Patricio D. Simari

Patient geometry-driven information retrieval for IMRT treatment plan quality control.

PURPOSE Intensity modulated radiation therapy (IMRT) treatment plan quality depends on the planners level of experience and the amount of time the planner invests in developing the plan. Planners often unwittingly accept plans when further sparing of the organs at risk (OARs) is possible. The authors propose a method of IMRT treatment plan quality control that helps planners to evaluate the doses of the OARs upon completion of a new plan. METHODS It is achieved by comparing the geometric configurations of the OARs and targets of a new patient with those of prior patients, whose plans are maintained in a database. They introduce the concept of a shape relationship descriptor and, specifically, the overlap volume histogram (OVH) to describe the spatial configuration of an OAR with respect to a target. The OVH provides a way to infer the likely DVHs of the OARs by comparing the relative spatial configurations between patients. A database of prior patients is built to serve as an external reference. At the conclusion of a new plan, planners search through the database and identify related patients by comparing the OAR-target geometric relationships of the new patient with those of prior patients. The treatment plans of these related patients are retrieved from the database and guide planners in determining whether lower doses delivered to the OARs in the new plan are feasible. RESULTS Preliminary evaluation is promising. In this evaluation, they applied the analysis to the parotid DVHs of 32 prior head-and-neck patients, whose plans are maintained in a database. Each parotid was queried against the other 63 parotids to determine whether a lower dose was possible. The 17 parotids that promised the greatest reduction in D50 (DVH dose at 50% volume) were flagged. These 17 parotids came from 13 patients. The method also indicated that the doses of the other nine parotids of the 13 patients could not be reduced, so they were included in the replanning process as controls. Replanning with an effort to reduce D50 was conducted on these 26 parotids. After replanning, the average reductions for D50 of the 17 flagged parotids and nine unflagged parotids were 6.6 and 1.9 Gy, respectively. These results demonstrate that the quality control method has accurately identified not only the parotids that require dose reductions but also those for which dose reductions are marginal. Originally, 11 of out the 17 flagged parotids did not meet the Radiation Therapy Oncology Group sparing goal of V(30 Gy) < 50%. Replanning reduced them to three. Additionally, PTV coverage and OAR sparing of the original plans were compared to those of the replans by using pairwise Wilcoxon p test. The statistical comparisons show that replanning compromised neither PTV coverage nor OAR sparing. CONCLUSIONS This method provides an effective quality control mechanism for evaluating the DVHs of the OARs. Adoption of such a method will advance the quality of current IMRT planning, providing better treatment plan consistency.

Folding meshes: hierarchical mesh segmentation based on planar symmetry

Meshes representing real world objects, both artist-created and scanned, contain a high level of redundancy due to (possibly approximate) planar reflection symmetries, either global or localized to different subregions. An algorithm is presented for detecting such symmetries and segmenting the mesh into the symmetric and remaining regions. The method, inspired by techniques in Computer Vision, has foundations in robust statistics and is resilient to structured outliers which are present in the form of the non symmetric regions of the data. Also introduced is an application of the method: the folding tree data structure. The structure encodes the non redundant regions of the original mesh as well as the reflection planes and is created by the recursive application of the detection method. This structure can then be unfolded to recover the original shape. Applications include mesh compression, repair, skeletal extraction of objects of known symmetry as well as mesh processing acceleration by limiting computation to non redundant regions and propagation of results.

Robust statistical estimation of curvature on discretized surfaces

A robust statistics approach to curvature estimation on discretely sampled surfaces, namely polygon meshes and point clouds, is presented. The method exhibits accuracy, stability and consistency even for noisy, non-uniformly sampled surfaces with irregular configurations. Within an M-estimation framework, the algorithm is able to reject noise and structured outliers by sampling normal variations in an adaptively reweighted neighborhood around each point. The algorithm can be used to reliably derive higher order differential attributes and even correct noisy surface normals while preserving the fine features of the normal and curvature field. The approach is compared with state-of-the-art curvature estimation methods and shown to improve accuracy by up to an order of magnitude across ground truth test surfaces under varying tessellation densities and types as well as increasing degrees of noise. Finally, the benefits of a robust statistical estimation of curvature are illustrated by applying it to the popular applications of mesh segmentation and suggestive contour rendering.

