Integrated morphologic analysis for the identification and characterization of disease subtypes

Tissue-Level Integration Overrides Gradations of Differentiating Cell Identity in Beetle Extraembryonic Tissue

During animal embryogenesis, one of the earliest specification events distinguishes extraembryonic (EE) from embryonic tissue fates: the serosa in the case of the insects. While it is well established that the homeodomain transcription factor Zen1 is the critical determinant of the serosa, the subsequent realization of this tissue's identity has not been investigated. Here, we examine serosal differentiation in the beetle Tribolium castaneum based on the quantification of morphological and morphogenetic features, comparing embryos from a Tc-zen1 RNAi dilution series, where complete knockdown results in amnion-only EE tissue identity. We assess features including cell density, tissue boundary morphology, and nuclear size as dynamic readouts for progressive tissue maturation. While some features exhibit an all-or-nothing outcome, other key features show dose-dependent phenotypic responses with trait-specific thresholds. Collectively, these findings provide nuance beyond the known status of Tc-Zen1 as a selector gene for serosal tissue patterning. Overall, our approach illustrates how the analysis of tissue maturation dynamics from live imaging extends but also challenges interpretations based on gene expression data, refining our understanding of tissue identity and when it is achieved.

Visual Intratumor Heterogeneity and Breast Tumor Progression

High intratumoral heterogeneity is thought to be a poor prognostic indicator. However, the source of heterogeneity may also be important, as genomic heterogeneity is not always reflected in histologic or 'visual' heterogeneity. We aimed to develop a predictor of histologic heterogeneity and evaluate its association with outcomes and molecular heterogeneity. We used VGG16 to train an image classifier to identify unique, patient-specific visual features in 1655 breast tumors (5907 core images) from the Carolina Breast Cancer Study (CBCS). Extracted features for images, as well as the epithelial and stromal image components, were hierarchically clustered, and visual heterogeneity was defined as a greater distance between images from the same patient. We assessed the association between visual heterogeneity, clinical features, and DNA-based molecular heterogeneity using generalized linear models, and we used Cox models to estimate the association between visual heterogeneity and tumor recurrence. Basal-like and ER-negative tumors were more likely to have low visual heterogeneity, as were the tumors from younger and Black women. Less heterogeneous tumors had a higher risk of recurrence (hazard ratio = 1.62, 95% confidence interval = 1.22-2.16), and were more likely to come from patients whose tumors were comprised of only one subclone or had a TP53 mutation. Associations were similar regardless of whether the image was based on stroma, epithelium, or both. Histologic heterogeneity adds complementary information to commonly used molecular indicators, with low heterogeneity predicting worse outcomes. Future work integrating multiple sources of heterogeneity may provide a more comprehensive understanding of tumor progression.

Geometrical analysis identified morphological features of hydrogel-induced cancer stem cells in synovial sarcoma model cells

Cancer stem cells (CSCs) has been a key target to cure cancer patients completely. Although many CSC markers have been identified, they are frequently cancer type-specific and those expressions are occasionally variable, which becomes an obstacle to elucidate the characteristics of the CSCs. Here we scrutinized the relationship between stemness elevation and geometrical features of single cells. The PAMPS hydrogel was utilized to create the CSCs from mouse myoblast C2C12 and its synovial sarcoma model cells. qRT-PCR analysis confirmed the significant increase in expression levels of Sox2, Nanog, and Oct3/4 on the PAMPS gel, which was higher in the synovial sarcoma model cells. Of note, the morphological heterogeneity was appeared on the PAMPS gel, mainly including flat spreading, elongated spindle, and small round cells, and the Sox2 expression was highest in the small round cells. To examine the role of morphological differences in the elevation of stemness, over 6,400 cells were segmented along with the Sox2 intensity, and 12 geometrical features were extracted at single cell level. A nonlinear mapping of the geometrical features by using uniform manifold approximation and projection (UMAP) clearly revealed the existence of relationship between morphological differences and the stemness elevation, especially for C2C12 and its synovial sarcoma model on the PAMPS gel in which the small round cells possess relatively high Sox2 expression on the PAMPS gel, which supports the strong relationship between morphological changes and the stemness elevation. Taken together, these geometrical features can be useful for morphological profiling of CSCs to classify and distinguish them for understanding of their role in disease progression and drug discovery.

NuCLS: A scalable crowdsourcing approach and dataset for nucleus classification and segmentation in breast cancer

BACKGROUND

Deep learning enables accurate high-resolution mapping of cells and tissue structures that can serve as the foundation of interpretable machine-learning models for computational pathology. However, generating adequate labels for these structures is a critical barrier, given the time and effort required from pathologists.

RESULTS

This article describes a novel collaborative framework for engaging crowds of medical students and pathologists to produce quality labels for cell nuclei. We used this approach to produce the NuCLS dataset, containing >220,000 annotations of cell nuclei in breast cancers. This builds on prior work labeling tissue regions to produce an integrated tissue region- and cell-level annotation dataset for training that is the largest such resource for multi-scale analysis of breast cancer histology. This article presents data and analysis results for single and multi-rater annotations from both non-experts and pathologists. We present a novel workflow that uses algorithmic suggestions to collect accurate segmentation data without the need for laborious manual tracing of nuclei. Our results indicate that even noisy algorithmic suggestions do not adversely affect pathologist accuracy and can help non-experts improve annotation quality. We also present a new approach for inferring truth from multiple raters and show that non-experts can produce accurate annotations for visually distinctive classes.

CONCLUSIONS

This study is the most extensive systematic exploration of the large-scale use of wisdom-of-the-crowd approaches to generate data for computational pathology applications.

Explainable nucleus classification using Decision Tree Approximation of Learned Embeddings

MOTIVATION

Nucleus detection, segmentation and classification are fundamental to high-resolution mapping of the tumor microenvironment using whole-slide histopathology images. The growing interest in leveraging the power of deep learning to achieve state-of-the-art performance often comes at the cost of explainability, yet there is general consensus that explainability is critical for trustworthiness and widespread clinical adoption. Unfortunately, current explainability paradigms that rely on pixel saliency heatmaps or superpixel importance scores are not well-suited for nucleus classification. Techniques like Grad-CAM or LIME provide explanations that are indirect, qualitative and/or nonintuitive to pathologists.

RESULTS

In this article, we present techniques to enable scalable nuclear detection, segmentation and explainable classification. First, we show how modifications to the widely used Mask R-CNN architecture, including decoupling the detection and classification tasks, improves accuracy and enables learning from hybrid annotation datasets like NuCLS, which contain mixtures of bounding boxes and segmentation boundaries. Second, we introduce an explainability method called Decision Tree Approximation of Learned Embeddings (DTALE), which provides explanations for classification model behavior globally, as well as for individual nuclear predictions. DTALE explanations are simple, quantitative, and can flexibly use any measurable morphological features that make sense to practicing pathologists, without sacrificing model accuracy. Together, these techniques present a step toward realizing the promise of computational pathology in computer-aided diagnosis and discovery of morphologic biomarkers.

AVAILABILITY AND IMPLEMENTATION

Relevant code can be found at github.com/CancerDataScience/NuCLS.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

The seen and the unseen: Molecular classification and image based-analysis of gastrointestinal cancers

Gastrointestinal cancers account for 22.5% of cancer related deaths worldwide and represent circa 20% of all cancers. In the last decades, we have witnessed a shift from histology-based to molecular-based classifications using genomic, epigenomic, and transcriptomic data. The molecular based classification revealed new prognostic markers and may aid the therapy selection. Because of the high-costs to perform a molecular classification, in recent years immunohistochemistry-based surrogate classification were developed which permit the stratification of patients, and in parallel multiple groups developed hematoxylin and eosin whole slide image analysis for sub-classifying these entities. Hence, we are witnessing a return to an image-based classification with the purpose to infer hidden information from routine histology images that would permit to detect the patients that respond to specific therapies and would be able to predict their outcome. In this review paper, we will discuss the current histological, molecular, and immunohistochemical classifications of the most common gastrointestinal cancers, gastric adenocarcinoma, and colorectal adenocarcinoma, and will present key aspects for developing a new artificial intelligence aided image-based classification of these malignancies.

