RESUMO
PURPOSE: Glioblastoma changes during chemoradiation therapy are inferred from magnetic resonance imaging (MRI) before and after treatment but are rarely investigated due to logistics of frequent MRI. Using a combination MRI-linear accelerator (MRI-linac), we evaluated changes during daily chemoradiation therapy. METHODS AND MATERIALS: Patients with glioblastoma were prospectively imaged daily during chemoradiation therapy on 0.35T MRI-linac and at 3 timepoints with and without contrast on standalone high-field MRI. Tumor or edema (lesion) and resection cavity dynamics throughout treatment were analyzed and compared with standalone T1 postcontrast (T1+C) and T2 volumes. RESULTS: Of 36 patients included in this analysis, 8 had cavity only, 12 had lesion only, and 16 had both cavity and lesion. Of these, 64% had lesion growth and 46% had cavity shrinkage during treatment on MRI-linac scans. The average MRI-linac migration distance was 1.3 cm (range, 0-4.1 cm) for lesion and 0.6 cm (range, 0.1-2.1 cm) for cavity. Standalone versus MRI-linac volumes correlated strongly with R2 values: 0.991 (T2 vs MRI-linac cavity), 0.972 (T1+C vs MRI-linac cavity), and 0.973 (T2 vs MRI-linac lesion). There was a moderate correlation between T1+C and MRI-linac lesion (R2 = 0.609), despite noncontrast MRI-linac inability to separate contrast enhancement from surrounding nonenhancing tumor and edema. From pretreatment to posttreatment in patients with all available scans (n = 35), T1+C and MRI-linac lesions changed together-shrank (n = 6), grew (n = 12), or unchanged (n = 8)-in 26 (74%) patients. Another 9 patients (26%) had growth on MRI-linac, although the T1+C component shrank. In no patient did T1+C lesion grow while MRI-linac lesion shrank. CONCLUSIONS: Anatomic changes are seen in patients with glioblastoma imaged daily on MRI-linac throughout the chemoradiation therapy course. As surgical resection cavities shrink, margins may be reduced to save normal brain. Patients with unresected or growing lesions may require margin expansions to cover changes. Limited volume glioblastoma boost trials could consider triggered gadolinium contrast administration for evaluation of adaptive radiation therapy when lesion growth is seen on noncontrast MRI-linac.
RESUMO
Glioblastoma changes during chemoradiotherapy are inferred from high-field MRI before and after treatment but are rarely investigated during radiotherapy. The purpose of this study was to develop a deep learning network to automatically segment glioblastoma tumors on daily treatment set-up scans from the first glioblastoma patients treated on MRI-linac. Glioblastoma patients were prospectively imaged daily during chemoradiotherapy on 0.35T MRI-linac. Tumor and edema (tumor lesion) and resection cavity kinetics throughout the treatment were manually segmented on these daily MRI. Utilizing a convolutional neural network, an automatic segmentation deep learning network was built. A nine-fold cross-validation schema was used to train the network using 80:10:10 for training, validation, and testing. Thirty-six glioblastoma patients were imaged pre-treatment and 30 times during radiotherapy (n = 31 volumes, total of 930 MRIs). The average tumor lesion and resection cavity volumes were 94.56 ± 64.68 cc and 72.44 ± 35.08 cc, respectively. The average Dice similarity coefficient between manual and auto-segmentation for tumor lesion and resection cavity across all patients was 0.67 and 0.84, respectively. This is the first brain lesion segmentation network developed for MRI-linac. The network performed comparably to the only other published network for auto-segmentation of post-operative glioblastoma lesions. Segmented volumes can be utilized for adaptive radiotherapy and propagated across multiple MRI contrasts to create a prognostic model for glioblastoma based on multiparametric MRI.