Increased organ sparing using shape-based treatment plan optimization for intensity modulated radiation therapy of pancreatic adenocarcinoma

PURPOSE To develop a model to assess the quality of an IMRT treatment plan using data of prior patients with pancreatic adenocarcinoma. METHODS The dose to an organ at risk (OAR) depends in large part on its orientation and distance to the planning target volume (PTV). A database of 33 previously treated patients with pancreatic cancer was queried to find patients with less favorable PTV-OAR configuration than a new case. The minimal achieved dose among the selected patients should also be achievable for the OAR of the new case. This way the achievable doses to the OARs of 25 randomly selected pancreas cancer patients were predicted. The patients were replanned to verify if the predicted dose could be achieved. The new plans were compared to their original clinical plans. RESULTS The predicted doses were achieved within 1 and 2 Gy for more than 82% and 94% of the patients, respectively, and were a good approximation of the minimal achievable doses. The improvement after replanning was 1.4 Gy (range 0-4.6 Gy) and 1.7 Gy (range 0-6.3 Gy) for the mean dose to the liver and the kidneys, respectively, without compromising target coverage or increasing radiation dose to the bowel, cord or stomach. CONCLUSIONS The model could accurately predict the achievable doses, leading to a considerable decrease in dose to the OARs and an increase in treatment planning efficiency.

Fully Automated Simultaneous Integrated Boosted–Intensity Modulated Radiation Therapy Treatment Planning Is Feasible for Head-and-Neck Cancer: A Prospective Clinical Study

PURPOSE To prospectively determine whether overlap volume histogram (OVH)-driven, automated simultaneous integrated boosted (SIB)-intensity-modulated radiation therapy (IMRT) treatment planning for head-and-neck cancer can be implemented in clinics. METHODS AND MATERIALS A prospective study was designed to compare fully automated plans (APs) created by an OVH-driven, automated planning application with clinical plans (CPs) created by dosimetrists in a 3-dose-level (70 Gy, 63 Gy, and 58.1 Gy), head-and-neck SIB-IMRT planning. Because primary organ sparing (cord, brain, brainstem, mandible, and optic nerve/chiasm) always received the highest priority in clinical planning, the study aimed to show the noninferiority of APs with respect to PTV coverage and secondary organ sparing (parotid, brachial plexus, esophagus, larynx, inner ear, and oral mucosa). The sample size was determined a priori by a superiority hypothesis test that had 85% power to detect a 4% dose decrease in secondary organ sparing with a 2-sided alpha level of 0.05. A generalized estimating equation (GEE) regression model was used for statistical comparison. RESULTS Forty consecutive patients were accrued from July to December 2010. GEE analysis indicated that in APs, overall average dose to the secondary organs was reduced by 1.16 (95% CI = 0.09-2.33) with P=.04, overall average PTV coverage was increased by 0.26% (95% CI = 0.06-0.47) with P=.02 and overall average dose to the primary organs was reduced by 1.14 Gy (95% CI = 0.45-1.8) with P=.004. A physician determined that all APs could be delivered to patients, and APs were clinically superior in 27 of 40 cases. CONCLUSIONS The application can be implemented in clinics as a fast, reliable, and consistent way of generating plans that need only minor adjustments to meet specific clinical needs.

Extracting lines of curvature from noisy point clouds

We present a robust framework for extracting lines of curvature from point clouds. First, we show a novel approach to denoising the input point cloud using robust statistical estimates of surface normal and curvature which automatically rejects outliers and corrects points by energy minimization. Then the lines of curvature are constructed on the point cloud with controllable density. Our approach is applicable to surfaces of arbitrary genus, with or without boundaries, and is statistically robust to noise and outliers while preserving sharp surface features. We show our approach to be effective over a range of synthetic and real-world input datasets with varying amounts of noise and outliers. The extraction of curvature information can benefit many applications in CAD, computer vision and graphics for point cloud shape analysis, recognition and segmentation. Here, we show the possibility of using the lines of curvature for feature-preserving mesh construction directly from noisy point clouds.

A Shape Relationship Descriptor for Radiation Therapy Planning

In this paper we address the challenge of matching patient geometry to facilitate the design of patient treatment plans in radiotherapy. To this end we propose a novel shape descriptor, the Overlap Volume Histogram, which provides a rotation and translation invariant representation of a patients organs at risk relative to the tumor volume. Using our descriptor, it is possible to accurately identify database patients with similar constellations of organ and tumor geometries, enabling the transfer of treatment plans between patients with similar geometries, We demonstrate the utility of our method for such tasks by outperforming state of the art shape descriptors in the retrieval of patients with similar treatment plans. We also preliminarily show its potential as a quality control tool by demonstrating how it is used to identify an organ at risk whose dose can be significantly reduced.

