Automatic classification for pathological prostate images based on fractal analysis

Exploring automatic prostate histopathology image Gleason grading via local structure modeling

Gleason-grading of prostate cancer pathology specimens reveal the malignancy of the cancer tissues, thus provides critical guidance for prostate cancer diagnoses and treatment. Computer-aided automatic grading methods have been providing efficient and result-consistent alternative to traditional manually slide reading approach, through statistical and structural feature analysis of the digitized pathology slides. In this paper, we propose a novel automatic Gleason grading algorithm through local structure model learning and classification. We use attributed graph to represent the tissue glandular structures in histopathology images; representative sub-graphs features were learned as bags-of-words features from labeled samples of each grades. Then structural similarity between sub-graphs in the unlabeled images and the representative sub-graphs were obtained using the learned codebook. Gleason grade was given based on an overall similarity score. We validated the proposed algorithm on 300 prostate histopathology images from the TCGA dataset, and the algorithm achieved average grading accuracy of 91.25%, 76.36% and 64.75% on images with Gleason grade 3, 4 and 5 respectively.

Computer-aided diagnosis from weak supervision: a benchmarking study

Supervised machine learning is a powerful tool frequently used in computer-aided diagnosis (CAD) applications. The bottleneck of this technique is its demand for fine grained expert annotations, which are tedious for medical image analysis applications. Furthermore, information is typically localized in diagnostic images, which makes representation of an entire image by a single feature set problematic. The multiple instance learning framework serves as a remedy to these two problems by allowing labels to be provided for groups of observations, called bags, and assuming the group label to be the maximum of the instance labels within the bag. This setup can effectively be applied to CAD by splitting a given diagnostic image into a Cartesian grid, treating each grid element (patch) as an instance by representing it with a feature set, and grouping instances belonging to the same image into a bag. We quantify the power of existing multiple instance learning methods by evaluating their performance on two distinct CAD applications: (i) Barrett's cancer diagnosis and (ii) diabetic retinopathy screening. In the experiments, mi-Graph appears as the best-performing method in bag-level prediction (i.e. diagnosis) for both of these applications that have drastically different visual characteristics. For instance-level prediction (i.e. disease localization), mi-SVM ranks as the most accurate method.

Selective invocation of shape priors for deformable segmentation and morphologic classification of prostate cancer tissue microarrays

Shape based active contours have emerged as a natural solution to overlap resolution. However, most of these shape-based methods are computationally expensive. There are instances in an image where no overlapping objects are present and applying these schemes results in significant computational overhead without any accompanying, additional benefit. In this paper we present a novel adaptive active contour scheme (AdACM) that combines boundary and region based energy terms with a shape prior in a multi level set formulation. To reduce the computational overhead, the shape prior term in the variational formulation is only invoked for those instances in the image where overlaps between objects are identified; these overlaps being identified via a contour concavity detection scheme. By not having to invoke all three terms (shape, boundary, region) for segmenting every object in the scene, the computational expense of the integrated active contour model is dramatically reduced, a particularly relevant consideration when multiple objects have to be segmented on very large histopathological images. The AdACM was employed for the task of segmenting nuclei on 80 prostate cancer tissue microarray images from 40 patient studies. Nuclear shape based, architectural and textural features extracted from these segmentations were extracted and found to able to discriminate different Gleason grade patterns with a classification accuracy of 86% via a quadratic discriminant analysis (QDA) classifier. On average the AdACM model provided 60% savings in computational times compared to a non-optimized hybrid active contour model involving a shape prior.

Computational hepatocellular carcinoma tumor grading based on cell nuclei classification

Hepatocellular carcinoma (HCC) is the most common histological type of primary liver cancer. HCC is graded according to the malignancy of the tissues. It is important to diagnose low-grade HCC tumors because these tissues have good prognosis. Image interpretation-based computer-aided diagnosis (CAD) systems have been developed to automate the HCC grading process. Generally, the HCC grade is determined by the characteristics of liver cell nuclei. Therefore, it is preferable that CAD systems utilize only liver cell nuclei for HCC grading. This paper proposes an automated HCC diagnosing method. In particular, it defines a pipeline-path that excludes nonliver cell nuclei in two consequent pipeline-modules and utilizes the liver cell nuclear features for HCC grading. The significance of excluding the nonliver cell nuclei for HCC grading is experimentally evaluated. Four categories of liver cell nuclear features were utilized for classifying the HCC tumors. Results indicated that nuclear texture is the dominant feature for HCC grading and others contribute to increase the classification accuracy. The proposed method was employed to classify a set of regions of interest selected from HCC whole slide images into five classes and resulted in a 95.97% correct classification rate.

