Multi-Institutional Development and Validation of Consensus Electron Density Look-up Tables for MR-Only Radiotherapy Planning
Abstract
Purpose
To develop and validate consensus electron density (ED) look-up tables (LUTs) for MR-only radiotherapy planning through multi-institutional collaboration.
Methods
Bulk relative ED (rED) values were extracted from 1,165 patients (711 from UT Southwestern, 454 from MD Anderson) who were treated on Elekta Unity MR-Linac systems. Two consensus LUT approaches were developed: Complex-rED with structure-specific assignments for 50 anatomical structures and Simple-rED with five tissue categories (air cavity, lung, soft tissue, soft bone, hard bone). Dosimetric validation was performed on 63 patients across four anatomical sites (CNS, lung, abdomen, pelvis) using gamma analysis (3%/2mm, 2%/2mm, 1%/1mm) and dose-volume histogram comparisons against CT-based plans. Wilcoxon signed-rank test was used for paired statistical comparisons.
Results
Inter-institutional ED values demonstrated excellent agreement (mean difference 0%±3%, R²=0.98). Both Complex-rED and Simple-rED approaches achieved clinically acceptable dosimetric accuracy. Gamma analysis showed pass rates >99% at 3%/2mm for all structures, >98% at 2%/2mm for most structures, and >90% at 1%/1mm for most cases, with no significant differences between approaches (p>0.05). Compared to CT-based plans, absolute mean dose differences for PTVs remained ≤0.37Gy across all sites, with the largest deviations observed in lung cases. While 13% of PTV mean dose comparisons showed statistical significance (p<0.05), including pelvis (p=0.02), pancreas (p=0.0015), and liver (p=0.048), absolute differences remained clinically negligible (<0.2Gy, <0.5% of prescription). For OARs, 22% of mean dose comparisons showed statistical significance (p<0.05), primarily in gastrointestinal and thoracic structures, yet absolute dose differences remained well within clinical tolerance: mean dose <0.2Gy and maximum dose <1.0Gy.
Conclusion
Multi-institutional consensus ED LUTs provide clinically acceptable dosimetric accuracy for MR-only radiotherapy planning via bulk-density overrides. Both Complex-rED and Simple-rED approaches maintain target coverage and OAR sparing with differences well within established clinical tolerances. The Simple-rED approach offers practical implementation advantages without compromising accuracy, supporting streamlined clinical adoption of MR-only workflows.