Using overlap volume histogram and IMRT plan data to guide and automate VMAT planning: A head‐and‐neck case study

PURPOSE To investigate whether an overlap volume histogram (OVH)-driven planning application using an intensity-modulated radiation therapy (IMRT) database can guide and automate volumetric-modulated arc therapy (VMAT) planning for head-and-neck cancer. METHODS Based on comparable head-and-neck dosimetric results between planner-generated VMAT and IMRT plans, an inhouse developed, OVH-driven automated planning application containing a database of prior clinical head-and-neck IMRT plans is built into Pinnacle(3) SmartArc for VMAT planning. Double-arc VMAT plans of four oropharynx, four nasopharynx, and four larynx patients are generated and compared with corresponding clinical IMRT plans. RESULTS Each VMAT plan is automatically generated in two optimization rounds, while the average number of optimization rounds in generating a clinical IMRT plan is 43. In VMAT plans, statistical superiority (p < 0.01) in sparing of the cord+4 mm, brainstem, brachial plexus, larynx, and inner ear is observed with a slight degradation in low-dose-level planning target volume (PTV) coverage. On average, D(0.1 cc) to the cord+4 mm, brainstem and brachial plexus is reduced by 3.7, 4.9, and 1.6 Gy, respectively; V(50 Gy) to the larynx is reduced by 5.3%; mean dose to the inner ear is reduced by 4.4 Gy; V(95) of low-dose-level PTV coverage is reduced by 0.3% with p = 0.25. CONCLUSIONS IMRT-data-driven VMAT planning offers a potential method for generating VMAT plans that are comparable to IMRT plans in terms of dosimetric quality.

Multi-objective shape segmentation and labeling

Shape segmentations designed for different applications show significant variation in the composition of their parts. In this paper, we introduce the segmentation and labeling of shape based on the simultaneous optimization of multiple heterogenous objectives that capture application‐specific segmentation criteria. We present a number of efficient objective functions that capture useful shape adjectives (compact, flat, narrow, perpendicular, etc.) Segmentation descriptions within our framework combine multiple such objective functions with optional labels to define each part. The optimization problem is simplified by proposing weighted Voronoi partitioning as a compact and continuous parametrization of spatially embedded shape segmentations. Separation of spatially close but geodesically distant parts is made possible using multi‐dimensional scaling prior to Voronoi partitioning. Optimization begins with an initial segmentation found using the centroids of a k‐means clustering of surface elements. This partition is automatically labeled to optimize heterogeneous part objectives and the Voronoi centers and their weights optimized using Generalized Pattern Search. We illustrate our framework using several diverse segmentation applications: consistent segmentations with semantic labels, bounding volume hierarchies for path tracing, and automatic rig and clothing transfer between animation characters.

Data-driven curvature for real-time line drawing of dynamic scenes

This article presents a method for real-time line drawing of deforming objects. Object-space line drawing algorithms for many types of curves, including suggestive contours, highlights, ridges, and valleys, rely on surface curvature and curvature derivatives. Unfortunately, these curvatures and their derivatives cannot be computed in real-time for animated, deforming objects. In a preprocessing step, our method learns the mapping from a low-dimensional set of animation parameters (e.g., joint angles) to surface curvatures for a deforming 3D mesh. The learned model can then accurately and efficiently predict curvatures and their derivatives, enabling real-time object-space rendering of suggestive contours and other such curves. This represents an order-of-magnitude speedup over the fastest existing algorithm capable of estimating curvatures and their derivatives accurately enough for many different types of line drawings. The learned model can generalize to novel animation sequences and is also very compact, typically requiring a few megabytes of storage at runtime. We demonstrate our method for various types of animated objects, including skeleton-based characters, cloth simulation, and blend-shape facial animation, using a variety of nonphotorealistic rendering styles. An important component of our system is the use of dimensionality reduction for differential mesh data. We show that Independent Component Analysis (ICA) yields localized basis functions, and gives superior generalization performance to that of Principal Component Analysis (PCA).