Building robust pathology image analyses with uncertainty quantification

BACKGROUND AND OBJECTIVE

Computerized pathology image analysis is an important tool in research and clinical settings, which enables quantitative tissue characterization and can assist a pathologist's evaluation. The aim of our study is to systematically quantify and minimize uncertainty in output of computer based pathology image analysis.

METHODS

Uncertainty quantification (UQ) and sensitivity analysis (SA) methods, such as Variance-Based Decomposition (VBD) and Morris One-At-a-Time (MOAT), are employed to track and quantify uncertainty in a real-world application with large Whole Slide Imaging datasets - 943 Breast Invasive Carcinoma (BRCA) and 381 Lung Squamous Cell Carcinoma (LUSC) patients. Because these studies are compute intensive, high-performance computing systems and efficient UQ/SA methods were combined to provide efficient execution. UQ/SA has been able to highlight parameters of the application that impact the results, as well as nuclear features that carry most of the uncertainty. Using this information, we built a method for selecting stable features that minimize application output uncertainty.

RESULTS

The results show that input parameter variations significantly impact all stages (segmentation, feature computation, and survival analysis) of the use case application. We then identified and classified features according to their robustness to parameter variation, and using the proposed features selection strategy, for instance, patient grouping stability in survival analysis has been improved from in 17% and 34% for BRCA and LUSC, respectively.

CONCLUSIONS

This strategy created more robust analyses, demonstrating that SA and UQ are important methods that may increase confidence digital pathology.

PRISM: A Platform for Imaging in Precision Medicine

PURPOSE

Precision medicine requires an understanding of individual variability, which can only be acquired from large data collections such as those supported by the Cancer Imaging Archive (TCIA). We have undertaken a program to extend the types of data TCIA can support. This, in turn, will enable TCIA to play a key role in precision medicine research by collecting and disseminating high-quality, state-of-the-art, quantitative imaging data that meet the evolving needs of the cancer research community.

METHODS

A modular technology platform is presented that would allow existing data resources, such as TCIA, to evolve into a comprehensive data resource that meets the needs of users engaged in translational research for imaging-based precision medicine. This Platform for Imaging in Precision Medicine (PRISM) helps streamline the deployment and improve TCIA's efficiency and sustainability. More importantly, its inherent modular architecture facilitates a piecemeal adoption by other data repositories.

RESULTS

PRISM includes services for managing radiology and pathology images and features and associated clinical data. A semantic layer is being built to help users explore diverse collections and pool data sets to create specialized cohorts. PRISM includes tools for image curation and de-identification. It includes image visualization and feature exploration tools. The entire platform is distributed as a series of containerized microservices with representational state transfer interfaces.

CONCLUSION

PRISM is helping modernize, scale, and sustain the technology stack that powers TCIA. Repositories can take advantage of individual PRISM services such as de-identification and quality control. PRISM is helping scale image informatics for cancer research at a time when the size, complexity, and demands to integrate image data with other precision medicine data-intensive commons are mounting.

Correlation of manual semi-quantitative and automated quantitative Ki-67 proliferative index with OncotypeDXTM recurrence score in invasive breast carcinoma

BACKGROUND

Ki-67 immunohistochemistry (IHC) staining is a widely used cancer proliferation assay; however, its limitations could be improved with automated scoring. The OncotypeDXTM Recurrence Score (ORS), which primarily evaluates cancer proliferation genes, is a prognostic indicator for breast cancer chemotherapy response; however, it is more expensive and slower than Ki-67.

OBJECTIVE

To compare manual Ki-67 (mKi-67) with automated Ki-67 (aKi-67) algorithm results based on manually selected Ki-67 "hot spots" in breast cancer, and correlate both with ORS.

METHODS

105 invasive breast carcinoma cases from 100 patients at our institution (2011-2013) with available ORS were evaluated. Concordance was assessed via Cohen's Kappa (κ).

RESULTS

57/105 cases showed agreement between mKi-67 and aKi-67 (κ 0.31, 95% CI 0.18-0.45), with 41 cases overestimated by aKi-67. Concordance was higher when estimated on the same image (κ 0.53, 95% CI 0.37-0.69). Concordance between mKi-67 score and ORS was fair (κ 0.27, 95% CI 0.11-0.42), and concordance between aKi-67 and ORS was poor (κ 0.10, 95% CI -0.03-0.23).

CONCLUSIONS

These results highlight the limits of Ki-67 algorithms that use manual "hot spot" selection. Due to suboptimal concordance, Ki-67 is likely most useful as a complement to, rather than a surrogate for ORS, regardless of scoring method.

Histopathological image and gene expression pattern analysis for predicting molecular features and prognosis of head and neck squamous cell carcinoma

BACKGROUND

Histopathological image features offer a quantitative measurement of cellular morphology, and probably help for better diagnosis and prognosis in head and neck squamous cell carcinoma (HNSCC).

METHODS

We first used histopathological image features and machine-learning algorithms to predict molecular features of 212 HNSCC patients from The Cancer Genome Atlas (TCGA). Next, we divided TCGA-HNSCC cohort into training set (n = 149) and test set (n = 63), and obtained tissue microarrays as an external validation set (n = 126). We identified the gene expression profile correlated to image features by bioinformatics analysis.

RESULTS

Histopathological image features combined with random forest may predict five somatic mutations, transcriptional subtypes, and methylation subtypes, with area under curve (AUC) ranging from 0.828 to 0.968. The prediction model based on image features could predict overall survival, with 5-year AUC of 0.831, 0.782, and 0.751 in training, test, and validation sets. We next established an integrative prognostic model of image features and gene expressions, which obtained better performance in training set (5-year AUC = 0.860) and test set (5-year AUC = 0.826). According to histopathological transcriptomics risk score (HTRS) generated by the model, high-risk and low-risk patients had different survival in training set (HR = 4.09, p < 0.001) and test set (HR=3.08, p = 0.019). Multivariate analysis suggested that HTRS was an independent predictor in training set (HR = 5.17, p < 0.001). The nomogram combining HTRS and clinical factors had higher net benefit than conventional clinical evaluation.

CONCLUSIONS

Histopathological image features provided a promising approach to predict mutations, molecular subtypes, and prognosis of HNSCC. The integration of image features and gene expression data had potential for improving prognosis prediction in HNSCC.

Deep learning-based image analysis methods for brightfield-acquired multiplex immunohistochemistry images

BACKGROUND

Multiplex immunohistochemistry (mIHC) permits the labeling of six or more distinct cell types within a single histologic tissue section. The classification of each cell type requires detection of the unique colored chromogens localized to cells expressing biomarkers of interest. The most comprehensive and reproducible method to evaluate such slides is to employ digital pathology and image analysis pipelines to whole-slide images (WSIs). Our suite of deep learning tools quantitatively evaluates the expression of six biomarkers in mIHC WSIs. These methods address the current lack of readily available methods to evaluate more than four biomarkers and circumvent the need for specialized instrumentation to spectrally separate different colors. The use case application for our methods is a study that investigates tumor immune interactions in pancreatic ductal adenocarcinoma (PDAC) with a customized mIHC panel.

METHODS

Six different colored chromogens were utilized to label T-cells (CD3, CD4, CD8), B-cells (CD20), macrophages (CD16), and tumor cells (K17) in formalin-fixed paraffin-embedded (FFPE) PDAC tissue sections. We leveraged pathologist annotations to develop complementary deep learning-based methods: (1) ColorAE is a deep autoencoder which segments stained objects based on color; (2) U-Net is a convolutional neural network (CNN) trained to segment cells based on color, texture and shape; and ensemble methods that employ both ColorAE and U-Net, collectively referred to as (3) ColorAE:U-Net. We assessed the performance of our methods using: structural similarity and DICE score to evaluate segmentation results of ColorAE against traditional color deconvolution; F1 score, sensitivity, positive predictive value, and DICE score to evaluate the predictions from ColorAE, U-Net, and ColorAE:U-Net ensemble methods against pathologist-generated ground truth. We then used prediction results for spatial analysis (nearest neighbor).

