Aortic dissections are a malformation of the inner wall of the human aorta. As this disease poses a significant risk to the patients life, it is important to analyze the progression of the dissection over the course of the disease. This leads to specialized treatment plans, since surgical intervention is not always the most optimal choice as it implies a certain risk to the patient’s health. We introduce four novel plots of aortic dissections showing the major diameter, the outflow blood volume of the aortic branches, the interventions happened so far, and the adverse-free event probabilities for the next, the second and the fifth year. By introducing aortic dissection maps as the composite visualization of these plots, we form a baseline comparison for assessing risk factors of aortic dissections. Following the vision of precision medicine, these maps not only allow radiologists to analyze the trend of aortic dissections over time to conclude on an individual, but suitable, treatment, but also to compare specific features across patients to analyze their expressive power. Finally, to strengthen our proposed approach, we conducted informal feedback from expert radiologists.

Time-dependent volumetric data has important applications in areas as diverse as medicine, climatology, and engineering. However, the simultaneous quantitative assessment of spatial and temporal features is very challenging. Common visualization techniques show either the whole volume in one time step (for example using direct volume rendering) or let the user select a region of interest (ROI) for which a collection of time intensity curves is shown. In this paper, we propose a novel approach that dynamically embeds quantitative detail views in a spatial layout. Inspired by the concept of small multiples, we generate view-dependent time-intensity graphs on-the-fly by aggregating per-ray information over time and image regions. Our method enables the detailed feature-aligned visual analysis of time-dependent volume data and allows interactive refinement and filtering. Temporal behaviors like frequency relations, aperiodic or periodic oscillations and their spatial context are easily perceived with our method. We demonstrate the power of our approach using examples from medicine and the natural sciences.

Multiple Sequence Alignments (MSA) of a set of DNA, RNA or protein sequences form the fundamental basis for various biological applications such as evolutionary heritage and protein structure prediction. The quality of an MSA is crucial in order to provide useful and correct results. The analysis of MSAs is, however, still a challenging task since constructing MSAs is an NP hard problem and thus the optimal MSA is usually unknown. Additionally, MSA quality analysis is often completely ignored, especially by non expert users, due to the lack of tools for the visual comparison and the intuitive quality assessment of alignments.In this paper, we present an interactive visual approach to simultaneously assess the quality of multiple alternative MSAs of the same sequence set that also enables non experts to perform MSA analysis in an intuitive way. We provide a direct assessment of the alignment quality using different highlighting techniques in combination with automatic quality measures and editing capabilities. An in-depth evaluation of our approach highlights its benefits for MSA quality assessment.

Visualization of structural biology data uses color to categorize or separate dense structures into particular semantic units. In multiscale models of viruses or bacteria, there are atoms on the finest level of detail, then amino-acids, secondary structures, macromolecules, up to the compartment level and, in all these levels, elements can be visually distinguished by color.However, currently only single scale coloring schemes are utilized that show information for one particular scale only. We present a novel technology which adaptively, based on the current scale level, adjusts the color scheme to depict or distinguish the currently best visible structural information. We treat the color as a visual resource that is distributed given a particular demand. The changes of the color scheme are seamlessly interpolated between the color scheme from the previous views into a given new one. With such dynamic multi-scale color mapping we ensure that the viewer is able to distinguish structural detail that is shown to her on any given scale. This technique has been tested by users with an expertise in structural biology and has been overall well received.

