Christoph Kubisch

Sinus Endoscopy - Application of Advanced GPU Volume Rendering for Virtual Endoscopy

For difficult cases in endoscopic sinus surgery, a careful planning of the intervention is necessary. Due to the reduced field of view during the intervention, the surgeons have less information about the surrounding structures in the working area compared to open surgery. Virtual endoscopy enables the visualization of the operating field and additional information, such as risk structures (e.g., optical nerve and skull base) and target structures to be removed (e.g., mucosal swelling). The Sinus Endoscopy system provides the functional range of a virtual endoscopic system with special focus on a realistic representation. Furthermore, by using direct volume rendering, we avoid time-consuming segmentation steps for the use of individual patient datasets. However, the image quality of the endoscopic view can be adjusted in a way that a standard computer with a modern standard graphics card achieves interactive frame rates with low CPU utilization. Thereby, characteristics of the endoscopic view are systematically used for the optimization of the volume rendering speed. The system design was based on a careful analysis of the endoscopic sinus surgery and the resulting needs for computer support. As a small standalone application it can be instantly used for surgical planning and patient education. First results of a clinical evaluation with ENT surgeons were employed to fine-tune the user interface, in particular to reduce the number of controls by using appropriate default values wherever possible. The system was used for preoperative planning in 102 cases, provides useful information for intervention planning (e.g., anatomic variations of the Rec. Frontalis), and closely resembles the intraoperative situation.

Adapted surface visualization of cerebral aneurysms with embedded blood flow information

Cerebral aneurysms are a vascular dilatation induced by a pathological change of the vessel wall and often require treatment to avoid rupture. Therefore, it is of main interest, to estimate the risk of rupture, to gain a deeper understanding of aneurysm genesis, and to plan an actual intervention, the surface morphology and the internal blood flow characteristics. Visual exploration is primarily used to understand such complex and variable type of data. Since the blood flow data is strongly influenced by the surrounding vessel morphology both have to be visually combined to efficiently support visual exploration. Since the flow is spatially embedded in the surrounding aneurysm surface, occlusion problems have to be tackled. Thereby, a meaningful visual reduction of the aneurysm surface that still provides morphological hints is necessary. We accomplish this by applying an adapted illustrative rendering style to the aneurysm surface. Our contribution lies in the combination and adaption of several rendering styles, which allow us to reduce the problem of occlusion and avoid most of the disadvantages of the traditional semi-transparent surface rendering, like ambiguities in perception of spatial relationships. In interviews with domain experts, we derived visual requirements. Later, we conducted an initial survey with 40 participants (13 medical experts of them), which leads to further improvements of our approach.

GPU-based smart visibility techniques for tumor surgery planning

PurposeThe rating of distances and infiltrations to vital structures is important for the planning of tumor surgery or interventional procedures. To support such an assessment, the target structures should be clearly emphasized in a 3D visualization by ensuring their visibility.MethodsSmart Visibility techniques such as Ghosting Views and Breakaway Views are employed. Ghosting Views highlight focus structures by fading out occluding structures and are often used in anatomical illustrations. Breakaway Views reveal the structure by cutting into surrounding structures. As a result, an intersection surface is created that allows relating the focus structure with its surroundings. In this contribution, a specialized GPU-based implementation of these techniques is presented for polygonal models derived from a segmentation of the anatomical structures.ResultsWe present different rendering styles of the techniques and apply them to highlight enlarged lymph nodes in the neck, as well as tumors inside the liver. Compared to other methods, we focus on polygonal models and optimizations. Thus, very high frame rates could be achieved on consumer graphics hardware. Furthermore, we employed markers that support the estimation of distances within the scene and possible infiltrations around the focus structures.ConclusionThe parameters for the techniques are defined automatically to aid the employment in clinical routine. Such an application is also supported by the combination and refinement of established rendering techniques.

Vessel Visualization with Volume Rendering

Volume rendering allows the direct visualization of scanned volume data, and can reveal vessel abnormalitiesmore faithfully. In this overview, we will present a pipeline model for direct volume rendering systems, which focus on vascular structures. We will cover the fields of data pre-processing, classification of the volume via transfer functions, and finally rendering the volume in 2D and 3D. For each stage in the pipeline, different techniques are discussed to support the diagnosis of vascular diseases. Next to various general methods we will present two case studies, in which the systems are optimized for two different medical issues. At the end, we discuss current trends in volume rendering and their implications for vessel visualization.

