Robust Tracing and Visualization of Heterogeneous Microvascular Networks

Abstract

Advances in high-throughput imaging allow researchers to collect three-dimensional images of whole organ microvascular networks. These images are extremely large, and contain networks that are highly complex. As a result, these data sets are time consuming to segment and difficult to visualize. In this paper, we present a framework for segmenting and visualizing vascular networks from terabyte-sized three-dimensional images collected using high-throughput microscopy. While these images require terabytes of storage, the volume devoted to the fiber network is of the total volume size. While the networks themselves are sparse, they are tremendously complex, interconnected, and vary widely in diameter. We describe a parallel GPU-based predictor-corrector method for tracing filaments that is robust to noise and sampling errors common in these data sets. We also propose a number of visualization techniques designed to convey the complex statistical descriptions of fibers across large tissue sections - including commonly studied microvascular characteristics, such as orientation and volume.

Three-Dimensional Microscopy by Milling with Ultraviolet Excitation

“…While the varying vessel thickness can introduce gaps in volume visualizations (Fig. 6d), the images are particularly high contrast and simple to segment using existing algorithms 19 . To better demonstrate this advantage, a region of the mouse cerebral cortex that has remarkable microvascular patterns was selected from an India-ink perfused mouse brain.…”

“…Images were collected at an effective resolution of 1.29 μm with a 3.0 μm section size. We applied an automated segmentation algorithm 19 to this data set, generating an explicit graph model with approximately 8,000 edges and 100,000 vertices of the cortical microvascular network (Fig. 7), which was visualized using ParaView (Kitware).…”

“…While the varying vessel thickness can introduce gaps in volume visualizations (Figure 6d), the images are particularly high contrast and simple to segment using existing algorithms [20]. This allowed us to create an explicit graph model with approximately 8,000 edges and 100,000 vertices of a cortical microvascular network (Figure 7), which was visualized using ParaView (Kitware).…”

VC-Net: Deep Volume-Composition Networks for Segmentation and Visualization of Highly Sparse and Noisy Image Data

“…Then the vessel label can be obtained based on an enhanced angiography map. Govyadinov et al [14] described a template-based predictor-corrector method for tracing filaments that is robust in microvascular datasets, and applied a number of glyph-based visualization techniques to represent the aggregated and biologically relevant information of the extracted microvascular network. Then, they developed a bi-modal visualization framework [15], leveraging graph-based and geometry-based techniques to achieve interactive visualization of microvascular networks.…”

“…The 3D structural / contextual information and quantitative metrics are still missing, although maximum intensity projection (MIP) [31], a widely-used approach for qualitatively visualizing and analyzing the 3D vasculature, has been employed to enhance the local vessel signal, allowing for geometric variability and scalability. The labor-intensive, time-consuming, and 3D global / contextual information-missing nature of the procedure makes it very challenging to fully take advantage of the In recent decades, the automatic model-driven vessel extraction and segmentation approaches have been proposed, such as multiscale filtering [12], region growing techniques [27], active contours [30], geometric flow [8], level-set approach [11], nonlinear subtraction (NLS) method [47], template-based predictor-corrector algorithm [14], etc. However, these approaches are easily overwhelmed by tons of low-level handcrafted features and complicated manual parameter adjustment to overcome aforementioned difficulties and subject variations.…”