Magnetic Resonance Imaging (MRI) enables volumetric and time-varying measurements of blood-flow data. Such data have shown potential to improve diagnosis and risk assessment of various cardiovascular diseases. Hereby, a unique way of analysing patient-specific haemodynamics becomes possible. However, these measurements are susceptible to artifacts, noise and a coarse spatio-temporal resolution. Furthermore, typical flow visualization techniques rely on interpolation. For example, using pathlines requires a high quality temporal resolution. While numerical simulations, based on mathematical flow models, address some of these limitations, the involved modelling assumptions (e.g., regarding the inflow and mesh) do not lead to patient-specific data — as actual measurements would. To overcome this issue, data assimilation techniques can be applied to use measured data in order to steer a physically-based simulation of the flow, combining the benefits of measured data and simulation. Our work builds upon such an existing solution to increase the temporal resolution of the measured data, but achieves significantly higher fidelity. We avoid the previous damping and interpolation bias towards one of the measurements, by simulating bidrectionally (forwards and backwards through time) and using sources and sinks. Our method is evaluated and compared to the, currently-used, conventional interpolation scheme and forward-only simulation using measured and analytical flow data. It reduces artifacts, noise, and interpolation error, while being closer to laminar flow, as is expected for flow in vessels.

We present a visualization application supporting interventional radiologists during analysis of simulated minimally invasive cancer treatment. The current clinical practice employs only rudimentary, manual measurement tools. Our system provides visual support throughout three evaluation stages, starting with determining prospective treatment success of the simulation parameterization. In case of insufficiencies, stage two includes a simulation scalar field for determining a new configuration of the simulation. For complex cases where stage two does not lead to a decisive strategy, stage three reinforces analysis of interdependencies of scalar fields via bivariate visualization. Our system is designed to be immediate applicable in medical practice. We analyze the design space of potentially useful visualization techniques and appraise their effectiveness in the context of our design goals. Furthermore, we present a user study, which reveals the disadvantages of manual analysis in the measurement stage of evaluation and highlight the demand for computer-support through our system.

We present a semi-immersive 3D User Interface to sketch complex vascular structures and vessel pathologies by drawing centerlines in 3D. Our framework comprises on-the-fly reconstruction of the corresponding vessel surface and subsequent local surface compression and expansion. Additionally, we allow the enrichment with an illustrative, plausible blood flow visualization. Our framework is designed for medical educators and students to support anatomy and pathology education. Anatomy educators can realize the step-by-step process of creating and explaining complex spatial relationships of interlinked vascular structures and blood flow behavior. Students can view this process and explore the created structures, which helps them in reproducing and memorizing them. To create a surface model based on the sketched centerlines, we employ implicit surfaces. This allows for easy adding, editing, and removing vessel branches and achieve continuous surfaces with smooth transitions at branchings. The blood flow can be interactively added and is realized with a topology-aware particle simulation. We qualitatively evaluated our framework and demonstrate the applicability and usability of our approach. Our framework is designed for medical educators and students to support anatomy and pathology education. Anatomy educators can realize the step-by-step process of creating and explaining complex spatial relationships of interlinked vascular structures and blood flow behavior. Students can view this process and explore the created structures, which helps them in reproducing and memorizing them. To create a surface model based on the sketched centerlines, we employ implicit surfaces. This allows for easy adding, editing, and removing vessel branches and maintain continuous surfaces with smooth transitions at branchings. The blood flow can be interactively added and is realized with a topology-aware particle simulation. We qualitatively evaluated our framework and discuss its possible integration in the anatomy curriculum.

For many clinical research applications as well as in clinical practice, vessel surface meshes are employed for visual exploration and evaluation of the patient-specific anatomy and pathologies. Most often, the segmentation itself is part of the application. Hence, a visual feedback how tuning parameters influences the segmentation result is desired. We adapted three comparative visualization techniques for the evaluation of cerebral aneurysm surface mesh volumes. Furthermore, we present a one factor, within-subject user study, which allows for comparison of these visualization techniques as well as identification of the most suitable one. The evaluation includes a qualitative as well as a comprehensive quantitative statistical analysis for comparison of significant differences between the techniques. As a result, we determine a color-coded map surface view as best suitable to depict the aneurysm volume changes. Furthermore, the presentation of the different stages of the evaluation pipeline allows for an easy adaption to other application areas of medical visualization. As a result, we hope that more researchers enrich their qualitative evaluations by the presented quantitative statistical analyses evaluations in the domain of medical visualization.