Automatic Transfer Function Specification for Visual Emphasis of Coronary Artery Plaque

Cardiovascular imaging with current multislice spiral computed tomography (MSCT) technology enables a non‐invasive evaluation of the coronary arteries. Contrast‐enhanced MSCT angiography with high spatial resolution allows for a segmentation of the coronary artery tree. We present an automatically adapted transfer function (TF) specification to highlight pathologic changes of the vessel wall based on the segmentation result of the coronary artery tree. The TFs are combined with common visualization techniques, such as multiplanar reformation and direct volume rendering for the evaluation of coronary arteries in MSCT image data. The presented TF‐based mapping of CT values in Hounsfield Units (HU) to color and opacity leads to a different color coding for different plaque types. To account for varying HU values of the vessel lumen caused by the contrast medium, the TFs are adapted to each dataset by local histogram analysis. We describe an informal evaluation with three board‐certified radiologists which indicates that the represented visualizations guide the users attention to pathologic changes of the vessel wall as well as provide an overview about spatial variations.

Interactive Mesh Smoothing for Medical Applications

Surface models derived from medical image data often exhibit artefacts, such as noise and staircases, which can be reduced by applying mesh smoothing filters. Usually, an iterative adaption of smoothing parameters to the specific data and continuous re‐evaluation of accuracy and curvature is required. Depending on the number of vertices and the filter algorithm, computation time may vary strongly and interfere with an interactive mesh generation procedure. In this paper, we present an approach to improve the handling of mesh smoothing filters. Based on a GPU mesh smoothing implementation of uniform and anisotropic filters, model quality is evaluated in real‐time and provided to the user to support the mental optimization of input parameters. This is achieved by means of quality graphs and quality bars. Moreover, this framework is used to find appropriate smoothing parameters automatically and to provide data‐specific parameter suggestions. These suggestions are employed to generate a preview gallery with different smoothing suggestions. The preview functionality is additionally used for the inspection of specific artefacts and their possible reduction with different parameter sets.

Visually Guided Mesh Smoothing for Medical Applications

Surface models derived from medical image data often exhibit artifacts, such as noise and staircases, which can be reduced by applying mesh smoothing filters. Usually, an iterative adaption of smoothing parameters to the specific data and continuous re-evaluation of accuracy and curvature is required. Depending on the number of vertices and the filter algorithm, computation time may vary strongly and interfere with an interactive mesh generation procedure. In this paper, we present an approach to improve the handling of mesh smoothing filters. Based on a GPU mesh smoothing implementation, model quality is evaluated in real-time and provided to the user as quality graphs to support the mental optimization of input parameters. Moreover, this framework is used to find optimal smoothing parameters automatically and to provide data-specific parameter suggestions.

GPU-basierte Smart Visibility Techniken für die Planung von Tumor-Operationen

Bei der Planung von Tumoroperationen ist die Einschatzung von Abstanden und Infiltrationen zu vitalen Strukturen wichtig. Im Bereich der medizinischen Visualisierung wurden hierfur bereits zahlreiche Techniken entwickelt, die unter dem Begriff Smart Visibility zusammengefasst werden. Zu diesen zahlen Ghost Views und Section Views. In diesem Beitrag wird eine GPU-basierte Realisierung dieser Techniken fur polygonale Daten vorgestellt. Die Parametrisierung der Techniken erfolgt automatisch, um einen klinischen Einsatz ermoglichen zu konnen.

Real-time surface analysis and tagged material cleansing for virtual colonoscopy

Virtual Colonoscopy is an important procedure for screening and detecting colorectal cancer. It increases patient comfort and reduces risks compared to optical colonoscopy. Oral contrast is used to emphasize the soft tissue border and avoid the need for physical colon cleansing. In order to ensure a reliable diagnosis, it is currently necessary to remove the fecal tagging in a time consuming pre-processing step. As the result can include artifacts and may effect polyp size, this paper proposes a novel technique that allows realistic visualization of the surface boundary based on unmodified CT images. A combined iso-surface reconstruction and direct volume rendering approach is developed to handle partial volume artifacts efficiently and allow on-the-fly surface reconstruction. The algorithm supports real-time analysis of detected surfaces and can differentiate material transitions between air, soft tissue and fluid. The surface-based rendering furthermore allows photo-realistic visualization through screen space shading to support procedure planning and interactive training.

Advanced GPU Volume Rendering for Virtual Endoscopy

For difficult cases in endoscopic sinus surgery, a careful planning of the intervention is necessary. Virtual endoscopy enables the visualization of the operating field and additional information, such as risk structures and target structures to be removed. The Sinus Endoscopy system provides the functional range of a virtual endoscopic system with special focus on a realistic representation. Furthermore, by using direct volume rendering, we avoid time-consuming segmentation steps for the use of individual patient datasets. However, the image quality of the endoscopic view can be adjusted in a way that a standard computer with a modern standard graphics card achieves interactive frame rates with low CPU utilization. Thereby, characteristics of the endoscopic view are systematically used for the optimization of the volume rendering speed. As a small standalone application it can be instantly used for surgical planning and patient education. The system was used for preoperative planning in 102 cases, provides useful information for intervention planning (e.g., anatomic variations of the Rec. Frontalis), and closely resembles the intraoperative situation.