Computer-Aided Prostate Cancer Diagnosis From Digitized Histopathology: A Review on Texture-Based Systems

Prostate cancer (PCa) is currently diagnosed by microscopic evaluation of biopsy samples. Since tissue assessment heavily relies on the pathologists level of expertise and interpretation criteria, it is still a subjective process with high intra- and interobserver variabilities. Computer-aided diagnosis (CAD) may have a major impact on detection and grading of PCa by reducing the pathologists reading time, and increasing the accuracy and reproducibility of diagnosis outcomes. However, the complexity of the prostatic tissue and the large volumes of data generated by biopsy procedures make the development of CAD systems for PCa a challenging task. The problem of automated diagnosis of prostatic carcinoma from histopathology has received a lot of attention. As a result, a number of CAD systems, have been proposed for quantitative image analysis and classification. This review aims at providing a detailed description of selected literature in the field of CAD of PCa, emphasizing the role of texture analysis methods in tissue description. It includes a review of image analysis tools for image preprocessing, feature extraction, classification, and validation techniques used in PCa detection and grading, as well as future directions in pursuit of better texture-based CAD systems.

Removing batch effects from histopathological images for enhanced cancer diagnosis

Researchers have developed computer-aided decision support systems for translational medicine that aim to objectively and efficiently diagnose cancer using histopathological images. However, the performance of such systems is confounded by nonbiological experimental variations or "batch effects" that can commonly occur in histopathological data, especially when images are acquired using different imaging devices and patient samples. This is even more problematic in large-scale studies in which cross-laboratory sharing of large volumes of data is necessary. Batch effects can change quantitative morphological image features and decrease the prediction performance. Using four batches of renal tumor images, we compare one image-level and five feature-level batch effect removal methods. Principal component variation analysis shows that batch is a large source of variance in image features. Results show that feature-level normalization methods reduce batch-contributed variance to almost zero. Moreover, feature-level normalization, especially ComBatN, improves cross-batch and combined-batch prediction performance. Compared to no normalization, ComBatN improves performance in 83% and 90% of cross-batch and combined-batch prediction models, respectively.

Max-AUC feature selection in computer-aided detection of polyps in CT colonography

We propose a feature selection method based on a sequential forward floating selection (SFFS) procedure to improve the performance of a classifier in computerized detection of polyps in CT colonography (CTC). The feature selection method is coupled with a nonlinear support vector machine (SVM) classifier. Unlike the conventional linear method based on Wilks' lambda, the proposed method selected the most relevant features that would maximize the area under the receiver operating characteristic curve (AUC), which directly maximizes classification performance, evaluated based on AUC value, in the computer-aided detection (CADe) scheme. We presented two variants of the proposed method with different stopping criteria used in the SFFS procedure. The first variant searched all feature combinations allowed in the SFFS procedure and selected the subsets that maximize the AUC values. The second variant performed a statistical test at each step during the SFFS procedure, and it was terminated if the increase in the AUC value was not statistically significant. The advantage of the second variant is its lower computational cost. To test the performance of the proposed method, we compared it against the popular stepwise feature selection method based on Wilks' lambda for a colonic-polyp database (25 polyps and 2624 nonpolyps). We extracted 75 morphologic, gray-level-based, and texture features from the segmented lesion candidate regions. The two variants of the proposed feature selection method chose 29 and 7 features, respectively. Two SVM classifiers trained with these selected features yielded a 96% by-polyp sensitivity at false-positive (FP) rates of 4.1 and 6.5 per patient, respectively. Experiments showed a significant improvement in the performance of the classifier with the proposed feature selection method over that with the popular stepwise feature selection based on Wilks' lambda that yielded 18.0 FPs per patient at the same sensitivity level.

Separation of benign and malignant glands in prostatic adenocarcinoma

This paper presents an analysis of the high resolution histopathology images of the prostate with a focus on the evolution of morphological gland features in prostatic adenocarcinoma. Here we propose a novel technique of labeling individual glands as malignant or benign. In the first step, the gland and nuclei objects of the images are automatically segmented. Individual gland units are segmented out by consolidating their lumina with the surrounding layers of epithelium and nuclei. The nuclei objects are segmented by using a marker controlled watershed algorithm. Two new features, Number of Nuclei Layer (N(NL)) and Ratio of Epithelial layer area to Lumen area (R(EL)) have been extracted from the segmented units. The main advantage of this approach is that it can detect individual malignant gland units, irrespective of neighboring histology and/or the spatial extent of the cancer. The proposed algorithm has been tested on 40 histopathology scenes taken from 10 high resolution whole mount images and achieved a sensitivity of 0.83 and specificity of 0.81 in a leave-75%-out cross-validation.