Cardiac flow is still not fully understood, and is currently an active research topic. Using phase-contrast magnetic resonance imaging (PC-MRI) blood flow can be measured. For the inspection of such flow, researchers often rely on methods that require additional scans produced by different imaging modalities to provide context. This requires labor-intensive registration and often manual segmentation before any exploration of the data is performed. This work provides a framework that allows for a quick exploration of cardiac flow without the need of additional imaging and time-consuming segmentation. To achieve this, only the 4D data from one PC-MRI scan is used. A context visualization is derived automatically from the data, and provides context for the flow. Instead of relying on segmentation to deliver an accurate context, the heart’s ventricles are approximated by half-ellipsoids that can be placed with minimal user interaction. Furthermore, seeding positions for flow visualization can be placed automatically in areas of interest defined by the user and based on derived flow features. The framework enables a user to do a fast initial exploration of cardiac flow, as is demonstrated by a use case and a user study involving cardiac blood flow researchers.

Blood flow data from direct measurements (4D-MRI) or numerical simulations opens new possibilities for the understanding of the development of cardiac diseases. However, before this new data can be used in clinical studies or for diagnosis, it is important to develop a notion of the characteristics of typical flow structures. To support this process we developed a novel blood flow clustering and exploration method. The method builds on the concept of coherent flow structures. Coherence maps for cross-sectional slices are defined to show the overall degree of coherence of the flow. In coherent regions the method summarizes the dominant blood flow using a small number of pathline representatives. In contrast to other clustering approaches the clustering is restricted to coherent regions and pathlines with low coherence values, which are not suitable for clustering and thus are not forced into clusters. The coherence map is based on the Finite-time Lyapunov Exponent (FTLE). It is created on selected planes in the inflow respective outflow area of a region of interest. The FTLE value measures the rate of separation of pathlines originating from this plane. Different to previous work using FTLE we do not focus on separating extremal lines but on local minima and regions of low FTLE intensities to extract coherent flow. The coherence map and the extracted clusters serve as basis for the flow exploration. The extracted clusters can be selected and inspected individually. Their flow rate and coherence provide a measure for their significance. Switching off clusters reduces the amount of occlusion and reveals the remaining part of the flow. The non-coherent regions can also be explored by interactive manual pathline seeding in the coherence map.

Cardiovascular disease is the leading cause of death worldwide according to the World Health Organization (WHO). Nearly half of all heart failures occur due to the decline in the performance of the left ventricle (LV). Therefore, early detection, monitoring, and accurate diagnosis of LV pathologies are of critical importance. Usually, global cardiac function parameters are used to assess the cardiac structure and function, although regional abnormalities are important biomarkers of several cardiac diseases. Regional motion of the myocardium, the muscular wall of the LV, can be captured in a non-invasive manner using the velocity-encoded magnetic resonance (MR) imaging method known as Tissue Phase Mapping (TPM). To analyze the complex motion pattern, one typically visualizes for each time step the radial, longitudinal, and circumferential velocities separately according to the American Heart Association (AHA) model, which makes the comprehension of the spatio-temporal pattern an extremely challenging cognitive task. We propose novel spatio-temporal visualization methods for LV myocardial motion analysis with less cognitive load. Our approach uses coordinated views for navigating through the data space. One view visualizes individual time steps, which can be scrolled or animated, while a second view visualizes the temporal evolution using the radial layout of a polar plot for the time dimension. Different designs for visual encoding were considered in both views and evaluated with medical experts to demonstrate and compare their effectiveness and intuitiveness for detecting and analyzing regional abnormalities.

