Tudda, A. and Castriconi, R. and Dusi, F. and Scaggion, A. (2023) Quantifying multi-Institution KB-plan prediction models transferability in breast cancer radiotherapy: A step forward toward benchmark. Il nuovo cimento C, 46 (4). pp. 1-4. ISSN 1826-9885
|
Text
ncc12703.pdf - Published Version Download (138kB) | Preview |
Abstract
Ten Institutes (INST1-INST10) of the MIKAPOCo consortium set their KB plan prediction models of whole breast irradiation (WBI) delivered with tangential fields (TF), by using RapidPlan (Varian Medical System, Inc.) and following the same criteria of contouring and plan model building and validation. DVH prediction bands of organs at risk (heart, ipsilateral lung, contralateral lung and contralateral breast) were exported on 20 new patients from the same Institutes (two patients each). SD of mean predicted DVHs among institutes was assessed as inter-institute variability. The estimated Principal Component (PC1) was con sidered for transferability models evaluation. Transferability cross-validation was further investigated in detail on a larger population for the model showing the poor est transferability (ModelINST6) against one of the models with high transferability (ModelINST3). Results show a limited inter-institute variability of plan prediction models (1.8% for DVH ipsilateral lung) and a satisfactory inter-institute transferability, excepting one institute, confirmed by the extended analyses on a larger cohort of test patients of INST6 (vs. INST3). These achievements pave the way for generating benchmarks for plan prediction in WBI with potentially relevant large-scale applications.
| Item Type: | Article |
|---|---|
| Subjects: | 500 Scienze naturali e Matematica > 530 Fisica |
| Depositing User: | Marina Spanti |
| Date Deposited: | 08 Mar 2024 13:51 |
| Last Modified: | 08 Mar 2024 13:51 |
| URI: | http://eprints.bice.rm.cnr.it/id/eprint/22719 |
Actions (login required)
 |
View Item |