RESULTS

We observed that (1) the performance of ColorAE is comparable to traditional color deconvolution for single-stain IHC images (note: traditional color deconvolution cannot be used for mIHC); (2) ColorAE and U-Net are complementary methods that detect 6 different classes of cells with comparable performance; (3) combinations of ColorAE and U-Net into ensemble methods outperform using either ColorAE and U-Net alone; and (4) ColorAE:U-Net ensemble methods can be employed for detailed analysis of the tumor microenvironment (TME). We developed a suite of scalable deep learning methods to analyze 6 distinctly labeled cell populations in mIHC WSIs. We evaluated our methods and found that they reliably detected and classified cells in the PDAC tumor microenvironment. We also present a use case, wherein we apply the ColorAE:U-Net ensemble method across 3 mIHC WSIs and use the predictions to quantify all stained cell populations and perform nearest neighbor spatial analysis. Thus, we provide proof of concept that these methods can be employed to quantitatively describe the spatial distribution immune cells within the tumor microenvironment. These complementary deep learning methods are readily deployable for use in clinical research studies.

Dataset of segmented nuclei in hematoxylin and eosin stained histopathology images of ten cancer types

The distribution and appearance of nuclei are essential markers for the diagnosis and study of cancer. Despite the importance of nuclear morphology, there is a lack of large scale, accurate, publicly accessible nucleus segmentation data. To address this, we developed an analysis pipeline that segments nuclei in whole slide tissue images from multiple cancer types with a quality control process. We have generated nucleus segmentation results in 5,060 Whole Slide Tissue images from 10 cancer types in The Cancer Genome Atlas. One key component of our work is that we carried out a multi-level quality control process (WSI-level and image patch-level), to evaluate the quality of our segmentation results. The image patch-level quality control used manual segmentation ground truth data from 1,356 sampled image patches. The datasets we publish in this work consist of roughly 5 billion quality controlled nuclei from more than 5,060 TCGA WSIs from 10 different TCGA cancer types and 1,356 manually segmented TCGA image patches from the same 10 cancer types plus additional 4 cancer types.

Histo-genomics: digital pathology at the forefront of precision medicine

The toughest challenge OMICs face is that they provide extremely high molecular resolution but poor spatial information. Understanding the cellular/histological context of the overwhelming genetic data is critical for a full understanding of the clinical behavior of a malignant tumor. Digital pathology can add an extra layer of information to help visualize in a spatial and microenvironmental context the molecular information of cancer. Thus, histo-genomics provide a unique chance for data integration. In the era of a precision medicine, a four-dimensional (4D) (temporal/spatial) analysis of cancer aided by digital pathology can be a critical step to understand the evolution/progression of different cancers and consequently tailor individual treatment plans. For instance, the integration of molecular biomarkers expression into a three-dimensional (3D) image of a digitally scanned tumor can offer a better understanding of its subtype, behavior, host immune response and prognosis. Using advanced digital image analysis, a larger spectrum of parameters can be analyzed as potential predictors of clinical behavior. Correlation between morphological features and host immune response can be also performed with therapeutic implications. Radio-histomics, or the interface of radiological images and histology is another emerging exciting field which encompasses the integration of radiological imaging with digital pathological images, genomics, and clinical data to portray a more holistic approach to understating and treating disease. These advances in digital slide scanning are not without technical challenges, which will be addressed carefully in this review with quick peek at its future.

Open access image repositories: high-quality data to enable machine learning research

Originally motivated by the need for research reproducibility and data reuse, large-scale, open access information repositories have become key resources for training and testing of advanced machine learning applications in biomedical and clinical research. To be of value, such repositories must provide large, high-quality data sets, where quality is defined as minimising variance due to data collection protocols and data misrepresentations. Curation is the key to quality. We have constructed a large public access image repository, The Cancer Imaging Archive, dedicated to the promotion of open science to advance the global effort to diagnose and treat cancer. Drawing on this experience and our experience in applying machine learning techniques to the analysis of radiology and pathology image data, we will review the requirements placed on such information repositories by state-of-the-art machine learning applications and how these requirements can be met.

CAESNet: Convolutional AutoEncoder based Semi-supervised Network for improving multiclass classification of endomicroscopic images

OBJECTIVE

This article presents a novel method of semisupervised learning using convolutional autoencoders for optical endomicroscopic images. Optical endomicroscopy (OE) is a newly emerged biomedical imaging modality that can support real-time clinical decisions for the grade of dysplasia. To enable real-time decision making, computer-aided diagnosis (CAD) is essential for its high speed and objectivity. However, traditional supervised CAD requires a large amount of training data. Compared with the limited number of labeled images, we can collect a larger number of unlabeled images. To utilize these unlabeled images, we have developed a Convolutional AutoEncoder based Semi-supervised Network (CAESNet) for improving the classification performance.

MATERIALS AND METHODS

We applied our method to an OE dataset collected from patients undergoing endoscope-based confocal laser endomicroscopy procedures for Barrett's esophagus at Emory Hospital, which consists of 429 labeled images and 2826 unlabeled images. Our CAESNet consists of an encoder with 5 convolutional layers, a decoder with 5 transposed convolutional layers, and a classification network with 2 fully connected layers and a softmax layer. In the unsupervised stage, we first update the encoder and decoder with both labeled and unlabeled images to learn an efficient feature representation. In the supervised stage, we further update the encoder and the classification network with only labeled images for multiclass classification of the OE images.

RESULTS

Our proposed semisupervised method CAESNet achieves the best average performance for multiclass classification of OE images, which surpasses the performance of supervised methods including standard convolutional networks and convolutional autoencoder network.

CONCLUSIONS

Our semisupervised CAESNet can efficiently utilize the unlabeled OE images, which improves the diagnosis and decision making for patients with Barrett's esophagus.

Innovative approaches for cancer treatment: current perspectives and new challenges

Every year, cancer is responsible for millions of deaths worldwide and, even though much progress has been achieved in medicine, there are still many issues that must be addressed in order to improve cancer therapy. For this reason, oncological research is putting a lot of effort towards finding new and efficient therapies which can alleviate critical side effects caused by conventional treatments. Different technologies are currently under evaluation in clinical trials or have been already introduced into clinical practice. While nanomedicine is contributing to the development of biocompatible materials both for diagnostic and therapeutic purposes, bioengineering of extracellular vesicles and cells derived from patients has allowed designing ad hoc systems and univocal targeting strategies. In this review, we will provide an in-depth analysis of the most innovative advances in basic and applied cancer research.

Integration of molecular features with clinical information for predicting outcomes for neuroblastoma patients

Background

Neuroblastoma is one of the most common types of pediatric cancer. In current neuroblastoma prognosis, patients can be stratified into high- and low-risk groups. Generally, more than 90% of the patients in the low-risk group will survive, while less than 50% for those with the high-risk disease will survive. Since the so-called “high-risk” patients still contain patients with mixed good and poor outcomes, more refined stratification needs to be established so that for the patients with poor outcome, they can receive prompt and individualized treatment to improve their long-term survival rate, while the patients with good outcome can avoid unnecessary over treatment.

Methods

We first mined co-expressed gene modules from microarray and RNA-seq data of neuroblastoma samples using the weighted network mining algorithm lmQCM, and summarize the resulted modules into eigengenes. Then patient similarity weight matrix was constructed with module eigengenes using two different approaches. At the last step, a consensus clustering method called Molecular Regularized Consensus Patient Stratification (MRCPS) was applied to aggregate both clinical information (clinical stage and clinical risk level) and multiple eigengene data for refined patient stratification.

Results

The integrative method MRCPS demonstrated superior performance to clinical staging or transcriptomic features alone for the NB cohort stratification. It successfully identified the worst prognosis group from the clinical high-risk group, with less than 40% survived in the first 50 months of diagnosis. It also identified highly differentially expressed genes between best prognosis group and worst prognosis group, which can be potential gene biomarkers for clinical testing.

Conclusions

To address the need for better prognosis and facilitate personalized treatment on neuroblastoma, we modified the recently developed bioinformatics workflow MRCPS for refined patient prognosis. It integrates clinical information and molecular features such as gene co-expression for prognosis. This clustering workflow is flexible, allowing the integration of both categorical and numerical data. The results demonstrate the power of survival prognosis with this integrative analysis workflow, with superior prognostic performance to only using transcriptomic data or clinical staging/risk information alone.

