Viability of Non-Coplanar VMAT for Liver SBRT as Compared to Coplanar VMAT and Beam Orientation Optimized 4π IMRT

Purpose

The 4π static noncoplanar radiation therapy delivery technique has demonstrated better normal tissue sparing and dose conformity than the clinically used volumetric modulated arc therapy (VMAT). It is unclear whether this is a fundamental limitation of VMAT delivery or the coplanar nature of its typical clinical plans. The dosimetry and the limits of normal tissue toxicity constrained dose escalation of coplanar VMAT, noncoplanar VMAT and 4π radiation therapy are quantified in this study.

Methods and materials

Clinical stereotactic body radiation therapy plans for 20 liver patients receiving 30 to 60 Gy using coplanar VMAT (cVMAT) were replanned using 3 to 4 partial noncoplanar arcs (nVMAT) and 4π with 20 intensity modulated noncoplanar fields. The conformity number, homogeneity index, 50% dose spillage volume, normal liver volume receiving >15 Gy, dose to organs at risk (OARs), and tumor control probability were compared for all 3 treatment plans. The maximum tolerable dose yielding a normal liver normal tissue control probability <1%, 5%, and 10% was calculated with the Lyman-Kutcher-Burman model for each plan as well as the resulting survival fractions at 1, 2, 3, and 4 years.

Results

Compared with cVMAT, the nVMAT and 4π plans reduced liver volume receiving >15 Gy by an average of 5 cm³ and 80 cm³, respectively. 4π reduced the 50% dose spillage volume by ∼23% compared with both VMAT plans, and either significantly decreased or maintained OAR doses. The 4π maximum tolerable doses and survival fractions were significantly higher than both cVMAT and nVMAT (P < .05) for all normal liver normal tissue control probability limits used in this study.

Conclusions

The 4π technique provides significantly better OAR sparing than both cVMAT and nVMAT and enables more clinically relevant dose escalation for tumor local control. Therefore, despite the current accessibility of nVMAT, it is not a viable alternative to 4π for liver SBRT.

Dose domain regularization of MLC leaf patterns for highly complex IMRT plans

PURPOSE

The advent of automated beam orientation and fluence optimization enables more complex intensity modulated radiation therapy (IMRT) planning using an increasing number of fields to exploit the expanded solution space. This has created a challenge in converting complex fluences to robust multileaf collimator (MLC) segments for delivery. A novel method to regularize the fluence map and simplify MLC segments is introduced to maximize delivery efficiency, accuracy, and plan quality.

METHODS

In this work, we implemented a novel approach to regularize optimized fluences in the dose domain. The treatment planning problem was formulated in an optimization framework to minimize the segmentation-induced dose distribution degradation subject to a total variation regularization to encourage piecewise smoothness in fluence maps. The optimization problem was solved using a first-order primal-dual algorithm known as the Chambolle-Pock algorithm. Plans for 2 GBM, 2 head and neck, and 2 lung patients were created using 20 automatically selected and optimized noncoplanar beams. The fluence was first regularized using Chambolle-Pock and then stratified into equal steps, and the MLC segments were calculated using a previously described level reducing method. Isolated apertures with sizes smaller than preset thresholds of 1-3 bixels, which are square units of an IMRT fluence map from MLC discretization, were removed from the MLC segments. Performance of the dose domain regularized (DDR) fluences was compared to direct stratification and direct MLC segmentation (DMS) of the fluences using level reduction without dose domain fluence regularization.

RESULTS

For all six cases, the DDR method increased the average planning target volume dose homogeneity (D95/D5) from 0.814 to 0.878 while maintaining equivalent dose to organs at risk (OARs). Regularized fluences were more robust to MLC sequencing, particularly to the stratification and small aperture removal. The maximum and mean aperture sizes using the DDR were consistently larger than those from DMS for all tested number of segments.

CONCLUSIONS

The fluence map to MLC segmentation conversion problem was formulated as a secondary optimization problem in the dose domain to minimize the smoothness-regularized dose discrepancy. The large scale optimization problem was solved using a primal-dual algorithm that transformed complicated fluences into maps that were more robust to the MLC segmentation and sequencing, affording fewer and larger segments with minimal degradation to dose distribution.

Iterative CT shading correction with no prior information

Shading artifacts in CT images are caused by scatter contamination, beam-hardening effect and other non-ideal imaging conditions. The purpose of this study is to propose a novel and general correction framework to eliminate low-frequency shading artifacts in CT images (e.g. cone-beam CT, low-kVp CT) without relying on prior information. The method is based on the general knowledge of the relatively uniform CT number distribution in one tissue component. The CT image is first segmented to construct a template image where each structure is filled with the same CT number of a specific tissue type. Then, by subtracting the ideal template from the CT image, the residual image from various error sources are generated. Since forward projection is an integration process, non-continuous shading artifacts in the image become continuous signals in a line integral. Thus, the residual image is forward projected and its line integral is low-pass filtered in order to estimate the error that causes shading artifacts. A compensation map is reconstructed from the filtered line integral error using a standard FDK algorithm and added back to the original image for shading correction. As the segmented image does not accurately depict a shaded CT image, the proposed scheme is iterated until the variation of the residual image is minimized. The proposed method is evaluated using cone-beam CT images of a Catphan©600 phantom and a pelvis patient, and low-kVp CT angiography images for carotid artery assessment. Compared with the CT image without correction, the proposed method reduces the overall CT number error from over 200 HU to be less than 30 HU and increases the spatial uniformity by a factor of 1.5. Low-contrast object is faithfully retained after the proposed correction. An effective iterative algorithm for shading correction in CT imaging is proposed that is only assisted by general anatomical information without relying on prior knowledge. The proposed method is thus practical and attractive as a general solution to CT shading correction.