Weakly supervised histopathology cancer image segmentation and classification

Labeling a histopathology image as having cancerous regions or not is a critical task in cancer diagnosis; it is also clinically important to segment the cancer tissues and cluster them into various classes. Existing supervised approaches for image classification and segmentation require detailed manual annotations for the cancer pixels, which are time-consuming to obtain. In this paper, we propose a new learning method, multiple clustered instance learning (MCIL) (along the line of weakly supervised learning) for histopathology image segmentation. The proposed MCIL method simultaneously performs image-level classification (cancer vs. non-cancer image), medical image segmentation (cancer vs. non-cancer tissue), and patch-level clustering (different classes). We embed the clustering concept into the multiple instance learning (MIL) setting and derive a principled solution to performing the above three tasks in an integrated framework. In addition, we introduce contextual constraints as a prior for MCIL, which further reduces the ambiguity in MIL. Experimental results on histopathology colon cancer images and cytology images demonstrate the great advantage of MCIL over the competing methods.

Nuclear morphometry, epigenetic changes, and clinical relevance in prostate cancer

Nuclear structure alterations in cancer involve global genetic (mutations, amplifications, copy number variations, translocations, etc.) and epigenetic (DNA methylation and histone modifications) events that dramatically and dynamically spatially change chromatin, nuclear body, and chromosome organization. In prostate cancer (CaP) there appears to be early (<50 years) versus late (>60 years) onset clinically significant cancers, and we have yet to clearly understand the hereditary and somatic-based molecular pathways involved. We do know that once cancer is initiated, dedifferentiation of the prostate gland occurs with significant changes in nuclear structure driven by numerous genetic and epigenetic processes. This review focuses upon the nuclear architecture and epigenetic dynamics with potential translational clinically relevant applications to CaP. Further, the review correlates changes in the cancer-driven epigenetic process at the molecular level and correlates these alterations to nuclear morphological quantitative measurements. Finally, we address how we can best utilize this knowledge to improve the efficacy of personalized treatment of cancer.

Prostate histopathology: learning tissue component histograms for cancer detection and classification

Radical prostatectomy is performed on approximately 40% of men with organ-confined prostate cancer. Pathologic information obtained from the prostatectomy specimen provides important prognostic information and guides recommendations for adjuvant treatment. The current pathology protocol in most centers involves primarily qualitative assessment. In this paper, we describe and evaluate our system for automatic prostate cancer detection and grading on hematoxylin & eosin-stained tissue images. Our approach is intended to address the dual challenges of large data size and the need for high-level tissue information about the locations and grades of tumors. Our system uses two stages of AdaBoost-based classification. The first provides high-level tissue component labeling of a superpixel image partitioning. The second uses the tissue component labeling to provide a classification of cancer versus noncancer, and low-grade versus high-grade cancer. We evaluated our system using 991 sub-images extracted from digital pathology images of 50 whole-mount tissue sections from 15 prostatectomy patients. We measured accuracies of 90% and 85% for the cancer versus noncancer and high-grade versus low-grade classification tasks, respectively. This system represents a first step toward automated cancer quantification on prostate digital histopathology imaging, which could pave the way for more accurately informed postprostatectomy patient care.

Statistical Shape Model for Manifold Regularization: Gleason grading of prostate histology

Gleason patterns of prostate cancer histopathology, characterized primarily by morphological and architectural attributes of histological structures (glands and nuclei), have been found to be highly correlated with disease aggressiveness and patient outcome. Gleason patterns 4 and 5 are highly correlated with more aggressive disease and poorer patient outcome, while Gleason patterns 1-3 tend to reflect more favorable patient outcome. Because Gleason grading is done manually by a pathologist visually examining glass (or digital) slides subtle morphologic and architectural differences of histological attributes, in addition to other factors, may result in grading errors and hence cause high inter-observer variability. Recently some researchers have proposed computerized decision support systems to automatically grade Gleason patterns by using features pertaining to nuclear architecture, gland morphology, as well as tissue texture. Automated characterization of gland morphology has been shown to distinguish between intermediate Gleason patterns 3 and 4 with high accuracy. Manifold learning (ML) schemes attempt to generate a low dimensional manifold representation of a higher dimensional feature space while simultaneously preserving nonlinear relationships between object instances. Classification can then be performed in the low dimensional space with high accuracy. However ML is sensitive to the samples contained in the dataset; changes in the dataset may alter the manifold structure. In this paper we present a manifold regularization technique to constrain the low dimensional manifold to a specific range of possible manifold shapes, the range being determined via a statistical shape model of manifolds (SSMM). In this work we demonstrate applications of the SSMM in (1) identifying samples on the manifold which contain noise, defined as those samples which deviate from the SSMM, and (2) accurate out-of-sample extrapolation (OSE) of newly acquired samples onto a manifold constrained by the SSMM. We demonstrate these applications of the SSMM in the context of distinguish between Gleason patterns 3 and 4 using glandular morphologic features in a prostate histopathology dataset of 58 patient studies. Identifying and eliminating noisy samples from the manifold via the SSMM results in a statistically significant improvement in area under the receiver operator characteristic curve (AUC), 0.832 ± 0.048 with removal of noisy samples compared to a AUC of 0.779 ± 0.075 without removal of samples. The use of the SSMM for OSE of newly acquired glands also shows statistically significant improvement in AUC, 0.834 ± 0.051 with the SSMM compared to 0.779 ± 0.054 without the SSMM. Similar results were observed for the synthetic Swiss Roll and Helix datasets.

