Pathologic review of tumor morphology in histologic sections is the traditional method for cancer classification and grading, yet human has limitations that can result low reproducibility inter-observer agreement. Computerized image analysis partially overcome these shortcomings due to its capacity quantitatively reproducibly measure structures on a large-scale. In this paper, we present an end-to-end data integration pipeline large-scale morphologic pathology images demonstrate ability...

This paper proposes and evaluates CUDAlign 4.0, a parallel strategy to obtain the optimal alignment of huge DNA sequences in multi-GPU platforms, using exact Smith–Waterman (SW) algorithm. In first phase Dynamic Programming (DP) matrix is computed by multiple GPUs, which asynchronously communicate border elements right neighbor order find score. After that, traceback SW executed. The efficient parallelization very challenging because high amount data dependency, particularly impacts...

GPUs have recently evolved into very fast parallel co-processors capable of executing general purpose computations extremely efficiently. At the same time, multi-core CPUs evolution continued and today's 4-8 cores. These two trends, however, followed independent paths in sense that we are aware few works consider both devices cooperating to solve computations. In this paper investigate coordinated use CPU GPU improve efficiency applications even further than using either device...

We study and characterize the performance of operations in an important class applications on GPUs Many Integrated Core (MIC) architectures. Our work is motivated by that analyze low-dimensional spatial datasets captured high resolution sensors, such as image obtained from whole slide tissue specimens using microscopy scanners. Common these involve detection extraction objects (object segmentation), computation features each extracted object (feature computation), characterization based...

Abstract Motivation Predicting pathological complete response (pCR) to neoadjuvant chemotherapy (NAC) in triple-negative breast cancer (TNBC) patients accurately is direly needed for clinical decision making. pCR also regarded as a strong predictor of overall survival. In this work, we propose deep learning system predict NAC based on serial pathology images stained with hematoxylin and eosin two immunohistochemical biomarkers (Ki67 PHH3). To support human prior domain knowledge-based...

Analysis of large pathology image datasets offers significant opportunities for the investigation disease morphology, but resource requirements analysis pipelines limit scale such studies. Motivated by a brain cancer study, we propose and evaluate parallel application pipeline high throughput computation resolution tissue images on distributed CPU-GPU platforms. To achieve efficient execution these hybrid systems, have built runtime support that allows us to express as hierarchical data...

The past decade has witnessed a major paradigm shift in high performance computing with the introduction of accelerators as general purpose processors. These devices make available very parallel power at low cost and consumption, transforming current platforms into heterogeneous CPU-GPU equipped systems. Although theoretical achieved by these hybrid systems is impressive, taking practical advantage this remains challenging problem. Most applications are still deployed to either GPU or CPU,...

For routine pathology diagnosis and imaging-based biomedical research, Whole-slide image (WSI) analyses have been largely limited to a 2D tissue space. more definitive representation support fine-resolution spatial integrative analyses, it is critical extend such tissue-based investigations 3D space with spatially aligned serial WSIs in different stains, as Hematoxylin Eosin (H&E) Immunohistochemistry (IHC) biomarkers. However, WSI registration technically challenged by the overwhelming...

The increases in multi-core processor parallelism and the flexibility of many-core accelerator processors, such as GPUs, have turned traditional SMP systems into hierarchical, heterogeneous computing environments. Fully exploiting these improvements parallel system design remains an open problem. Moreover, most current tools for development applications hierarchical concentrate on use only a single type (e.g., accelerators) do not coordinate several processors. Here, we show that making all...

A large number of cell-oriented cancer investigations require an effective and reliable cell segmentation method on three dimensional (3D) fluorescence microscopic images for quantitative analysis biological properties. In this paper, we present a fully automated that can detect cells from 3D images. Enlightened by imaging techniques, regulated the image gradient field vector flow (GVF) with interpolated smoothed data volume, grouped voxels based modes identified tracking GVF field. Adaptive...

Biological sequence alignment is a very popular application in Bioinformatics, used routinely worldwide. Many implementations of biological algorithms have been proposed for multicores, GPUs, FPGAs and CellBEs. These are platform-specific; porting them to other systems requires considerable programming effort. This article proposes evaluates MASA, flexible customizable software architecture that enables the execution applications with three variants (local, global, semiglobal) multiple...

Sensitivity analysis and parameter tuning are important processes in large-scale image analysis. They very costly because the workflows required to be executed several times systematically correlate output variations with changes or tune parameters. An integrated solution minimum user interaction that uses effective methodologies high performance computing is scale these studies large imaging datasets expensive workflows.The experiments two segmentation show proposed approach can (i) quickly...

Oncotype DX Recurrence Score (RS) has been widely used to predict chemotherapy benefits in patients with estrogen receptor-positive breast cancer. Studies showed that the features Magee equations correlate RS. We aimed examine whether deep learning (DL)-based histology image analyses can enhance such correlations.We retrieved 382 cases RS diagnosed between 2011 and 2015 from Emory University Ohio State University. All received surgery. DL models were developed detect nuclei of tumor cells...

We describe a suite of tools and methods that form core set capabilities for researchers clinical investigators to evaluate multiple analytical pipelines quantify sensitivity variability the results while conducting large-scale studies in investigative pathology oncology. The overarching objective current investigation is address challenges large data sizes high computational demands. proposed take advantage state-of-the-art parallel machines efficient content-based image searching...

We carry out a comparative performance study of multi-core CPUs, GPUs and Intel Xeon Phi (Many Integrated Core (MIC)) with microscopy image analysis application. experimentally evaluate the computing devices on core operations correlate observed characteristics data access patterns, computation complexities, parallelization forms operations. The results show significant variability in respect to device used. performances regular are comparable or sometimes better MIC than that GPU. more...

High-throughput serial histology imaging provides a new avenue for the routine study of micro-anatomical structures in 3D space. However, emergence whole slide poses registration challenge, as gigapixel image size precludes direct application conventional techniques. In this paper, we develop three-stage with multi-resolution mapping and propagation method to dynamically produce registered subvolumes from images. We validate our algorithm images brain tumor sections synthetic volumes. The...

Liver steatosis is known as the abnormal accumulation of lipids within cells. An accurate quantification area liver histopathological microscopy images plays an important role in disease diagnosis and transplantation assessment. Such a analysis often requires precise segmentation that challenging due to abundant presence highly overlapped droplets. In this paper, deep learning model Mask-RCNN used segment droplets clumps. Extended from Faster R-CNN, can predict object masks addition bounding...