Machine learning has emerged as a crucial tool for medical image analysis, largely due to recent developments in deep artificial neural networks addressing numerous, diverse clinical problems. As with any conceptual tool, the effective use of machine learning should be predicated on an understanding of its underlying motivations just as much as algorithms or theory — and to do so, we need to explore its philosophical foundations. One of these foundations is the understanding of how values, despite being non-empirical, nevertheless affect scientific research. This article has three goals: to introduce the reader to values in a way that is specific to medical image analysis; to characterise a particular set of technical decisions (what we call the end-to-end vs. separable learning spectrum ) that are fundamental to machine learning for medical image analysis; and to create a simple and structured method to show how these values can be rigorously connected to these technical decisions. This better understanding of how the philosophy of science can clarify fundamental elements of how medical image analysis research is performed and can be improved. • Research values indirectly but fundamentally affect technical decisions in machine learning for medical image analysis. • Machine learning frameworks can be characterised on a spectrum regarding how much they use explicit intermediate representations, i.e. sub-problems. • Structured approaches to values can show what motivates research towards either end of said spectrum.

Training and planning for neurosurgeries necessitate many requirements from junior neurosurgeons, including perceptual capacities. An effective method of deliberate training is to replicate the required procedures using neurosurgical simulation tools and visualizing a three-dimensional (3D) workspace. However, Augmented Reality (AR) neurosurgical simulators become obsolete for a variety of reasons, including users' distance underestimation. Few investigations have been conducted for improving users' depth perception in AR systems with perceptual motion cues through neurosurgical simulation tools for planning aid purposes. In this poster, we are reporting a user study about whether head motion and perceptual motion cues have any an influence on users' depth perception.

We present a prototype of a system in development for pre-operative planning. The proposed NeuroTable uses a combination of traditional rendering and novel visualization techniques rendered to facilitate real-time collaboration between neurosurgeons during intervention planning. A set of multimodal 2D and 3D renderings convey the relation between the region of interest and the surrounding anatomical structures. A haptic device is used for interaction with the NeuroTable to facilitate immersive control over the 3D cursor and navigation modes for the neurosurgeons during their discourse of planning. A pilot experimental study was conducted to assess the performance of users in targeting points within the preoperative 3D scan. Then, two clinicians participated in the evaluation of the table in discussing and planning a case. Results indicate that the NeuroTable facilitated the discourse and we discuss the results of the speed and accuracy for the specification of entry and target points.

The placement of an external ventricular drain is one of the most commonly performed neurosurgical procedures, and consequently, is an essential skill to be mastered by neurosurgical trainees. The optimal placement of the drain involves choosing an appropriate burr hole on the skull and blindly placing a catheter through the burr hole to intersect a lateral ventricle in order to drain cerebrospinal fluid and relieve intracranial pressure. Undesirable trajectories lead to multiple tries in order to hit the ventricle, with potential risk of damaging eloquent brain areas. In this paper, we describe the development of a simulation environment to train residents on the acquisition of these targeting skills before attempting the placement on live patients. The platform is coupled with an augmented reality image-guidance tool, developed in our lab, to help with the visualization of the ventricles in the patient's head. Performance is evaluated using Fitts' methodology (Fitts, 1954), which respects the users ability to trade-off speed and accuracy.

The primary purpose of medical image visualization is to improve patient outcomes by facilitating the inspection, analysis, and interpretation of patient data. This is only possible if the users’ perceptual and cognitive limitations are taken into account during every step of design, implementation, and evaluation of interactive displays. Visualization of medical images, if executed effectively and efficiently, can empower physicians to explore patient data rapidly and accurately with minimal cognitive effort. This article describes a specific case study in biomedical visualization system design and evaluation, which is the visualization of MR angiography images for planning arteriovenous malformation (AVM) interventions. The success of an AVM intervention greatly depends on the surgeon gaining a full understanding of the anatomy of the malformation and its surrounding structures. Accordingly, the purpose of this study was to investigate the usability of visualization modalities involving contour enhancement and stereopsis in the identification and localization of vascular structures using objective user studies. Our preliminary results indicate that contour enhancement, particularly when combined with stereopsis, results in improved performance enhancement of the perception of connectivity and relative depth between different structures.