Pathology imaging informatics for quantitative analysis of whole-slide images

Objectives

With the objective of bringing clinical decision support systems to reality, this article reviews histopathological whole-slide imaging informatics methods, associated challenges, and future research opportunities.

Target audience

This review targets pathologists and informaticians who have a limited understanding of the key aspects of whole-slide image (WSI) analysis and/or a limited knowledge of state-of-the-art technologies and analysis methods.

Scope

First, we discuss the importance of imaging informatics in pathology and highlight the challenges posed by histopathological WSI. Next, we provide a thorough review of current methods for: quality control of histopathological images; feature extraction that captures image properties at the pixel, object, and semantic levels; predictive modeling that utilizes image features for diagnostic or prognostic applications; and data and information visualization that explores WSI for de novo discovery. In addition, we highlight future research directions and discuss the impact of large public repositories of histopathological data, such as the Cancer Genome Atlas, on the field of pathology informatics. Following the review, we present a case study to illustrate a clinical decision support system that begins with quality control and ends with predictive modeling for several cancer endpoints. Currently, state-of-the-art software tools only provide limited image processing capabilities instead of complete data analysis for clinical decision-making. We aim to inspire researchers to conduct more research in pathology imaging informatics so that clinical decision support can become a reality.

New morphological features for grading pancreatic ductal adenocarcinomas

Pathological diagnosis is influenced by subjective factors such as the individual experience and knowledge of doctors. Therefore, it may be interpreted in different ways for the same symptoms. The appearance of digital pathology has created good foundation for objective diagnoses based on quantitative feature analysis. Recently, numerous studies are being done to develop automated diagnosis based on the digital pathology. But there are as of yet no general automated methods for pathological diagnosis due to its specific nature. Therefore, specific methods according to a type of disease and a lesion could be designed. This study proposes quantitative features that are designed to diagnose pancreatic ductal adenocarcinomas. In the diagnosis of pancreatic ductal adenocarcinomas, the region of interest is a duct that consists of lumen and epithelium. Therefore, we first segment the lumen and epithelial nuclei from a tissue image. Then, we extract the specific features to diagnose the pancreatic ductal adenocarcinoma from the segmented objects. The experiment evaluated the classification performance of the SVM learned by the proposed features. The results showed an accuracy of 94.38% in the experiment distinguishing between pancreatic ductal adenocarcinomas and normal tissue and a classification accuracy of 77.03% distinguishing between the stages of pancreatic ductal adenocarcinomas.

How bioinformatics influences health informatics: usage of biomolecular sequences, expression profiles and automated microscopic image analyses for clinical needs and public health

ABSTRACT

The currently hyped expectation of personalized medicine is often associated with just achieving the information technology led integration of biomolecular sequencing, expression and histopathological bioimaging data with clinical records at the individual patients' level as if the significant biomedical conclusions would be its more or less mandatory result. It remains a sad fact that many, if not most biomolecular mechanisms that translate the human genomic information into phenotypes are not known and, thus, most of the molecular and cellular data cannot be interpreted in terms of biomedically relevant conclusions. Whereas the historical trend will certainly be into the general direction of personalized diagnostics and cures, the temperate view suggests that biomedical applications that rely either on the comparison of biomolecular sequences and/or on the already known biomolecular mechanisms have much greater chances to enter clinical practice soon. In addition to considering the general trends, we exemplarily review advances in the area of cancer biomarker discovery, in the clinically relevant characterization of patient-specific viral and bacterial pathogens (with emphasis on drug selection for influenza and enterohemorrhagic E. coli) as well as progress in the automated assessment of histopathological images. As molecular and cellular data analysis will become instrumental for achieving desirable clinical outcomes, the role of bioinformatics and computational biology approaches will dramatically grow.

