, median treatments per week was 4 with a maximum of 11. The distribution of treatment sites by fraction was 128 (33%) pancreas, 111 (28%) liver, 60 (15%) lung, 50 (12%) other, 20 (5%) prostate, 13 (3%) bone, and 10 (3%) kidney. Overall adaptive planning frequency was 44%, with the highest adaption rate for pancreas (81%) and lowest for prostate (0%). Median deviation in couch position relative to the initial fraction were 0.65, 0.77, and 0.28 cm for x, y, and z directions respectively and intrafractional couch adjustments were made in 48 (12%) fractions. Median duration of beam delivery for completed fractions was 14.7 minutes and varied by treatment site with 17.1, 16.5, 14.0, 13.3, 12.6, 10.9, and 10.0 minutes for pancreas, lung, liver, kidney, other, bone, and prostate respectively. Total delivery time was positively correlated with the number of beam segments delivered (Pearson’s correlation test, r=045, p<0.01).
Conclusion: Automated analysis of delivery logs can inform the adoption and improve operational efficiencies of MRgSBRT.
Keywords
Not Applicable / None Entered.
Taxonomy
TH- Dataset Analysis/Biomathematics: Informatics
Contact Email