Hierarchical Octree and Sub-Volume Texture Block Projection for GPU Accelerated Ray Casting Volume Rendering

Volume processing of fluorescent laser confocal microscopic cellular images

“…Watershed segmentation has been refined by combing with other enhancement procedures and used for separating connected nuclei or cells Irinopoulou et al, 1997;Lin et al, 2003;Malpica et al, 1997;Nilsson & Heyden, 2005;Schmitt & Hasse, 2009;Solorzano et al, 1999;Thomas & Warren, 2007;Wählby et al, 2004). Rigaut et al (1991) developed a segmentation method composed of gray level opening, top hat transformation, geodesic reconstruction, particle deagglomeration by the watershed algorithm, and final discrimination by adaptive thresholding to segment 3D tissue samples of normal rat liver and in situ carcinoma of the esophagus.…”

“…(1997) presented a 3D segmentation method that used global thresholding followed by morphological filtering and segregation by watershed algorithm to segment each 2D slice of the 3D image. Marker controlled or seeded watershed transformation has been proposed to define singular seed or marker for each cell or nuclei object using gray scale mathematical morphological operations, which simultaneously utilize both the morphological and intensity information (Lin et al, 2003;Malpica et al, 1997;Nilsson & Heyden, 2005;Schmitt & Hasse, 2009;Solorzano et al, 1999;Wählby et al, 2004) The watershed method, however, often suffers from the over-segmentation problem, and heuristic rules are frequently used for region merging (Adiga & Chaudhuri, 2001b;. Adiga (2001) developed a region-based segmentation method involving a watershed algorithm to separate clustered objects, followed by a rule-based merging technique to reduce the over-segmentation of cells caused by the watershed algorithm discontinuities.…”

“…2001introduced an accelerated volume ray casting using 2D texture to store the ray terminals. The idea of hardware ray casting has recently gained increased attention because it has become feasible on GPUs (Kr¨uger & Westermann, 2003;Scharsach, 2005;Stegmaier et al, 2005;Gobbetti et al, 2008;Ma & Zhang, 2010;Zhanfang et al, 2010).…”

“…Recent generations of consumer GPU architectures are designed for high efficiency and high computational load, along with effective use of coherency and extensive memory bandwidth requirements (Natalya et al, 2008). In recent days, advances and innovations in latest GPU architectures (such as ATI Radeon HD 2000 and beyond, NVIDIA's GeForce graphics card series) has been exploited extensively to perform real-time volume rendering and visualization of high-resolution volumetric datasets (Engel et al, 2001;Levin et al, 2005;Ma et al, 2003, Kruger & Westermann, 2003Scharsach, 2005;Stegmaier et al, 2005;Gobbetti et al, 2008;Ma & Zhang, 2010;Zhanfang et al, 2010). Furthermore, with the advent of programmable GPU pipelines, researchers take advantage of the massive parallelism of GPUs, rather than relying on significantly slower-scaling CPU multi-core architecture (Natalya et al, 2008).…”

“…One of the major complications in the automatic segmentation of cellular images arises due to the fact that the cells are often closely clustered. Cluster division to isolate individual cell objects is a critical task in the automatic evaluation of cytological and histological images in order to study the morphology of cells or nuclei and/or the spatial distribution of cells in the tissue specimen (Pal et al, 1998;Malpica et al, 1997;Nilsson & Heyden, 2005;Schmitt & Hasse, 2009).…”