AUTHOR SUMMARY

With DNA sequencing and computers becoming increasingly cheap and accessible to the layman, the idea of integrating biomolecular and clinical patient data seems to become a realistic, short-term option that will lead to patient-specific diagnostics and treatment design for many diseases such as cancer, metabolic disorders, inherited conditions, etc. These hyped expectations will fail since many, if not most biomolecular mechanisms that translate the human genomic information into phenotypes are not known yet and, thus, most of the molecular and cellular data collected will not lead to biomedically relevant conclusions. At the same time, less spectacular biomedical applications based on biomolecular sequence comparison and/or known biomolecular mechanisms have the potential to unfold enormous potential for healthcare and public health. Since the analysis of heterogeneous biomolecular data in context with clinical data will be increasingly critical, the role of bioinformatics and computational biology will grow correspondingly in this process.

Automatic differential diagnosis of pancreatic serous and mucinous cystadenomas based on morphological features

Generally, pathological diagnosis using an electron microscope is time-consuming and likely to result in a subjective judgment, because pathologists perform manual screening of tissue slides at high magnifications. Recently, the advent of digital pathology technology has provided the basis for convenient screening and quantitative analysis by digitizing tissue slides through a computer system. However, a screening process with high magnification still takes quite a long time. To solve these problems, recently the use of computer-aided design techniques for performing pathologic diagnosis has been increasing in digital pathology. For pathological diagnosis, we need different diagnostic methods for different regions with different characteristics. Therefore, in order to effectively diagnose different lesions and types of diseases, a quantitative method for extracting specific features is required in computerized pathologic diagnosis. This study is about an automated differential diagnosis system to differentiate between benign serous cystadenoma and possibly-malignant mucinous cystadenoma. In order to diagnose cystic tumors, the first step is identifying a cystic region and inspecting its epithelial cells. First, we identify the lumen boundary of a cyst using the Direction Cumulative Map considering 8-ways. Then, the Epithelial Nuclei Identification algorithm is used to discern epithelial nuclei. After that, three morphological features for the differential diagnosis of mucinous and serous cystadenomas are extracted. To demonstrate the superiority of the proposed features, the experiments compared performance of the classifiers learned by using the proposed morphological features and the classical morphological features based on nuclei. The classifiers in the simulations are as follows; Bayesian Classifier, k-Nearest Neighbors, Support Vector Machine, and Artificial Neural Network. The results show that all classifiers using the proposed features have the best classification performance.

A parallel solution for high resolution histological image analysis

This paper describes a general methodology for developing parallel image processing algorithms based on message passing for high resolution images (on the order of several Gigabytes). These algorithms have been applied to histological images and must be executed on massively parallel processing architectures. Advances in new technologies for complete slide digitalization in pathology have been combined with developments in biomedical informatics. However, the efficient use of these digital slide systems is still a challenge. The image processing that these slides are subject to is still limited both in terms of data processed and processing methods. The work presented here focuses on the need to design and develop parallel image processing tools capable of obtaining and analyzing the entire gamut of information included in digital slides. Tools have been developed to assist pathologists in image analysis and diagnosis, and they cover low and high-level image processing methods applied to histological images. Code portability, reusability and scalability have been tested by using the following parallel computing architectures: distributed memory with massive parallel processors and two networks, INFINIBAND and Myrinet, composed of 17 and 1024 nodes respectively. The parallel framework proposed is flexible, high performance solution and it shows that the efficient processing of digital microscopic images is possible and may offer important benefits to pathology laboratories.

Quantifying optical microangiography images obtained from a spectral domain optical coherence tomography system

The blood vessel morphology is known to correlate with several diseases, such as cancer, and is important for describing several tissue physiological processes, like angiogenesis. Therefore, a quantitative method for characterizing the angiography obtained from medical images would have several clinical applications. Optical microangiography (OMAG) is a method for obtaining three-dimensional images of blood vessels within a volume of tissue. In this study we propose to quantify OMAG images obtained with a spectral domain optical coherence tomography system. A technique for determining three measureable parameters (the fractal dimension, the vessel length fraction, and the vessel area density) is proposed and validated. Finally, the repeatability for acquiring OMAG images is determined, and a new method for analyzing small areas from these images is proposed.

Automatic detection of melanoma progression by histological analysis of secondary sites