We present a unique simulator-based methodology for assessing both technical and nontechnical (cognitive) skills for surgical trainees while immersed in a complete medical simulation environment. Further, we have included two crisis scenarios which allow for the evaluation of the effect of cognitive strategy selection on the low-level surgical skills. Training these mixed-mode scenarios can thereby be evaluated on our platform, allowing for improved assessment and a stronger foundation for credentialing, with the potential to reduce the occurrence of adverse events in the operating room. Scientific evaluation and validation of our work is conducted together with 19 junior surgeons in order to achieve the following goals: 1) to provide a qualitative measure of usability, 2) to assess vertebroplasty technical performance of the surgeon, and 3) to explore the relationship between mental workload and surgical performance during crisis. Our results indicate that: 1) the surgeons scored the face validity of our modeled simulation environment very highly ( 4.68 ±0.48, using a 5-point Likert scale), 2) surgeon training enabled completion of tasks more quickly, and 3) the introduction of crisis scenarios negatively affected the surgeons' objective performance. Taken together, our results underscore the need to develop realistic simulation environments that prepare young residents to respond to emergent events in the operating room.

Acquisition and fusion of multiple imaging modalities is becoming an increasingly desired clinical practice. This is particularly the case in radiation therapy, where dosage must be determined using electron density calculations from CT images, which lack the contrast to resolve soft-tissue structures. This often necessitates the fusion of corresponding MRI images in order to plot radiation trajectories safely around critical tissues. Simultaneous visualization of multiple volumes using direct volume rendering (DVR) techniques offers a number of advantages over traditional visualization methods. Specifically, fused visualization enhances the relational aspects of volumes, providing improved context [cite the first one]. However, there are many challenges involved in implementing DVR using fused data sets. The primary challenge is determining how images overlap to provide meaningful information. Additionally, there is increased computational complexity beyond standard DVR techniques, threatening real-time applications of fused DVR. These difficulties are evident to users of such systems as they must manage complex user interfaces and poor application performance. In this work, we introduce a user-centric multi-volume DVR technique which addresses issues of performance and ease of use. Through an intuitive interface, users are able to spatially define regions of interest, determining the relative contributions of each modality in the output rendering.

In minimally invasive image-guided interventions, different imaging modalities, such as magnetic resonance imaging (MRI), computed tomography (CT), and 3-D ultrasound (US), can provide complementary, multispectral image information. Dynamic image registration is a well-established approach that permits real-time diagnostic information to be enhanced by placing lower-quality real-time images within a high quality anatomical context. For the guidance of cardiac interventions, it would be valuable to register dynamic MRI or CT with intra-operative US. However, in practice, either the high computational cost prohibits such real-time visualization, or else the resulting image quality is not satisfactory for accurate interventional guidance. Modern graphics processing units (GPUs) provide the programmability, parallelism and increased computational precision to address this problem. In this paper, we first outline our research on dynamic 3-D cardiac MR and US image acquisition, real-time dual-modality registration and US tracking. Next, we describe our contributions on image processing and optimization techniques for 4-D (3-D + time) cardiac image rendering, and our GPU-accelerated methodologies for multimodality 4-D medical image visualization and optical blending, along with real-time synchronization of dual-modality dynamic cardiac images. Finally, multiple transfer functions, various image composition schemes, and an extended window-level setting and adjustment approach are proposed and applied to facilitate the dynamic volumetric MR and US cardiac data exploration and enhance the feature of interest of US image that is usually restricted to a narrow voxel intensity range.

The authors investigated the kinematics of human visually guided prehension in a situation similar to that typically used in visual servo control of autonomous robots. They found that kinematic parameters of the human grasp, such as transport velocity, path, and grip aperture, were determined by the 3-dimensional geometric structure of the target object, not the 2-dimensional projected image of the object. The results of this study have important implications for the design of reliable and efficient control systems for robots with human-like capabilities.

We use the left and right monocular motion and structure parameters of two stereo image sequences (direction of translation, relative depth, observer rotation and rotational acceleration) to compute the absolute depth, absolute translation and absolute translational acceleration for each pair of left and right images. Individual translation parameters computed at each frame are integrated over time using a Kalman filter to provide more accuracy and a "best" estimate of the absolute translation at each time.