Search

Purpose/Objective(s) Daily Cone beam CT (CBCT) imaging provides necessary anatomical information for accurate patient setup. Image quality of CBCT is usually far inferior to simulation CT scans. A workaround is to register the CT to the CBCT such that the contours and Hounsfield Unit (HU) values of the CT can be propagated to the CBCT. However, the inconsistent HU values across CT and CBCT make it less effective to use conventional image similarity measures. We aim to develop an unsupervised registration network to overcome this challenge in multimodal CT-CBCT image registration. Materials/Methods We propose to integrate directional local structural similarity into an unsupervised learning framework to perform abdominal CT-CBCT image registration. Directional local structural similarity measures the image's self-similarity which reflects the underlying structural similarity regardless of the modality in use. The CBCT and CT images were separately processed to extract directional local structural similarity feature maps in different directions. We concatenated the directional local structural similarity feature maps and the original images as network input. Taking both the original images and their respective structural similarity feature maps as input allows the network to fully explore the potential correlations between CBCT and CT for accurate deformation vector field (DVF) prediction. Salient features learnt through previous iterations were highlighted by attention gates across layers to expedite the learning process. A 3D bicubic interpolation was used to up-sample and smooth the predicted DVF. We performed a leave-one-out cross validation with an image dataset of 45 patients to evaluate the proposed registration method. Normalized cross correlation (NCC) and target registration error (TRE) between CBCT and deformed CT were calculated to quantify the registration accuracy. Results Our results show that the alignment between the abdominal soft tissues has been greatly improved after registration for all patients. The mean and standard deviation of NCC and TRE were 0.97 (range 0.95-0.99) and 1.88 (range 1.03-2.67) mm. The proposed network allows for many datasets to be used as training datasets since ground truth DVF is not needed for the training process. The proposed network can predict a final DVF via a single forward prediction, which is faster than the conventional iterative registration algorithms. Conclusion We have developed a novel unsupervised multimodal image registration method for CT-CBCT abdominal image registration, which does not need ground truth DVF for training, and demonstrated its feasibility. Taking both CBCT and CT images and their respective directional local structural similarity features as input, the proposed network performs direct DVF prediction to register the abdominal CT to CBCT images. This tool will be useful for future CBCT-guided radiotherapy of abdominal malignancies.

To determine the dosimetric effects of rotational errors on target coverage using volumetric modulated arc therapy (VMAT) for multitarget stereotactic radiosurgery (SRS).This retrospective study included 50 SRS cases, each with 2 intracranial planning target volumes (PTVs). Both PTVs were planned for simultaneous treatment to 21 Gy using a single-isocenter, noncoplanar VMAT SRS technique. Rotational errors of 0.5°, 1.0°, and 2.0° were simulated about all axes. The dose to 95% of the PTV (D95) and the volume covered by 95% of the prescribed dose (V95) were evaluated using multivariate analysis to determine how PTV coverage was related to PTV volume, PTV separation, and rotational error.At 0.5° rotational error, D95 values and V95 coverage rates were ≥95% in all cases. For rotational errors of 1.0°, 7% of targets had D95 and V95 values95%. Coverage worsened substantially when the rotational error increased to 2.0°: D95 and V95 values were95% for only 63% of the targets. Multivariate analysis showed that PTV volume and distance to isocenter were strong predictors of target coverage.The effects of rotational errors on target coverage were studied across a broad range of SRS cases. In general, the risk of compromised coverage increased with decreasing target volume, increasing rotational error and increasing distance between targets. Multivariate regression models from this study may be used to quantify the dosimetric effects of rotational errors on target coverage given patient-specific input parameters of PTV volume and distance to isocenter.

Purpose To describe the use of radiosurgery (RS) alone to the resection cavity after resection of brain metastases as an alternative to adjuvant whole-brain radiotherapy (WBRT). Methods and Materials Sixty-two patients with 64 cavities were treated with linear accelerator–based RS alone to the resection cavity after surgical removal of brain metastases between March 2007 and August 2010. Fifty-two patients (81%) had a gross total resection. Median cavity volume was 8.5 cm 3 . Forty-four patients (71%) had a single metastasis. Median marginal and maximum doses were 18 Gy and 20.4 Gy, respectively. Sixty-one cavities (95%) had gross tumor volume to planning target volume expansion of ≥1 mm. Results Six-month and 1-year actuarial local recurrence rates were 14% and 22%, respectively, with a median follow-up period of 9.7 months. Six-month and 1-year actuarial distant brain recurrence, total intracranial recurrence, and freedom from WBRT rates were 31% and 51%, 41% and 63%, and 91% and 74%, respectively. The symptomatic cavity radiation necrosis rate was 8%, with 2 patients (3%) undergoing surgery. Of the 11 local failures, 8 were in-field, 1 was marginal, and 2 were both (defined as in-field if ≥90% of recurrence within the prescription isodose and marginal if ≥90% outside of the prescription isodose). Conclusions The high rate of in-field cavity failure suggests that geographic misses with highly conformal RS are not a major contributor to local recurrence. The current dosing regimen derived from Radiation Therapy Oncology Group protocol 90-05 should be optimized in this patient population before any direct comparison with WBRT.

Purpose Computed tomography (CT) artifacts can severely degrade dose calculation accuracy in proton therapy. Prompted by the recently increased popularity of magnetic resonance imaging (MRI) in the radiation therapy clinic, we developed an MRI-based CT artifact correction method for improving the accuracy of proton range calculations. Methods and materials The proposed method replaces corrupted CT data by mapping CT Hounsfield units (HU number) from a nearby artifact-free slice, using a coregistered MRI. MRI and CT volumetric images were registered with use of 3-dimensional (3D) deformable image registration (DIR). The registration was fine-tuned on a slice-by-slice basis by using 2D DIR. Based on the intensity of paired MRI pixel values and HU from an artifact-free slice, we performed a comprehensive analysis to predict the correct HU for the corrupted region. For a proof-of-concept validation, metal artifacts were simulated on a reference data set. Proton range was calculated using reference, artifactual, and corrected images to quantify the reduction in proton range error. The correction method was applied to 4 unique clinical cases. Results The correction method resulted in substantial artifact reduction, both quantitatively and qualitatively. On respective simulated brain and head and neck CT images, the mean error was reduced from 495 and 370 HU to 108 and 92 HU after correction. Correspondingly, the absolute mean proton range errors of 2.4 cm and 1.7 cm were reduced to less than 2 mm in both cases. Conclusions Our MRI-based CT artifact correction method can improve CT image quality and proton range calculation accuracy for patients with severe CT artifacts.