We present results from machine classification of melanoma biopsies sectioned and stained with hematoxylin/eosin (H&E) on tissue microarrays (TMA). The four stages of melanoma progression were represented by seven tissue types, including benign nevus, primary tumors with radial and vertical growth patterns (stage I) and four secondary metastatic tumors: subcutaneous (stage II), lymph node (stage III), gastrointestinal and soft tissue (stage IV). Our experiment setup comprised 14,208 image samples based on 164 TMA cores. In our experiments, we constructed an HE color space by digitally deconvolving the RGB images into separate H (hematoxylin) and E (eosin) channels. We also compared three different classifiers: Weighted Neighbor Distance (WND), Radial Basis Functions (RBF), and k-Nearest Neighbors (kNN). We found that the HE color space consistently outperformed other color spaces with all three classifiers, while the different classifiers did not have as large of an effect on accuracy. This showed that a more physiologically relevant representation of color can have a larger effect on correct image interpretation than downstream processing steps. We were able to correctly classify individual fields of view with an average of 96% accuracy when randomly splitting the dataset into training and test fields. We also obtained a classification accuracy of 100% when testing entire cores that were not previously used in training (four random trials with one test core for each of 7 classes, 28 tests total). Because each core corresponded to a different patient, this test more closely mimics a clinically relevant setting where new patients are evaluated based on training with previous cases. The analysis method used in this study contains no parameters or adjustments that are specific to melanoma morphology, suggesting it can be used for analyzing other tissues and phenotypes, as well as potentially different image modalities and contrast techniques.

Classifying prostate cancer malignancy by quantitative histomorphometry

PURPOSE

Prostate cancer is routinely graded according to the Gleason grading scheme. This scheme is predominantly based on the textural appearance of aberrant glandular structures. Gleason grade is difficult to standardize and often leads to discussion due to interrater and intrarater disagreement. Thus, we investigated whether digital image based automated quantitative histomorphometry could be used to achieve a more standardized, reproducible classification outcome.

MATERIALS AND METHODS

In a proof of principle study we developed a method to evaluate digitized histological images of single prostate cancer regions in hematoxylin and eosin stained sections. Preprocessed color images were subjected to color deconvolution, followed by the binarization of obtained hematoxylin related image channels. Highlighted neoplastic epithelial gland related objects were morphometrically assessed by a classifier based on 2 calculated quantitative and objective geometric measures, that is inverse solidity and inverse compactness. The procedure was then applied to the prostate cancer probes of 125 patients. Each probe was independently classified for Gleason grade 3, 4 or 5 by an experienced pathologist blinded to image analysis outcome.

RESULTS

Together inverse compactness and inverse solidity were adequate discriminatory features for a powerful classifier that distinguished Gleason grade 3 from grade 4/5 histology. The classifier was robust on sensitivity analysis.

CONCLUSIONS

Results suggest that quantitative and interpretable measures can be obtained from image based analysis, permitting algorithmic differentiation of prostate Gleason grades. The method must be validated in a large independent series of specimens.

A fully automated approach to prostate biopsy segmentation based on level-set and mean filtering

With modern automated microscopes and digital cameras, pathologists no longer have to examine samples looking through microscope binoculars. Instead, the slide is digitized to an image, which can then be examined on a screen. This creates the possibility for computers to analyze the image. In this work, a fully automated approach to region of interest (ROI) segmentation in prostate biopsy images is proposed. This will allow the pathologists to focus on the most important areas of the image. The method proposed is based on level-set and mean filtering techniques for lumen centered expansion and cell density localization respectively. The novelty of the technique lies in the ability to detect complete ROIs, where a ROI is composed by the conjunction of three different structures, that is, lumen, cytoplasm, and cells, as well as regions with a high density of cells. The method is capable of dealing with full biopsies digitized at different magnifications. In this paper, results are shown with a set of 100 H and E slides, digitized at 5×, and ranging from 12 MB to 500 MB. The tests carried out show an average specificity above 99% across the board and average sensitivities of 95% and 80%, respectively, for the lumen centered expansion and cell density localization. The algorithms were also tested with images at 10× magnification (up to 1228 MB) obtaining similar results.

Histological Image Feature Mining Reveals Emergent Diagnostic Properties for Renal Cancer

Computer-aided histological image classification systems are important for making objective and timely cancer diagnostic decisions. These systems use combinations of image features that quantify a variety of image properties. Because researchers tend to validate their diagnostic systems on specific cancer endpoints, it is difficult to predict which image features will perform well given a new cancer endpoint. In this paper, we define a comprehensive set of common image features (consisting of 12 distinct feature subsets) that quantify a variety of image properties. We use a data-mining approach to determine which feature subsets and image properties emerge as part of an "optimal" diagnostic model when applied to specific cancer endpoints. Our goal is to assess the performance of such comprehensive image feature sets for application to a wide variety of diagnostic problems. We perform this study on 12 endpoints including 6 renal tumor subtype endpoints and 6 renal cancer grade endpoints. Keywords-histology, image mining, computer-aided diagnosis.