Radiotherapy is one of the most common treatments for prostate tumors. To create a radiotherapy dose plan, accurate segmentation of the prostate and the surrounding organs is required. However, the automatic model-based methods that are often employed, may produce inaccurate segmentations. These, if used as input to the dose plan, can result in inadequate dose administration — either with side effects to healthy tissues, or with under-treatment of the tumor. Currently, an analysis to predict which anatomic regions are prone to inaccuracies, and to determine how to improve segmentation algorithms, cannot be performed. We propose a visual tool to enable experts, working on model-based segmentation algorithms, to explore and analyze the outcomes and errors of their methods. Our approach supports the global exploration of errors in a cohort of pelvic organ segmentations, where the performance of an algorithm can be assessed. Also, it enables the detailed exploration and assessment of segmentation errors, in individual subjects. To the best of our knowledge, there is no other tool with comparable functionality. A usage scenario is employed to explore and illustrate the capabilities of our visual tool. To further assess the value of the proposed tool, we performed an evaluation with five segmentation experts. The evaluation participants confirmed the potential of the tool in providing new insight into their data and employed algorithms. They also gave feedback for future improvements.

A challenging problem in biology is how to address the incompleteness of acquired information when visualizing biological phenomena. Structural biology generates detailed models of viruses or bacteria at different development stages, while the processes that relate one stage to another are often not clear. Similarly, the entire life cycle of a biological entity might be available as a quantitative model, while only one structural model is available. If the relation between two models is specified at a lower level of detail than the actual models themselves, the two models cannot be interpolated correctly. We propose a method that deals with the visualization of incomplete data information in the developmental or evolutionary states of biological mesoscale models, such as viruses or microorganisms. The central tool in our approach is visual abstraction. Instead of directly interpolating between two models that show different states of an organism, we gradually forward transform the models into a level of visual abstraction that matches the level of detail of the modeled relation between them. At this level they can be interpolated without conveying false information. After the interpolation to the new state, we apply the inverse transformation to the model’s original level of abstraction. To show the flexibility of our approach, we demonstrate our method on the basis of molecular data, in particular data of the HIV virion and mycoplasma.

In medical visualization of surface information, problems often arise when visualizing several overlapping structures simultaneously. There is a trade-off between visualizing multiple structures in a detailed way and limiting visual clutter, in order to allow users to focus on the main structures. Illustrative visualization techniques can help alleviate these problems by defining a level of abstraction per structure. However, clinical uptake of these advanced visualization techniques so far has been limited due to the complex parameter settings required. To bring advanced medical visualization closer to clinical application, we propose a novel illustrative technique that offers a seamless transition between various levels of abstraction and detail. Using a single comprehensive parameter, users are able to quickly define a visual representation per structure that fits the visualization requirements for focus and context structures. This technique can be applied to any biomedical context in which multiple surfaces are routinely visualized, such as neurosurgery, radiotherapy planning or drug design. Additionally, we introduce a novel hatching technique, that runs in real-time and does not require texture coordinates. An informal evaluation with experts from different biomedical domains reveals that our technique allows users to design focus-and-context visualizations in a fast and intuitive manner.

The presence of tunnels in protein structures substantially influences their reactivity with other molecules. Therefore, studying their properties and changes over time has been in the scope of biochemists for decades. In this paper we introduce a novel approach for comparative visualization and exploration of ensembles of tunnels. Our goal is to overcome occlusion problems present in traditional tunnel representations while providing users a quick way to navigate through the input dataset to identify potentially interesting tunnels. First, we unfold the input tunnels to a 2D representation enabling to observe the mutual position of amino acids forming the tunnel surface and the amount of surface they influence. These 2D images are subsequently described by image moments commonly used in image processing. This way we are able to detect similarities and outliers in the dataset, which are visualized as clusters in a scatterplot graph. The same coloring scheme is used in the linked bar chart enabling to detect the position of the cluster members over time. These views provide a way to select a subset of potentially interesting tunnels that can be further explored in detail using the 2D unfolded view and also traditional 3D representation. The usability of our approach is demonstrated on case studies conducted by the domain experts.

