Premium
SU‐G‐JeP2‐08: Image‐Guided Radiation Therapy Using Synthetic CTs in Brain Cancer
Author(s) -
Price R.G.,
Kim J.,
Zheng W.,
Chetty I.J.,
GlideHurst C.
Publication year - 2016
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4957028
Subject(s) - image guided radiation therapy , imaging phantom , image registration , nuclear medicine , voxel , medicine , cone beam computed tomography , medical imaging , computer science , medical physics , artificial intelligence , computed tomography , radiology , image (mathematics)
Purpose: Synthetic‐CTs(synCTs) are essential for MR‐only treatment planning. However, the performance of synCT for IGRT must be carefully assessed. This work evaluated the accuracy of synCT and synCT‐generated DRRs and determined their performance for IGRT in brain cancer radiation therapy. Methods: MR‐SIM and CT‐SIM images were acquired of a novel anthropomorphic phantom and a cohort of 12 patients. SynCTs were generated by combining an ultra‐short echo time (UTE) sequence with other MRI datasets using voxel‐based weighted summation. For the phantom, DRRs from synCT and CT were compared via bounding box and landmark analysis. Planar (MV/KV) and volumetric (CBCT) IGRT performance were evaluated across several platforms. In patients, retrospective analysis was conducted to register CBCTs (n=34) to synCTs and CTs using automated rigid registration in the treatment planning system using whole brain and local registration techniques. A semi‐automatic registration program was developed and validated to rigidly register planar MV/KV images (n=37) to synCT and CT DRRs. Registration reproducibility was assessed and margin differences were characterized using the van Herk formalism. Results: Bounding box and landmark analysis of phantom synCT DRRs were within 1mm of CT DRRs. Absolute 2D/2D registration shift differences ranged from 0.0–0.7mm for phantom DRRs on all treatment platforms and 0.0–0.4mm for volumetric registrations. For patient planar registrations, mean shift differences were 0.4±0.5mm (range: −0.6–1.6mm), 0.0±0.5mm, (range: −0.9–1.2mm), and 0.1±0.3mm (range: −0.7–0.6mm) for the superior‐inferior(S‐I), left‐right(L–R), and anterior‐posterior(A‐P) axes, respectively. Mean shift differences in volumetric registrations were 0.6±0.4mm (range: −0.2–1.6mm), 0.2±0.4mm (range: −0.3–1.2mm), and 0.2±0.3mm (range: −0.2–1.2mm) for S‐I, L–R, and A–P axes, respectively. CT‐SIM and synCT derived margins were within 0.3mm. Conclusion: DRRs generated via synCT agreed well with CT‐SIM. Planar and volumetric registrations to synCT‐derived targets were comparable to CT. This validation is the next step toward clinical implementation of MR‐only planning for the brain. The submitting institution has research agreements with Philips Healthcare. Research sponsored by a Henry Ford Health System Internal Mentored Grant.