We introduce a novel vascular skeleton extraction and decomposition technique for computer-assisted diagnosis and analysis. We start by addressing the problem of vascular decomposition as a cluster optimization problem and present a methodology for weighted convex approximations. Decomposed vessel structures are then grouped using the vessel skeleton, extracted using a Laplace-based operator. The method is validated using presegmented sections of vasculature archived for 98 aneurysms in 112 patients. We test first for vascular decomposition and next for vessel skeleton extraction. The proposed method produced promising results with an estimated 80.5% of the vessel sections correctly decomposed and 92.9% of the vessel sections having the correct number of skeletal branches, identified by a clinical radiological expert. Next, the method was validated on longitudinal study data from n = 4 subjects, where vascular skeleton extraction and decomposition was performed. Volumetric and surface area comparisons were made between expert segmented sections and the proposed approach on sections containing aneurysms. Results suggest that the method is able to detect changes in aneurysm volumes and surface areas close to that segmented by an expert.

In this work we propose a new Prediction from Expert Demonstration (PED) methodology to provide haptic assistance in robot assisted trocar surgery. Data was collected from expert (clinician) demonstrations for the procedure of trocar insertion. We encode a set of force, torque and penetration trajectories by using a Gaussian Mixture Model (GMM). A generalization of these profiles and associated parameters are retrieved by Gaussian Mixture Regression (GMR). A haptic assistance mode was devised to help novices perform the procedure based on the proposed PED model. We validated the methodology for surgical assistance on (n= 15) participants. The PED haptic model was tested for instrument deviation, penetration force and penetration depth. Preliminary study results showed that participants with PED haptic assistance performed the task with more consistency and exerted lesser penetration force than subjects without assistance.

In this paper, a shape decomposition methodology to decompose the cervical spine using an approximate convex methodology is proposed. The proposed methodology identifies the most concave L-ring neighborhood in a decomposition of a surface (manifold-2) and partitions it in order to reduce its concavity. A close approximation of the original surface by a set of convex surfaces hulls is then produced.

The generation of handwriting is a complex task. In order to accommodate for the large variations involved in handwritten words deformable templates need to be used. In this paper we propose a handwriting model, based on Active shape modeling (ASM). In a two-step generation process, a template-based ASM generates characters and a Gaussian mixture regression (GMR) model concatenates the generated characters. For real time generation of cursive handwriting an adaptation of Global local approximation (GLA) methodology is used to fit the generated models.

Computer based training has become an increasingly attractive alternative to traditional training methods for skill acquisition, and the topic of skill modeling has become one of great interest. One of the key problems in computer-based training is automatic skill evaluation, which requires precise and accurate skill modeling. In order to evaluate and model human skills it is necessary to identify a relevant set of attributes by which skill can be measured, interpreted and evaluated by computers. In this paper we present our work on attributed skill transfer and focus on the specific task of writing. Identifying the central attributes associated with writing and analyzing the consistencies of these attributes within and among subjects determines whether one unique expert model can be derived from a pool of experts. In our study key attributes of hand writing were identified and statistical analysis on subject data was conducted. In our analysis we found experts' behavior can be modeled using the parameters and the same could be used to distinguish between experts and novices, lending itself as an evaluation tool.