Ray-based simulations have been shown to generate impressively realistic ultrasound images in interactive frame rates. Recent efforts used GPU-based surface ray tracing to simulate complex ultrasound interactions such as multiple reflections and refractions. These methods are restricted to perfectly specular reflections (i.e., following only a single reflective/refractive ray), whereas real tissue exhibits roughness of varying degree at tissue interfaces, causing partly diffuse reflections and refractions. Such surface interactions are significantly more complex and can in general not be handled by such deterministic ray tracing approaches. However, they can be efficiently computed by Monte-Carlo sampling techniques, where many ray paths are generated with respect to a probability distribution. In this paper we introduce Monte-carlo ray tracing for ultrasound. This enables the realistic simulation of ultrasound interactions such as soft shadows and fuzzy reflections. We discuss how to properly weight the contribution of each ray path in order to simulate the behaviour of a beamformed ultrasound signal. Tracing many rays per transducer element is easily parallelizable on modern GPUs, as opposed to previous US ray tracing approaches based on recursive binary trees.

We present an illustrative rendering pipeline which combines anatomical information from CT-scans with functional information from PET-scans. To treat lung metastases with Selective Internal Radiation Therapies, combined PET/CT-recordings were used for treatment planning and intervention validation. For our result, we firstly extract surface meshes of the anatomical structures including the lung lobes, the trachea and the bronchial arteries from the CT-scan. In addition, the radiation activity of the therapeutic agent Y90 is acquired from the PET data. To convey all this information in one view, we use illustrative rendering techniques, combining Order-Independant Transparencies with Boundary Enhancements and Silhouettes. The capabilities of our results are evaluated by clinical and visualization domain experts. This study indicates an excellent spatial perception and evaluation of tumor position, metabolic and therapeutic agent activity, when transparencies and boundary enhancements are used to render the surrounding lung lobes.

Digital Pathology is a recent clinical environment in which Electronic Health Records (EHRs), biopsy data and whole-slideimages (WSI) come together to provide pathologists the necessary information for making a diagnosis. Integration of this heterogeneous data into a single application is still one of the challenges in the evolution of pathology to a digital practice. While pathologists can perform diagnoses routinely on digital slides only, this is not the case in clinical research. For such purposes, the link between clinicopathological information of patients and images is essential. For example, image analysis researchers who develop automated diagnostic (support) algorithms need to select a representative set of slides to evaluate their methods. To achieve this, they need applications that combine cohort specification, slide image exploration, and selection of suitable images. We present the visualization tool PATHONE, which enables users to perform these steps on a single screen, integrating cohort and WSI selection.

Four-dimensional phase-contrast magnetic resonance imaging (4D PC-MRI) is a method to non-invasively acquire in-vivo blood flow, e.g. in the aorta. It produces three-dimensional, time-resolved datasets containing both flow speed and direction for each voxel. In order to perform qualitative and quantitative data analysis on these datasets, a vessel segmentation is often required. These segmentations are mostly performed manually or semi-automatically, based on three-dimensional intensity images containing the maximal flow speed over all time steps. While working well on datasets with good overall quality, segmenting low-contrast images usually requires time-consuming and exhausting manual input. To allow for a faster segmentation, we propose a method that, in addition to intensity, incorporates the flow trajectories into the segmentation process. This is accomplished by extracting Lagrangian Coherent Structures (LCS) from the flow data, which indicate physical boundaries in a dynamical system. In order to approximate LCS in our discrete images, we employ Finite Time Lyapunov Exponent (FTLE) fields to quantify the rate of separation of neighboring flow trajectories. LCS appear as ridges or valleys in FTLE images, indicating the presence of either a flow structure boundary or physical boundary such as the aortic wall. We will show that the process of segmenting low-contrast 4D PC-MRI datasets can be simplified by using the generated FLTE data in combination with intensity images.