Reviewers

This article was reviewed by Lan Hu, Haibo Liu, Julie Zhu and Aleksandra Gruca.

Electronic supplementary material

The online version of this article (10.1186/s13062-019-0244-y) contains supplementary material, which is available to authorized users.

Correlation Analysis of Histopathology and Proteogenomics Data for Breast Cancer

Tumors are heterogeneous tissues with different types of cells such as cancer cells, fibroblasts, and lymphocytes. Although the morphological features of tumors are critical for cancer diagnosis and prognosis, the underlying molecular events and genes for tumor morphology are far from being clear. With the advancement in computational pathology and accumulation of large amount of cancer samples with matched molecular and histopathology data, researchers can carry out integrative analysis to investigate this issue. In this study, we systematically examine the relationships between morphological features and various molecular data in breast cancers. Specifically, we identified 73 breast cancer patients from the TCGA and CPTAC projects matched whole slide images, RNA-seq, and proteomic data. By calculating 100 different morphological features and correlating them with the transcriptomic and proteomic data, we inferred four major biological processes associated with various interpretable morphological features. These processes include metabolism, cell cycle, immune response, and extracellular matrix development, which are all hallmarks of cancers and the associated morphological features are related to area, density, and shapes of epithelial cells, fibroblasts, and lymphocytes. In addition, protein specific biological processes were inferred solely from proteomic data, suggesting the importance of proteomic data in obtaining a holistic understanding of the molecular basis for tumor tissue morphology. Furthermore, survival analysis yielded specific morphological features related to patient prognosis, which have a strong association with important molecular events based on our analysis. Overall, our study demonstrated the power for integrating multiple types of biological data for cancer samples in generating new hypothesis as well as identifying potential biomarkers predicting patient outcome. Future work includes causal analysis to identify key regulators for cancer tissue development and validating the findings using more independent data sets.

The Promise of Digital Biopsy for the Prediction of Tumor Molecular Features and Clinical Outcomes Associated With Immunotherapy

Immunotherapy by immune checkpoint inhibitors has emerged as an effective treatment for a slight proportion of patients with aggressive tumors. Currently, some molecular determinants, such as the expression of the programmed cell death ligand-1 (PD-L1) or the tumor mutational burden (TMB) have been used in the clinical practice as predictive biomarkers, although they fail in consistency, applicability, or reliability to precisely identify the responding patients mainly because of their spatial intratumoral heterogeneity. Therefore, new biomarkers for early prediction of patient response to immunotherapy, that could integrate several approaches, are eagerly sought. Novel methods of quantitative image analysis (such as radiomics or pathomics) might offer a comprehensive approach providing spatial and temporal information from macroscopic imaging features potentially predictive of underlying molecular drivers, tumor-immune microenvironment, tumor-related prognosis, and clinical outcome (in terms of response or toxicity) following immunotherapy. Preliminary results from radiomics and pathomics analysis have demonstrated their ability to correlate image features with PD-L1 tumor expression, high CD3 cell infiltration or CD8 cell expression, or to produce an image signature concordant with gene expression. Furthermore, the predictive power of radiomics and pathomics can be improved by combining information from other modalities, such as blood values or molecular features, leading to increase the accuracy of these models. Thus, "digital biopsy," which could be defined by non-invasive and non-consuming digital techniques provided by radiomics and pathomics, may have the potential to allow for personalized approach for cancer patients treated with immunotherapy.

Multi-objective Parameter Auto-tuning for Tissue Image Segmentation Workflows

We propose a software platform that integrates methods and tools for multi-objective parameter auto-tuning in tissue image segmentation workflows. The goal of our work is to provide an approach for improving the accuracy of nucleus/cell segmentation pipelines by tuning their input parameters. The shape, size, and texture features of nuclei in tissue are important biomarkers for disease prognosis, and accurate computation of these features depends on accurate delineation of boundaries of nuclei. Input parameters in many nucleus segmentation workflows affect segmentation accuracy and have to be tuned for optimal performance. This is a time-consuming and computationally expensive process; automating this step facilitates more robust image segmentation workflows and enables more efficient application of image analysis in large image datasets. Our software platform adjusts the parameters of a nuclear segmentation algorithm to maximize the quality of image segmentation results while minimizing the execution time. It implements several optimization methods to search the parameter space efficiently. In addition, the methodology is developed to execute on high-performance computing systems to reduce the execution time of the parameter tuning phase. These capabilities are packaged in a Docker container for easy deployment and can be used through a friendly interface extension in 3D Slicer. Our results using three real-world image segmentation workflows demonstrate that the proposed solution is able to (1) search a small fraction (about 100 points) of the parameter space, which contains billions to trillions of points, and improve the quality of segmentation output by × 1.20, × 1.29, and × 1.29, on average; (2) decrease the execution time of a segmentation workflow by up to 11.79× while improving output quality; and (3) effectively use parallel systems to accelerate parameter tuning and segmentation phases.

MaReIA: A Cloud MapReduce Based High Performance Whole Slide Image Analysis Framework

Recent advancements in systematic analysis of high resolution whole slide images have increase efficiency of diagnosis, prognosis and prediction of cancer and important diseases. Due to the enormous sizes and dimensions of whole slide images, the analysis requires extensive computing resources which are not commonly available. Images have to be tiled for processing due to computer memory limitations, which lead to inaccurate results due to the ignorance of boundary crossing objects. Thus, we propose a generic and highly scalable cloud-based image analysis framework for whole slide images. The framework enables parallelized integration of image analysis steps, such as segmentation and aggregation of micro-structures in a single pipeline, and generation of final objects manageable by databases. The core concept relies on the abstraction of objects in whole slide images as different classes of spatial geometries, which in turn can be handled as text based records in MapReduce. The framework applies an overlapping partitioning scheme on images, and provides parallelization of tiling and image segmentation based on MapReduce architecture. It further provides robust object normalization, graceful handling of boundary objects with an efficient spatial indexing based matching method to generate accurate results. Our experiments on Amazon EMR show that MaReIA is highly scalable, generic and extremely cost effective by benchmark tests.

Robust Histopathology Image Analysis: to Label or to Synthesize?

Detection, segmentation and classification of nuclei are fundamental analysis operations in digital pathology. Existing state-of-the-art approaches demand extensive amount of supervised training data from pathologists and may still perform poorly in images from unseen tissue types. We propose an unsupervised approach for histopathology image segmentation that synthesizes heterogeneous sets of training image patches, of every tissue type. Although our synthetic patches are not always of high quality, we harness the motley crew of generated samples through a generally applicable importance sampling method. This proposed approach, for the first time, re-weighs the training loss over synthetic data so that the ideal (unbiased) generalization loss over the true data distribution is minimized. This enables us to use a random polygon generator to synthesize approximate cellular structures (i.e., nuclear masks) for which no real examples are given in many tissue types, and hence, GAN-based methods are not suited. In addition, we propose a hybrid synthesis pipeline that utilizes textures in real histopathology patches and GAN models, to tackle heterogeneity in tissue textures. Compared with existing state-of-the-art supervised models, our approach generalizes significantly better on cancer types without training data. Even in cancer types with training data, our approach achieves the same performance without supervision cost. We release code and segmentation results on over 5000 Whole Slide Images (WSI) in The Cancer Genome Atlas (TCGA) repository, a dataset that would be orders of magnitude larger than what is available today.

Sensitivity analysis in digital pathology: Handling large number of parameters with compute expensive workflows

Digital pathology imaging enables valuable quantitative characterizations of tissue state at the sub-cellular level. While there is a growing set of methods for analysis of whole slide tissue images, many of them are sensitive to changes in input parameters. Evaluating how analysis results are affected by variations in input parameters is important for the development of robust methods. Executing algorithm sensitivity analyses by systematically varying input parameters is an expensive task because a single evaluation run with a moderate number of tissue images may take hours or days. Our work investigates the use of Surrogate Models (SMs) along with parallel execution to speed up parameter sensitivity analysis (SA). This approach significantly reduces the SA cost, because the SM execution is inexpensive. The evaluation of several SM strategies with two image segmentation workflows demonstrates that a SA study with SMs attains results close to a SA with real application runs (mean absolute error lower than 0.022), while the SM accelerates the SA execution by 51 × . We also show that, although the number of parameters in the example workflows is high, most of the uncertainty can be associated with a few parameters. In order to identify the impact of variations in segmentation results to downstream analyses, we carried out a survival analysis with 387 Lung Squamous Cell Carcinoma cases. This analysis was repeated using 3 values for the most significant parameters identified by the SA for the two segmentation algorithms; about 600 million cell nuclei were segmented per run. The results show that significance of the survival correlations of patient groups, assessed by a logrank test, are strongly affected by the segmentation parameter changes. This indicates that sensitivity analysis is an important tool for evaluating the stability of conclusions from image analyses.