A resampling-based Markovian model for automated colon cancer diagnosis

In recent years, there has been a great effort in the research of implementing automated diagnostic systems for tissue images. One major challenge in this implementation is to design systems that are robust to image variations. In order to meet this challenge, it is important to learn the systems on a large number of labeled images from a different range of variation. However, acquiring labeled images is quite difficult in this domain, and hence, the labeled training data are typically very limited. Although the issue of having limited labeled data is acknowledged by many researchers, it has rarely been considered in the system design. This paper successfully addresses this issue, introducing a new resampling framework to simulate variations in tissue images. This framework generates multiple sequences from an image for its representation and models them using a Markov process. Working with colon tissue images, our experiments show that this framework increases the generalization capacity of a learner by increasing the size and variation of the training data and improves the classification performance of a given image by combining the decisions obtained on its sequences.

Computer-aided identification of prostatic adenocarcinoma: Segmentation of glandular structures

Background:

Identification of individual prostatic glandular structures is an important prerequisite to quantitative histological analysis of prostate cancer with the aid of a computer. We have developed a computer method to segment individual glandular units and to extract quantitative image features, for computer identification of prostatic adenocarcinoma.

Methods:

Two sets of digital histology images were used: database I (n = 57) for developing and testing the computer technique, and database II (n = 116) for independent validation. The segmentation technique was based on a k-means clustering and a region-growing method. Computer segmentation results were evaluated subjectively and also compared quantitatively against manual gland outlines, using the Jaccard similarity measure. Quantitative features that were extracted from the computer segmentation results include average gland size, spatial gland density, and average gland circularity. Linear discriminant analysis (LDA) was used to combine quantitative image features. Classification performance was evaluated with receiver operating characteristic (ROC) analysis and the area under the ROC curve (AUC).

Results:

Jaccard similarity coefficients between computer segmentation and manual outlines of individual glands were between 0.63 and 0.72 for non-cancer and between 0.48 and 0.54 for malignant glands, respectively, similar to an interobserver agreement of 0.79 for non-cancer and 0.75 for malignant glands, respectively. The AUC value for the features of average gland size and gland density combined via LDA was 0.91 for database I and 0.96 for database II.

Conclusions:

Using a computer, we are able to delineate individual prostatic glands automatically and identify prostatic adenocarcinoma accurately, based on the quantitative image features extracted from computer-segmented glandular structures.

Multimodal microscopy for automated histologic analysis of prostate cancer

Background

Prostate cancer is the single most prevalent cancer in US men whose gold standard of diagnosis is histologic assessment of biopsies. Manual assessment of stained tissue of all biopsies limits speed and accuracy in clinical practice and research of prostate cancer diagnosis. We sought to develop a fully-automated multimodal microscopy method to distinguish cancerous from non-cancerous tissue samples.

Methods

We recorded chemical data from an unstained tissue microarray (TMA) using Fourier transform infrared (FT-IR) spectroscopic imaging. Using pattern recognition, we identified epithelial cells without user input. We fused the cell type information with the corresponding stained images commonly used in clinical practice. Extracted morphological features, optimized by two-stage feature selection method using a minimum-redundancy-maximal-relevance (mRMR) criterion and sequential floating forward selection (SFFS), were applied to classify tissue samples as cancer or non-cancer.

Results

We achieved high accuracy (area under ROC curve (AUC) >0.97) in cross-validations on each of two data sets that were stained under different conditions. When the classifier was trained on one data set and tested on the other data set, an AUC value of ~0.95 was observed. In the absence of IR data, the performance of the same classification system dropped for both data sets and between data sets.

Conclusions

We were able to achieve very effective fusion of the information from two different images that provide very different types of data with different characteristics. The method is entirely transparent to a user and does not involve any adjustment or decision-making based on spectral data. By combining the IR and optical data, we achieved high accurate classification.

Histopathology tissue segmentation by combining fuzzy clustering with multiphase vector level sets

High resolution, multispectral, and multimodal imagery of tissue biopsies is an indispensable source of information for diagnosis and prognosis of diseases. Automatic extraction of relevant features from these imagery is a valuable assistance for medical experts. A primary step in computational histology is accurate image segmentation to detect the number and spatial distribution of cell nuclei in the tissue, along with segmenting other structures such as lumen and epithelial regions which together make up a gland structure. This chapter presents an automatic segmentation system for histopathology imaging. Spatial constraint fuzzy C-means provides an unsupervised initialization. An active contour algorithm that combines multispectral edge and region informations through a vector multiphase level set framework and Beltrami color metric tensors refines the segmentation. An improved iterative kernel filtering approach detects individual nuclei centers and decomposes densely clustered nuclei structures. The obtained results show high performances for nuclei detection compared to the human annotation.