The characterization of cytoplasmic protein aggregates based on time-lapse fluorescence microscopy imaging data is important for research in neuro-degenerative diseases such as Parkinson. As the manual assessment is time-consuming and subject to significant inter-observer variability, incentive for the development of an objective automated system is provided. We propose and evaluate a pipeline consisting of cell-segmentation, tracking and characterization (i.e. classification) of neurological cells. Focus is specifically on the novel and challenging classification problem which is covered by relying on feature extraction followed by a hybrid classification approach incorporating a support vector machine focusing on mainly stationary information and a hidden Markov model to incorporate temporal context. Several image representations are experimentally evaluated to identify cell properties that are important for discrimination. Relying on the proposed approach, classification accuracies up to 80 % are reached. By extensively analyzing the outcomes, we discuss about strengths and weaknesses of our method as a quantitative assessment tool.

In this paper, we investigate the use of Principal Component Analysis (PCA) for image-based volume visualization. Firstly we compute a high-dimensional eigenspace using training images, pre-rendered using a standard ray-caster, from a spherically distributed range of camera positions. Then, our system is able to synthesize arbitrary views of the dataset with minimal computation at run time. We propose a perceptually-adaptive technique to minimize data size and computational complexity whilst preserving perceptual quality of the visualization, in comparison to corresponding ray-cast images. Results indicate that PCA is able to sufficiently learn the full view-independent volumetric model through a finite number of training images and generalize the computed eigenspace to produce high quality images from arbitrary viewpoints, on demand. The approach has potential application in client-server volume visualization or where results of a computationally-complex 3D imaging process need to be interactively visualized on a display device of limited specification.

Perfusion imaging is an important diagnostic tool used mostly in oncology, neurology and cardiology, to assess the perfusion status of the tissue on a capillary level, e.g. assessment of angiogenesis, ischemic regions and inflammation. This contribution is a review of recent advances in dynamic contrast-enhanced magnetic-resonance and ultrasound imaging (DCE-MRI and DCE-US). In these methods, contrast-agent concentration time curves are derived from the acquired image sequences for each tissue region of interest. These tissue curves are then approximated by a pharmacokinetic model parametrized by perfusion parameters, such as blood flow, blood volume, vessel permeability-surface product, and extravascular-extracellular space volume. In DCE-MRI, the pharmacokinetic model is a convolution of the arterial input function (AIF) and the impulse residue function (IRF). Selection of the appropriate IRF model for a given tissue remains an open question. Selection of the appropriate pharmacokinetic model for a given tissue has been supported by simulations and real data. Estimation of an examination-specific AIF has been one of the main challenges in DCE-MRI. Single- and multi-channel blind-deconvolution approaches based on non-parametric and parametric AIF formulation are reviewed. The recent trend is to use more complex IRF models parametrized by more perfusion parameters which leads to the problem of ill-posed approximation task. This problem requires higher signal-to-noise ratio (SNR), higher temporal resolution and the need for additional information. One approach to improve the SNR and temporal resolution is to use compressed sensing acquisition and image reconstruction techniques. Additional information has been recently gained by application of several boluses of contrast agents with a different molecular weight. In DCE-US, the challenge of absolute quantification of perfusion parameters has not been addressed by many research groups. One approach to absolute perfusion-parameter quantification is based on the same convolutional model as in DCE-MRI (and in DCE-CT, PET, SPECT), with simpler IRF models. This unified model can be applied to the burst-replenishment, bolus and bolus&burst acquisition techniques. The AIF can be measured or estimated by single-channel blind deconvolution.

The Image Projection onto Patients (IPoP) system is work in progress intended to assist medical practitioners perform procedures such as biopsies, or provide a novel anatomical education tool, by projecting anatomy and other relevant information from the operating room directly onto a patient’s skin. This approach is not currently used widely in hospitals but has the benefit of providing effective procedure guidance without the practitioner having to look away from the patient. Developmental work towards the alpha-phase of IPoP is presented including tracking methods for tools such as biopsy needles, patient tracking, image registration and problems encountered with the multi-mirror effect.