Lukas Marsalek

High-speed volume ray casting with CUDA

Volume ray casting experiences a renewed interest in the last decade. Largely due to the graphics hardware, which enabled real-time implementations competitive in speed with slicing. However these implementations need specialized shader languages and are forced to use graphics APIs. It makes implementation of advanced methods difficult and hinders performance, bending the programming and execution model for something it was not designed to. In late 2006 a new generation of GPUs has been introduced together with CUDA, C-language API.

Real-Time Ray Tracing of Complex Molecular Scenes

Molecular visualization is one of the cornerstones in structural bioinformatics and related fields. Today, rasterization is typically used for the interactive display of molecular scenes, while ray tracing aims at generating high-quality images, taking typically minutes to hours to generate and requiring the usage of an external off-line program. Recently, real-time ray tracing evolved to combine the interactivity of rasterization-based approaches with the superb image quality of ray tracing techniques. We demonstrate how real-time ray tracing integrated into a molecular modelling and visualization tool allows for better understanding of the structural arrangement of biomolecules and natural creation of publication-quality images in real-time. However, unlike most approaches, our technique naturaly integrates into the full-featured molecular modelling and visualization tool BALLView, seamlessly extending a standard workflow with interactive high-quality rendering.

Measuring properties of molecular surfaces using ray casting

Molecular geometric properties, such as volume, exposed surface area, and occurrence of internal cavities, are important inputs to many applications in molecular modeling. In this work we describe a very general and highly efficient approach for the accurate computation of such properties, which is applicable to arbitrary molecular surface models. The technique relies on a high performance ray casting framework that can be easily adapted to the computation of further quantities of interest at interactive speed, even for huge models.

On geometric artifacts in cryo electron tomography

Single-tilt scheme is nowadays the prevalent acquisition geometry in electron tomography and subtomogram averaging experiments. Being an incomplete scheme that induces ill-posedness in the sense of the X-ray or Radon transform inverse problem, it introduces a number of artifacts that directly influence the quality of tomographic reconstructions. Though individually described by different authors before, a systematic study of these acquisition geometry-related artifacts in one place and across representative set of reconstruction methods has not been, to our knowledge, performed before. Moreover, the effects of these artifacts on the reconstructed density are sometimes misinterpreted, attributing them to the wrong cause, especially if their effects accumulate. In this work, we systematically study the major artifacts of single-tilt geometry known as the missing wedge (incomplete projection set problem), the missing information and the specimen-level interior problem (long-object problem). First, we illustratively describe, using a unified terminology, how and why these artifacts arise and when they can be avoided. Next, we describe the effects of these artifacts on the reconstructions across all major classes of reconstruction methods, including newly-appeared methods like the Iterative Nonuniform fast Fourier transform based Reconstruction method (INFR) and the Progressive Stochastic Reconstruction Technique (PSRT). Finally, we draw conclusions and recommendations on numerous points, especially regarding the mutual influence of the geometric artifacts, ability of different reconstruction methods to suppress them as well as implications to the interpretation of both electron tomography and subtomogram averaging experiments.

Progressive Stochastic Reconstruction Technique (PSRT) for cryo electron tomography.

Cryo Electron Tomography (cryoET) plays an essential role in Structural Biology, as it is the only technique that allows to study the structure of large macromolecular complexes in their close to native environment in situ. The reconstruction methods currently in use, such as Weighted Back Projection (WBP) or Simultaneous Iterative Reconstruction Technique (SIRT), deliver noisy and low-contrast reconstructions, which complicates the application of high-resolution protocols, such as Subtomogram Averaging (SA). We propose a Progressive Stochastic Reconstruction Technique (PSRT) - a novel iterative approach to tomographic reconstruction in cryoET based on Monte Carlo random walks guided by Metropolis-Hastings sampling strategy. We design a progressive reconstruction scheme to suit the conditions present in cryoET and apply it successfully to reconstructions of macromolecular complexes from both synthetic and experimental datasets. We show how to integrate PSRT into SA, where it provides an elegant solution to the region-of-interest problem and delivers high-contrast reconstructions that significantly improve template-based localization without any loss of high-resolution structural information. Furthermore, the locality of SA is exploited to design an importance sampling scheme which significantly speeds up the otherwise slow Monte Carlo approach. Finally, we design a new memory efficient solution for the specimen-level interior problem of cryoET, removing all associated artifacts.

Ettention: Building Blocks for Iterative Reconstruction Algorithms

We present a novel software package for tomographic reconstruction in electron microscopy, named Ettention [1]. The software consists of a set of modular building-blocks for iterative reconstruction algorithms. Ettention simultaneously features (1) a modular, object-oriented software design, (2) optimized access to high-performance computing (HPC) platforms such as graphic processing units (GPU) or many-core architectures like Xeon Phi, and (3) accessibility to microscopy end-users via integration in the IMOD package and user interface. We provide developers with a clean application programming interface (API) that allows for extending the software easily and thus makes it an ideal platform for algorithmic research while hiding most of the technical details of high-performance computing. Several case studies are provided to demonstrate the feasibility of the concept [2].

Realistic lighting simulation for interactive VR applications

In the field of aircraft design, interior illumination increasingly becomes an important design element. Different illumination scenarios inside an aircraft cabin are considered to influence the mood of air passengers, help passengers to be better prepared for time lags and to create an overall positive environment. Consequently, a physically correct and realistic lighting simulation becomes essential during the design process. Available tools are producing videos or still images of illumination settings. The main reason for this is that realistic lighting simulation is believed to require heavy offline processing and unfeasible to do from within a real-time system. On the other hand, interactive Virtual Reality (VR) applications are an appropriate tool to experience an aircraft cabin under different illuminations. The ability to integrate lighting simulations into VR applications would simplify the design process remarkably by skipping time-consuming context and tool switches. In this paper, we present a solution for integrating realistic lighting simulation with interactive performance into a single VR application. We explain our integration of real-time ray tracing, interactive global illumination, and measured point lights in a VR system, and its combination with classic rasterization techniques. We describe suitable interaction metaphors to enable realistic lighting simulation, high interactivity and intuitive interaction in an application for light design inside an aircraft cabin.

PSRT: Progressive Stochastic Reconstruction Technique for Cryo Electron Tomography

Cryo Electron Tomography (cryoET) is one of the essential techniques in Structural Biology, as it allows us to study the structure of macromolecular complexes in their native environment in situ. The tomographic reconstruction in cryoET is a particularly challenging task as the input data suffers from very low contrast, high noise, and limited tilt range. Moreover, the scanned specimen is larger than the detector, introducing the interior problem into the reconstruction process, which causes vignetting artifacts on the edges of the reconstructions. To alleviate some of these limitations, high-resolution protocols such as Subtomogram Averaging (SA) are applied to obtain structures of individual macromolecular complexes from a tomogram. Results of these protocols are highly dependent on the quality of the reconstruction. Current state-of-the-art methods such as Weighted Back Projection (WBP) or Simultaneous Algebraic Reconstruction Technique (SART) deliver noisy and low-contrast reconstructions and thus manual intervention is often needed during SA.

Progressive stochastic reconstruction technique for cryo electron tomography

Cryo Electron Tomography (cryoET) plays an essential role in Structural Biology, as it is the only technique that allows to study the structure and intracellular distribution of large macromolecular complexes in their (close to) native environment. A major limitation of cryoET is the highest achievable resolution, currently at around 3 nm, which prevents its application to smaller complexes and in turn to a wider range of important biological questions.