Image analysis with deep learning to predict breast cancer grade, ER status, histologic subtype, and intrinsic subtype

RNA-based, multi-gene molecular assays are available and widely used for patients with ER-positive/HER2-negative breast cancers. However, RNA-based genomic tests can be costly and are not available in many countries. Methods for inferring molecular subtype from histologic images may identify patients most likely to benefit from further genomic testing. To identify patients who could benefit from molecular testing based on H&E stained histologic images, we developed an image analysis approach using deep learning. A training set of 571 breast tumors was used to create image-based classifiers for tumor grade, ER status, PAM50 intrinsic subtype, histologic subtype, and risk of recurrence score (ROR-PT). The resulting classifiers were applied to an independent test set (n = 288), and accuracy, sensitivity, and specificity of each was assessed on the test set. Histologic image analysis with deep learning distinguished low-intermediate vs. high tumor grade (82% accuracy), ER status (84% accuracy), Basal-like vs. non-Basal-like (77% accuracy), Ductal vs. Lobular (94% accuracy), and high vs. low-medium ROR-PT score (75% accuracy). Sampling considerations in the training set minimized bias in the test set. Incorrect classification of ER status was significantly more common for Luminal B tumors. These data provide proof of principle that molecular marker status, including a critical clinical biomarker (i.e., ER status), can be predicted with accuracy >75% based on H&E features. Image-based methods could be promising for identifying patients with a greater need for further genomic testing, or in place of classically scored variables typically accomplished using human-based scoring.

Fusion of encoder-decoder deep networks improves delineation of multiple nuclear phenotypes

BACKGROUND

Nuclear segmentation is an important step for profiling aberrant regions of histology sections. If nuclear segmentation can be resolved, then new biomarkers of nuclear phenotypes and their organization can be predicted for the application of precision medicine. However, segmentation is a complex problem as a result of variations in nuclear geometry (e.g., size, shape), nuclear type (e.g., epithelial, fibroblast), nuclear phenotypes (e.g., vesicular, aneuploidy), and overlapping nuclei. The problem is further complicated as a result of variations in sample preparation (e.g., fixation, staining). Our hypothesis is that (i) deep learning techniques can learn complex phenotypic signatures that rise in tumor sections, and (ii) fusion of different representations (e.g., regions, boundaries) contributes to improved nuclear segmentation.

RESULTS

We have demonstrated that training of deep encoder-decoder convolutional networks overcomes complexities associated with multiple nuclear phenotypes, where we evaluate alternative architecture of deep learning for an improved performance against the simplicity of the design. In addition, improved nuclear segmentation is achieved by color decomposition and combining region- and boundary-based features through a fusion network. The trained models have been evaluated against approximately 19,000 manually annotated nuclei, and object-level Precision, Recall, F1-score and Standard Error are reported with the best F1-score being 0.91. Raw training images, annotated images, processed images, and source codes are released as a part of the Additional file 1.

CONCLUSIONS

There are two intrinsic barriers in nuclear segmentation to H&E stained images, which correspond to the diversity of nuclear phenotypes and perceptual boundaries between adjacent cells. We demonstrate that (i) the encoder-decoder architecture can learn complex phenotypes that include the vesicular type; (ii) delineation of overlapping nuclei is enhanced by fusion of region- and edge-based networks; (iii) fusion of ENets produces an improved result over the fusion of UNets; and (iv) fusion of networks is better than multitask learning. We suggest that our protocol enables processing a large cohort of whole slide images for applications in precision medicine.

Image-based surrogate biomarkers for molecular subtypes of colorectal cancer

Motivation

Whole genome expression profiling of large cohorts of different types of cancer led to the identification of distinct molecular subcategories (subtypes) that may partially explain the observed inter-tumoral heterogeneity. This is also the case of colorectal cancer (CRC) where several such categorizations have been proposed. Despite recent developments, the problem of subtype definition and recognition remains open, one of the causes being the intrinsic heterogeneity of each tumor, which is difficult to estimate from gene expression profiles. However, one of the observations of these studies indicates that there may be links between the dominant tumor morphology characteristics and the molecular subtypes. Benefiting from a large collection of CRC samples, comprising both gene expression and histopathology images, we investigated the possibility of building image-based classifiers able to predict the molecular subtypes. We employed deep convolutional neural networks for extracting local descriptors which were then used for constructing a dictionary-based representation of each tumor sample. A set of support vector machine classifiers were trained to solve different binary decision problems, their combined outputs being used to predict one of the five molecular subtypes.

Results

A hierarchical decomposition of the multi-class problem was obtained with an overall accuracy of 0.84 (95%CI=0.79-0.88). The predictions from the image-based classifier showed significant prognostic value similar to their molecular counterparts.

Contact

popovici@iba.muni.cz.

Availability and Implementation

Source code used for the image analysis is freely available from https://github.com/higex/qpath .

Supplementary information

Supplementary data are available at Bioinformatics online.

Unsupervised morphological segmentation of tissue compartments in histopathological images

Algorithmic segmentation of histologically relevant regions of tissues in digitized histopathological images is a critical step towards computer-assisted diagnosis and analysis. For example, automatic identification of epithelial and stromal tissues in images is important for spatial localisation and guidance in the analysis and characterisation of tumour micro-environment. Current segmentation approaches are based on supervised methods, which require extensive training data from high quality, manually annotated images. This is often difficult and costly to obtain. This paper presents an alternative data-independent framework based on unsupervised segmentation of oropharyngeal cancer tissue micro-arrays (TMAs). An automated segmentation algorithm based on mathematical morphology is first applied to light microscopy images stained with haematoxylin and eosin. This partitions the image into multiple binary ‘virtual-cells’, each enclosing a potential ‘nucleus’ (dark basins in the haematoxylin absorbance image). Colour and morphology measurements obtained from these virtual-cells as well as their enclosed nuclei are input into an advanced unsupervised learning model for the identification of epithelium and stromal tissues. Here we exploit two Consensus Clustering (CC) algorithms for the unsupervised recognition of tissue compartments, that consider the consensual opinion of a group of individual clustering algorithms. Unlike most unsupervised segmentation analyses, which depend on a single clustering method, the CC learning models allow for more robust and stable detection of tissue regions. The proposed framework performance has been evaluated on fifty-five hand-annotated tissue images of oropharyngeal tissues. Qualitative and quantitative results of the proposed segmentation algorithm compare favourably with eight popular tissue segmentation strategies. Furthermore, the unsupervised results obtained here outperform those obtained with individual clustering algorithms.

A Containerized Software System for Generation, Management, and Exploration of Features from Whole Slide Tissue Images

Well-curated sets of pathology image features will be critical to clinical studies that aim to evaluate and predict treatment responses. Researchers require information synthesized across multiple biological scales, from the patient to the molecular scale, to more effectively study cancer. This article describes a suite of services and web applications that allow users to select regions of interest in whole slide tissue images, run a segmentation pipeline on the selected regions to extract nuclei and compute shape, size, intensity, and texture features, store and index images and analysis results, and visualize and explore images and computed features. All the services are deployed as containers and the user-facing interfaces as web-based applications. The set of containers and web applications presented in this article is used in cancer research studies of morphologic characteristics of tumor tissues. The software is free and open source. Cancer Res; 77(21); e79-82. ©2017 AACR.