A nonvoxel-based dose convolution/superposition algorithm optimized for scalable GPU architectures

PURPOSE

Real-time adaptive planning and treatment has been infeasible due in part to its high computational complexity. There have been many recent efforts to utilize graphics processing units (GPUs) to accelerate the computational performance and dose accuracy in radiation therapy. Data structure and memory access patterns are the key GPU factors that determine the computational performance and accuracy. In this paper, the authors present a nonvoxel-based (NVB) approach to maximize computational and memory access efficiency and throughput on the GPU.

METHODS

The proposed algorithm employs a ray-tracing mechanism to restructure the 3D data sets computed from the CT anatomy into a nonvoxel-based framework. In a process that takes only a few milliseconds of computing time, the algorithm restructured the data sets by ray-tracing through precalculated CT volumes to realign the coordinate system along the convolution direction, as defined by zenithal and azimuthal angles. During the ray-tracing step, the data were resampled according to radial sampling and parallel ray-spacing parameters making the algorithm independent of the original CT resolution. The nonvoxel-based algorithm presented in this paper also demonstrated a trade-off in computational performance and dose accuracy for different coordinate system configurations. In order to find the best balance between the computed speedup and the accuracy, the authors employed an exhaustive parameter search on all sampling parameters that defined the coordinate system configuration: zenithal, azimuthal, and radial sampling of the convolution algorithm, as well as the parallel ray spacing during ray tracing. The angular sampling parameters were varied between 4 and 48 discrete angles, while both radial sampling and parallel ray spacing were varied from 0.5 to 10 mm. The gamma distribution analysis method (γ) was used to compare the dose distributions using 2% and 2 mm dose difference and distance-to-agreement criteria, respectively. Accuracy was investigated using three distinct phantoms with varied geometries and heterogeneities and on a series of 14 segmented lung CT data sets. Performance gains were calculated using three 256 mm cube homogenous water phantoms, with isotropic voxel dimensions of 1, 2, and 4 mm.

RESULTS

The nonvoxel-based GPU algorithm was independent of the data size and provided significant computational gains over the CPU algorithm for large CT data sizes. The parameter search analysis also showed that the ray combination of 8 zenithal and 8 azimuthal angles along with 1 mm radial sampling and 2 mm parallel ray spacing maintained dose accuracy with greater than 99% of voxels passing the γ test. Combining the acceleration obtained from GPU parallelization with the sampling optimization, the authors achieved a total performance improvement factor of >175 000 when compared to our voxel-based ground truth CPU benchmark and a factor of 20 compared with a voxel-based GPU dose convolution method.

CONCLUSIONS

The nonvoxel-based convolution method yielded substantial performance improvements over a generic GPU implementation, while maintaining accuracy as compared to a CPU computed ground truth dose distribution. Such an algorithm can be a key contribution toward developing tools for adaptive radiation therapy systems.

Denoised and texture enhanced MVCT to improve soft tissue conspicuity

PURPOSE

MVCT images have been used in TomoTherapy treatment to align patients based on bony anatomies but its usefulness for soft tissue registration, delineation, and adaptive radiation therapy is limited due to insignificant photoelectric interaction components and the presence of noise resulting from low detector quantum efficiency of megavoltage x-rays. Algebraic reconstruction with sparsity regularizers as well as local denoising methods has not significantly improved the soft tissue conspicuity. The authors aim to utilize a nonlocal means denoising method and texture enhancement to recover the soft tissue information in MVCT (DeTECT).

METHODS

A block matching 3D (BM3D) algorithm was adapted to reduce the noise while keeping the texture information of the MVCT images. Following imaging denoising, a saliency map was created to further enhance visual conspicuity of low contrast structures. In this study, BM3D and saliency maps were applied to MVCT images of a CT imaging quality phantom, a head and neck, and four prostate patients. Following these steps, the contrast-to-noise ratios (CNRs) were quantified.

RESULTS

By applying BM3D denoising and saliency map, postprocessed MVCT images show remarkable improvements in imaging contrast without compromising resolution. For the head and neck patient, the difficult-to-see lymph nodes and vein in the carotid space in the original MVCT image became conspicuous in DeTECT. For the prostate patients, the ambiguous boundary between the bladder and the prostate in the original MVCT was clarified. The CNRs of phantom low contrast inserts were improved from 1.48 and 3.8 to 13.67 and 16.17, respectively. The CNRs of two regions-of-interest were improved from 1.5 and 3.17 to 3.14 and 15.76, respectively, for the head and neck patient. DeTECT also increased the CNR of prostate from 0.13 to 1.46 for the four prostate patients. The results are substantially better than a local denoising method using anisotropic diffusion.

CONCLUSIONS

The authors showed that it is feasible to extract more soft tissue contrast information from the noisy MVCT images using a nonlocal means 3D block matching method in combination with saliency maps, revealing information that was originally unperceivable to human observers.