An Assessment of Imaging Informatics for Precision Medicine in Cancer

Objectives: Precision medicine requires the measurement, quantification, and cataloging of medical characteristics to identify the most effective medical intervention. However, the amount of available data exceeds our current capacity to extract meaningful information. We examine the informatics needs to achieve precision medicine from the perspective of quantitative imaging and oncology. Methods: The National Cancer Institute (NCI) organized several workshops on the topic of medical imaging and precision medicine. The observations and recommendations are summarized herein. Results: Recommendations include: use of standards in data collection and clinical correlates to promote interoperability; data sharing and validation of imaging tools; clinician's feedback in all phases of research and development; use of open-source architecture to encourage reproducibility and reusability; use of challenges which simulate real-world situations to incentivize innovation; partnership with industry to facilitate commercialization; and education in academic communities regarding the challenges involved with translation of technology from the research domain to clinical utility and the benefits of doing so. Conclusions: This article provides a survey of the role and priorities for imaging informatics to help advance quantitative imaging in the era of precision medicine. While these recommendations were drawn from oncology, they are relevant and applicable to other clinical domains where imaging aids precision medicine.

Development of a Framework for Large Scale Three-Dimensional Pathology and Biomarker Imaging and Spatial Analytics

With the rapid advancement in large-throughput scanning technologies, digital pathology has emerged as platform with promise for diagnostic approaches, but also for high-throughput quantitative data extraction and analysis for translational research. Digital pathology and biomarker images are rich sources of information on tissue architecture, cell diversity and morphology, and molecular pathway activation. However, the understanding of disease in three-dimension (3D) has been hampered by their traditional two-dimension (2D) representations on histologic slides. In this paper, we propose a scalable image processing framework to quantitatively investigate 3D phenotypic and cell-specific molecular features from digital pathology and biomarker images in information- lossless 3D tissue space. We also develop a generalized 3D spatial data management framework with multi-level parallelism and provide a sustainable infrastructure for rapid spatial queries through scalable and efficient spatial data processing. The developed framework can facilitate biomedical research by efficiently processing large-scale, 3D pathology and in-situ biomarker imaging data.

Towards Generation, Management, and Exploration of Combined Radiomics and Pathomics Datasets for Cancer Research

Cancer is a complex multifactorial disease state and the ability to anticipate and steer treatment results will require information synthesis across multiple scales from the host to the molecular level. Radiomics and Pathomics, where image features are extracted from routine diagnostic Radiology and Pathology studies, are also evolving as valuable diagnostic and prognostic indicators in cancer. This information explosion provides new opportunities for integrated, multi-scale investigation of cancer, but also mandates a need to build systematic and integrated approaches to manage, query and mine combined Radiomics and Pathomics data. In this paper, we describe a suite of tools and web-based applications towards building a comprehensive framework to support the generation, management and interrogation of large volumes of Radiomics and Pathomics feature sets and the investigation of correlations between image features, molecular data, and clinical outcome.

Multiview boosting digital pathology analysis of prostate cancer

BACKGROUND AND OBJECTIVE

Various digital pathology tools have been developed to aid in analyzing tissues and improving cancer pathology. The multi-resolution nature of cancer pathology, however, has not been fully analyzed and utilized. Here, we develop an automated, cooperative, and multi-resolution method for improving prostate cancer diagnosis.

METHODS

Digitized tissue specimen images are obtained from 5 tissue microarrays (TMAs). The TMAs include 70 benign and 135 cancer samples (TMA1), 74 benign and 89 cancer samples (TMA2), 70 benign and 115 cancer samples (TMA3), 79 benign and 82 cancer samples (TMA4), and 72 benign and 86 cancer samples (TMA5). The tissue specimen images are segmented using intensity- and texture-based features. Using the segmentation results, a number of morphological features from lumens and epithelial nuclei are computed to characterize tissues at different resolutions. Applying a multiview boosting algorithm, tissue characteristics, obtained from differing resolutions, are cooperatively combined to achieve accurate cancer detection.

RESULTS

In segmenting prostate tissues, the multiview boosting method achieved≥ 0.97 AUC using TMA1. For detecting cancers, the multiview boosting method achieved an AUC of 0.98 (95% CI: 0.97-0.99) as trained on TMA2 and tested on TMA3, TMA4, and TMA5. The proposed method was superior to single-view approaches, utilizing features from a single resolution or merging features from all the resolutions. Moreover, the performance of the proposed method was insensitive to the choice of the training dataset. Trained on TMA3, TMA4, and TMA5, the proposed method obtained an AUC of 0.97 (95% CI: 0.96-0.98), 0.98 (95% CI: 0.96-0.99), and 0.97 (95% CI: 0.96-0.98), respectively.

CONCLUSIONS

The multiview boosting method is capable of integrating information from multiple resolutions in an effective and efficient fashion and identifying cancers with high accuracy. The multiview boosting method holds a great potential for improving digital pathology tools and research.

Evaluation of nucleus segmentation in digital pathology images through large scale image synthesis

Digital histopathology images with more than 1 Gigapixel are drawing more and more attention in clinical, biomedical research, and computer vision fields. Among the multiple observable features spanning multiple scales in the pathology images, the nuclear morphology is one of the central criteria for diagnosis and grading. As a result it is also the mostly studied target in image computing. Large amount of research papers have devoted to the problem of extracting nuclei from digital pathology images, which is the foundation of any further correlation study. However, the validation and evaluation of nucleus extraction have yet been formulated rigorously and systematically. Some researches report a human verified segmentation with thousands of nuclei, whereas a single whole slide image may contain up to million. The main obstacle lies in the difficulty of obtaining such a large number of validated nuclei, which is essentially an impossible task for pathologist. We propose a systematic validation and evaluation approach based on large scale image synthesis. This could facilitate a more quantitatively validated study for current and future histopathology image analysis field.

Whole Slide Imaging for Analytical Anatomic Pathology and Telepathology: Practical Applications Today, Promises, and Perils

Whole slide imaging (WSI) offers a convenient, tractable platform for measuring features of routine and special-stain histology or in immunohistochemistry staining by using digital image analysis (IA). We now routinely use IA for quantitative and qualitative analysis of theranostic markers such as human epidermal growth factor 2 (HER2/neu), estrogen and progesterone receptors, and Ki-67. Quantitative IA requires extensive validation, however, and may not always be the best approach, with pancreatic neuroendocrine tumors being one example in which a semiautomated approach may be preferable for patient care. We find that IA has great utility for objective assessment of gastrointestinal tract dysplasia, microvessel density in hepatocellular carcinoma, hepatic fibrosis and steatosis, renal fibrosis, and general quality analysis/quality control, although the applications of these to daily practice are still in development. Collaborations with bioinformatics specialists have explored novel applications to gliomas, including in silico approaches for mining histologic data and correlating with molecular and radiologic findings. We and many others are using WSI for rapid, remote-access slide reviews (telepathology), though technical factors currently limit its utility for routine, high-volume diagnostics. In our experience, the greatest current practical impact of WSI lies in facilitating long-term storage and retrieval of images while obviating the need to keep slides on site. Once the existing barriers of capital cost, validation, operator training, software design, and storage/back-up concerns are overcome, these technologies appear destined to be a cornerstone of precision medicine and personalized patient care, and to become a routine part of pathology practice.

Application Performance Analysis and Efficient Execution on Systems with multi-core CPUs, GPUs and MICs: A Case Study with Microscopy Image Analysis

We carry out a comparative performance study of multi-core CPUs, GPUs and Intel Xeon Phi (Many Integrated Core-MIC) with a microscopy image analysis application. We experimentally evaluate the performance of computing devices on core operations of the application. We correlate the observed performance with the characteristics of computing devices and data access patterns, computation complexities, and parallelization forms of the operations. The results show a significant variability in the performance of operations with respect to the device used. The performances of operations with regular data access are comparable or sometimes better on a MIC than that on a GPU. GPUs are more efficient than MICs for operations that access data irregularly, because of the lower bandwidth of the MIC for random data accesses. We propose new performance-aware scheduling strategies that consider variabilities in operation speedups. Our scheduling strategies significantly improve application performance compared to classic strategies in hybrid configurations.

Patch-based Convolutional Neural Network for Whole Slide Tissue Image Classification

Convolutional Neural Networks (CNN) are state-of-the-art models for many image classification tasks. However, to recognize cancer subtypes automatically, training a CNN on gigapixel resolution Whole Slide Tissue Images (WSI) is currently computationally impossible. The differentiation of cancer subtypes is based on cellular-level visual features observed on image patch scale. Therefore, we argue that in this situation, training a patch-level classifier on image patches will perform better than or similar to an image-level classifier. The challenge becomes how to intelligently combine patch-level classification results and model the fact that not all patches will be discriminative. We propose to train a decision fusion model to aggregate patch-level predictions given by patch-level CNNs, which to the best of our knowledge has not been shown before. Furthermore, we formulate a novel Expectation-Maximization (EM) based method that automatically locates discriminative patches robustly by utilizing the spatial relationships of patches. We apply our method to the classification of glioma and non-small-cell lung carcinoma cases into subtypes. The classification accuracy of our method is similar to the inter-observer agreement between pathologists. Although it is impossible to train CNNs on WSIs, we experimentally demonstrate using a comparable non-cancer dataset of smaller images that a patch-based CNN can outperform an image-based CNN.

SparkGIS: Efficient Comparison and Evaluation of Algorithm Results in Tissue Image Analysis Studies

Algorithm evaluation provides a means to characterize variability across image analysis algorithms, validate algorithms by comparison of multiple results, and facilitate algorithm sensitivity studies. The sizes of images and analysis results in pathology image analysis pose significant challenges in algorithm evaluation. We present SparkGIS, a distributed, in-memory spatial data processing framework to query, retrieve, and compare large volumes of analytical image result data for algorithm evaluation. Our approach combines the in-memory distributed processing capabilities of Apache Spark and the efficient spatial query processing of Hadoop-GIS. The experimental evaluation of SparkGIS for heatmap computations used to compare nucleus segmentation results from multiple images and analysis runs shows that SparkGIS is efficient and scalable, enabling algorithm evaluation and algorithm sensitivity studies on large datasets.

Patch-based Convolutional Neural Network for Whole Slide Tissue Image Classification

Convolutional Neural Networks (CNN) are state-of-the-art models for many image classification tasks. However, to recognize cancer subtypes automatically, training a CNN on gigapixel resolution Whole Slide Tissue Images (WSI) is currently computationally impossible. The differentiation of cancer subtypes is based on cellular-level visual features observed on image patch scale. Therefore, we argue that in this situation, training a patch-level classifier on image patches will perform better than or similar to an image-level classifier. The challenge becomes how to intelligently combine patch-level classification results and model the fact that not all patches will be discriminative. We propose to train a decision fusion model to aggregate patch-level predictions given by patch-level CNNs, which to the best of our knowledge has not been shown before. Furthermore, we formulate a novel Expectation-Maximization (EM) based method that automatically locates discriminative patches robustly by utilizing the spatial relationships of patches. We apply our method to the classification of glioma and non-small-cell lung carcinoma cases into subtypes. The classification accuracy of our method is similar to the inter-observer agreement between pathologists. Although it is impossible to train CNNs on WSIs, we experimentally demonstrate using a comparable non-cancer dataset of smaller images that a patch-based CNN can outperform an image-based CNN.

A robust nonlinear tissue-component discrimination method for computational pathology

Advances in digital pathology, specifically imaging instrumentation and data management, have allowed for the development of computational pathology tools with the potential for better, faster, and cheaper diagnosis, prognosis, and prediction of disease. Images of tissue sections frequently vary in color appearance across research laboratories and medical facilities because of differences in tissue fixation, staining protocols, and imaging instrumentation, leading to difficulty in the development of robust computational tools. To address this challenge, we propose a novel nonlinear tissue-component discrimination (NLTD) method to register automatically the color space of histopathology images and visualize individual tissue components, independent of color differences between images. Our results show that the NLTD method could effectively discriminate different tissue components from different types of tissues prepared at different institutions. Further, we demonstrate that NLTD can improve the accuracy of nuclear detection and segmentation algorithms, compared with using conventional color deconvolution methods, and can quantitatively analyze immunohistochemistry images. Together, the NLTD method is objective, robust, and effective, and can be easily implemented in the emerging field of computational pathology.

Scalable analysis of Big pathology image data cohorts using efficient methods and high-performance computing strategies

BACKGROUND

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

RESULTS

The proposed tools and methods take advantage of state-of-the-art parallel machines and efficient content-based image searching strategies. The content based image retrieval (CBIR) algorithms can quickly detect and retrieve image patches similar to a query patch using a hierarchical analysis approach. The analysis component based on high performance computing can carry out consensus clustering on 500,000 data points using a large shared memory system.

CONCLUSIONS

Our work demonstrates efficient CBIR algorithms and high performance computing can be leveraged for efficient analysis of large microscopy images to meet the challenges of clinically salient applications in pathology. These technologies enable researchers and clinical investigators to make more effective use of the rich informational content contained within digitized microscopy specimens.

TissueCypher(™): A systems biology approach to anatomic pathology

Background:

Current histologic methods for diagnosis are limited by intra- and inter-observer variability. Immunohistochemistry (IHC) methods are frequently used to assess biomarkers to aid diagnoses, however, IHC staining is variable and nonlinear and the manual interpretation is subjective. Furthermore, the biomarkers assessed clinically are typically biomarkers of epithelial cell processes. Tumors and premalignant tissues are not composed only of epithelial cells but are interacting systems of multiple cell types, including various stromal cell types that are involved in cancer development. The complex network of the tissue system highlights the need for a systems biology approach to anatomic pathology, in which quantification of system processes is combined with informatics tools to produce actionable scores to aid clinical decision-making.

Aims:

Here, we describe a quantitative, multiplexed biomarker imaging approach termed TissueCypher™ that applies systems biology to anatomic pathology. Applications of TissueCypher™ in understanding the tissue system of Barrett's esophagus (BE) and the potential use as an adjunctive tool in the diagnosis of BE are described.

Patients and Methods:

The TissueCypher™ Image Analysis Platform was used to assess 14 epithelial and stromal biomarkers with known diagnostic significance in BE in a set of BE biopsies with nondysplastic BE with reactive atypia (RA, n = 22) and Barrett's with high-grade dysplasia (HGD, n = 17). Biomarker and morphology features were extracted and evaluated in the confirmed BE HGD cases versus the nondysplastic BE cases with RA.

Results:

Multiple image analysis features derived from epithelial and stromal biomarkers, including immune biomarkers and morphology, showed significant differences between HGD and RA.

Conclusions:

The assessment of epithelial cell abnormalities combined with an assessment of cellular changes in the lamina propria may serve as an adjunct to conventional pathology in the assessment of BE.

Computational Pathology: A Path Ahead

CONTEXT

We define the scope and needs within the new discipline of computational pathology, a discipline critical to the future of both the practice of pathology and, more broadly, medical practice in general.

OBJECTIVE

To define the scope and needs of computational pathology.

DATA SOURCES

A meeting was convened in Boston, Massachusetts, in July 2014 prior to the annual Association of Pathology Chairs meeting, and it was attended by a variety of pathologists, including individuals highly invested in pathology informatics as well as chairs of pathology departments.

CONCLUSIONS

The meeting made recommendations to promote computational pathology, including clearly defining the field and articulating its value propositions; asserting that the value propositions for health care systems must include means to incorporate robust computational approaches to implement data-driven methods that aid in guiding individual and population health care; leveraging computational pathology as a center for data interpretation in modern health care systems; stating that realizing the value proposition will require working with institutional administrations, other departments, and pathology colleagues; declaring that a robust pipeline should be fostered that trains and develops future computational pathologists, for those with both pathology and nonpathology backgrounds; and deciding that computational pathology should serve as a hub for data-related research in health care systems. The dissemination of these recommendations to pathology and bioinformatics departments should help facilitate the development of computational pathology.

Novel genotype-phenotype associations in human cancers enabled by advanced molecular platforms and computational analysis of whole slide images

Technological advances in computing, imaging, and genomics have created new opportunities for exploring relationships between histology, molecular events, and clinical outcomes using quantitative methods. Slide scanning devices are now capable of rapidly producing massive digital image archives that capture histological details in high resolution. Commensurate advances in computing and image analysis algorithms enable mining of archives to extract descriptions of histology, ranging from basic human annotations to automatic and precisely quantitative morphometric characterization of hundreds of millions of cells. These imaging capabilities represent a new dimension in tissue-based studies, and when combined with genomic and clinical endpoints, can be used to explore biologic characteristics of the tumor microenvironment and to discover new morphologic biomarkers of genetic alterations and patient outcomes. In this paper, we review developments in quantitative imaging technology and illustrate how image features can be integrated with clinical and genomic data to investigate fundamental problems in cancer. Using motivating examples from the study of glioblastomas (GBMs), we demonstrate how public data from The Cancer Genome Atlas (TCGA) can serve as an open platform to conduct in silico tissue-based studies that integrate existing data resources. We show how these approaches can be used to explore the relation of the tumor microenvironment to genomic alterations and gene expression patterns and to define nuclear morphometric features that are predictive of genetic alterations and clinical outcomes. Challenges, limitations, and emerging opportunities in the area of quantitative imaging and integrative analyses are also discussed.

Mapping spatial heterogeneity in the tumor microenvironment: a new era for digital pathology

The emergent field of digital pathology employing automated image analysis techniques is to revolutionize traditional pathology at the center of clinical diagnostics. Histological images provide important tumor features unavailable in molecular profiling or omics data- the spatial context of tumor and stromal cells at single-cell resolution. Methods to map the spatial and morphological patterns of cancer and normal cells can contribute to a more comprehensive understanding of the highly heterogeneous tumor microenvironment. This review focuses on methods that help expand our knowledge of intra-tumoral spatial heterogeneity of the tumor microenvironment and their potential synergies with molecular profiling technologies.

Region Templates: Data Representation and Management for High-Throughput Image Analysis

We introduce a region template abstraction and framework for the efficient storage, management and processing of common data types in analysis of large datasets of high resolution images on clusters of hybrid computing nodes. The region template abstraction provides a generic container template for common data structures, such as points, arrays, regions, and object sets, within a spatial and temporal bounding box. It allows for different data management strategies and I/O implementations, while providing a homogeneous, unified interface to applications for data storage and retrieval. A region template application is represented as a hierarchical dataflow in which each computing stage may be represented as another dataflow of finer-grain tasks. The execution of the application is coordinated by a runtime system that implements optimizations for hybrid machines, including performance-aware scheduling for maximizing the utilization of computing devices and techniques to reduce the impact of data transfers between CPUs and GPUs. An experimental evaluation on a state-of-the-art hybrid cluster using a microscopy imaging application shows that the abstraction adds negligible overhead (about 3%) and achieves good scalability and high data transfer rates. Optimizations in a high speed disk based storage implementation of the abstraction to support asynchronous data transfers and computation result in an application performance gain of about 1.13×. Finally, a processing rate of 11,730 4K×4K tiles per minute was achieved for the microscopy imaging application on a cluster with 100 nodes (300 GPUs and 1,200 CPU cores). This computation rate enables studies with very large datasets.

NCI Workshop Report: Clinical and Computational Requirements for Correlating Imaging Phenotypes with Genomics Signatures

The National Cancer Institute (NCI) Cancer Imaging Program organized two related workshops on June 26–27, 2013, entitled “Correlating Imaging Phenotypes with Genomics Signatures Research” and “Scalable Computational Resources as Required for Imaging-Genomics Decision Support Systems.” The first workshop focused on clinical and scientific requirements, exploring our knowledge of phenotypic characteristics of cancer biological properties to determine whether the field is sufficiently advanced to correlate with imaging phenotypes that underpin genomics and clinical outcomes, and exploring new scientific methods to extract phenotypic features from medical images and relate them to genomics analyses. The second workshop focused on computational methods that explore informatics and computational requirements to extract phenotypic features from medical images and relate them to genomics analyses and improve the accessibility and speed of dissemination of existing NIH resources. These workshops linked clinical and scientific requirements of currently known phenotypic and genotypic cancer biology characteristics with imaging phenotypes that underpin genomics and clinical outcomes. The group generated a set of recommendations to NCI leadership and the research community that encourage and support development of the emerging radiogenomics research field to address short-and longer-term goals in cancer research.

Automated grading of renal cell carcinoma using whole slide imaging

INTRODUCTION

Recent technology developments have demonstrated the benefit of using whole slide imaging (WSI) in computer-aided diagnosis. In this paper, we explore the feasibility of using automatic WSI analysis to assist grading of clear cell renal cell carcinoma (RCC), which is a manual task traditionally performed by pathologists.

MATERIALS AND METHODS

Automatic WSI analysis was applied to 39 hematoxylin and eosin-stained digitized slides of clear cell RCC with varying grades. Kernel regression was used to estimate the spatial distribution of nuclear size across the entire slides. The analysis results were correlated with Fuhrman nuclear grades determined by pathologists.

RESULTS

The spatial distribution of nuclear size provided a panoramic view of the tissue sections. The distribution images facilitated locating regions of interest, such as high-grade regions and areas with necrosis. The statistical analysis showed that the maximum nuclear size was significantly different (P < 0.001) between low-grade (Grades I and II) and high-grade tumors (Grades III and IV). The receiver operating characteristics analysis showed that the maximum nuclear size distinguished high-grade and low-grade tumors with a false positive rate of 0.2 and a true positive rate of 1.0. The area under the curve is 0.97.

CONCLUSION

The automatic WSI analysis allows pathologists to see the spatial distribution of nuclei size inside the tumors. The maximum nuclear size can also be used to differentiate low-grade and high-grade clear cell RCC with good sensitivity and specificity. These data suggest that automatic WSI analysis may facilitate pathologic grading of renal tumors and reduce variability encountered with manual grading.

Quantitative modeling of clinical, cellular, and extracellular matrix variables suggest prognostic indicators in cancer: a model in neuroblastoma

BACKGROUND

Risk classification and treatment stratification for cancer patients is restricted by our incomplete picture of the complex and unknown interactions between the patient's organism and tumor tissues (transformed cells supported by tumor stroma). Moreover, all clinical factors and laboratory studies used to indicate treatment effectiveness and outcomes are by their nature a simplification of the biological system of cancer, and cannot yet incorporate all possible prognostic indicators.

METHODS

A multiparametric analysis on 184 tumor cylinders was performed. To highlight the benefit of integrating digitized medical imaging into this field, we present the results of computational studies carried out on quantitative measurements, taken from stromal and cancer cells and various extracellular matrix fibers interpenetrated by glycosaminoglycans, and eight current approaches to risk stratification systems in patients with primary and nonprimary neuroblastoma.

RESULTS

New tumor tissue indicators from both fields, the cellular and the extracellular elements, emerge as reliable prognostic markers for risk stratification and could be used as molecular targets of specific therapies.

CONCLUSION

The key to dealing with personalized therapy lies in the mathematical modeling. The use of bioinformatics in patient-tumor-microenvironment data management allows a predictive model in neuroblastoma.

High-Performance Computational Analysis of Glioblastoma Pathology Images with Database Support Identifies Molecular and Survival Correlates

In this paper, we present a novel framework for microscopic image analysis of nuclei, data management, and high performance computation to support translational research involving nuclear morphometry features, molecular data, and clinical outcomes. Our image analysis pipeline consists of nuclei segmentation and feature computation facilitated by high performance computing with coordinated execution in multi-core CPUs and Graphical Processor Units (GPUs). All data derived from image analysis are managed in a spatial relational database supporting highly efficient scientific queries. We applied our image analysis workflow to 159 glioblastomas (GBM) from The Cancer Genome Atlas dataset. With integrative studies, we found statistics of four specific nuclear features were significantly associated with patient survival. Additionally, we correlated nuclear features with molecular data and found interesting results that support pathologic domain knowledge. We found that Proneural subtype GBMs had the smallest mean of nuclear Eccentricity and the largest mean of nuclear Extent, and MinorAxisLength. We also found gene expressions of stem cell marker MYC and cell proliferation maker MKI67 were correlated with nuclear features. To complement and inform pathologists of relevant diagnostic features, we queried the most representative nuclear instances from each patient population based on genetic and transcriptional classes. Our results demonstrate that specific nuclear features carry prognostic significance and associations with transcriptional and genetic classes, highlighting the potential of high throughput pathology image analysis as a complementary approach to human-based review and translational research.