Detection and classification of thyroid follicular lesions based on nuclear structure from histopathology images

Follicular lesions of the thyroid are traditionally difficult and tedious challenges in diagnostic surgical pathology in part due to lack of obvious discriminatory cytological and microarchitectural features. We describe a computerized method to detect and classify follicular adenoma of the thyroid, follicular carcinoma of the thyroid, and normal thyroid based on the nuclear chromatin distribution from digital images of tissue obtained by routine histological methods. Our method is based on determining whether a set of nuclei, obtained from histological images using automated image segmentation, is most similar to sets of nuclei obtained from normal or diseased tissues. This comparison is performed utilizing numerical features, a support vector machine, and a simple voting strategy. We also describe novel methods to identify unique and defining chromatin patterns pertaining to each class. Unlike previous attempts in detecting and classifying these thyroid lesions using computational imaging, our results show that our method can automatically classify the data pertaining to 10 different human cases with 100% accuracy after blind cross validation using at most 43 nuclei randomly selected from each patient. We conclude that nuclear structure alone contains enough information to automatically classify the normal thyroid, follicular carcinoma, and follicular adenoma, as long as groups of nuclei (instead of individual ones) are used. We also conclude that the distribution of nuclear size and chromatin concentration (how tightly packed it is) seem to be discriminating features between nuclei of follicular adenoma, follicular carcinoma, and normal thyroid.

A boosted Bayesian multiresolution classifier for prostate cancer detection from digitized needle biopsies

Diagnosis of prostate cancer (CaP) currently involves examining tissue samples for CaP presence and extent via a microscope, a time-consuming and subjective process. With the advent of digital pathology, computer-aided algorithms can now be applied to disease detection on digitized glass slides. The size of these digitized histology images (hundreds of millions of pixels) presents a formidable challenge for any computerized image analysis program. In this paper, we present a boosted Bayesian multiresolution (BBMR) system to identify regions of CaP on digital biopsy slides. Such a system would serve as an important preceding step to a Gleason grading algorithm, where the objective would be to score the invasiveness and severity of the disease. In the first step, our algorithm decomposes the whole-slide image into an image pyramid comprising multiple resolution levels. Regions identified as cancer via a Bayesian classifier at lower resolution levels are subsequently examined in greater detail at higher resolution levels, thereby allowing for rapid and efficient analysis of large images. At each resolution level, ten image features are chosen from a pool of over 900 first-order statistical, second-order co-occurrence, and Gabor filter features using an AdaBoost ensemble method. The BBMR scheme, operating on 100 images obtained from 58 patients, yielded: 1) areas under the receiver operating characteristic curve (AUC) of 0.84, 0.83, and 0.76, respectively, at the lowest, intermediate, and highest resolution levels and 2) an eightfold savings in terms of computational time compared to running the algorithm directly at full (highest) resolution. The BBMR model outperformed (in terms of AUC): 1) individual features (no ensemble) and 2) a random forest classifier ensemble obtained by bagging multiple decision tree classifiers. The apparent drop-off in AUC at higher image resolutions is due to lack of fine detail in the expert annotation of CaP and is not an artifact of the classifier. The implicit feature selection done via the AdaBoost component of the BBMR classifier reveals that different classes and types of image features become more relevant for discriminating between CaP and benign areas at different image resolutions.

High-throughput detection of prostate cancer in histological sections using probabilistic pairwise Markov models

In this paper we present a high-throughput system for detecting regions of carcinoma of the prostate (CaP) in HSs from radical prostatectomies (RPs) using probabilistic pairwise Markov models (PPMMs), a novel type of Markov random field (MRF). At diagnostic resolution a digitized HS can contain 80Kx70K pixels - far too many for current automated Gleason grading algorithms to process. However, grading can be separated into two distinct steps: (1) detecting cancerous regions and (2) then grading these regions. The detection step does not require diagnostic resolution and can be performed much more quickly. Thus, we introduce a CaP detection system capable of analyzing an entire digitized whole-mount HS (2x1.75cm(2)) in under three minutes (on a desktop computer) while achieving a CaP detection sensitivity and specificity of 0.87 and 0.90, respectively. We obtain this high-throughput by tailoring the system to analyze the HSs at low resolution (8microm per pixel). This motivates the following algorithm: (Step 1) glands are segmented, (Step 2) the segmented glands are classified as malignant or benign, and (Step 3) the malignant glands are consolidated into continuous regions. The classification of individual glands leverages two features: gland size and the tendency for proximate glands to share the same class. The latter feature describes a spatial dependency which we model using a Markov prior. Typically, Markov priors are expressed as the product of potential functions. Unfortunately, potential functions are mathematical abstractions, and constructing priors through their selection becomes an ad hoc procedure, resulting in simplistic models such as the Potts. Addressing this problem, we introduce PPMMs which formulate priors in terms of probability density functions, allowing the creation of more sophisticated models. To demonstrate the efficacy of our CaP detection system and assess the advantages of using a PPMM prior instead of the Potts, we alternately incorporate both priors into our algorithm and rigorously evaluate system performance, extracting statistics from over 6000 simulations run across 40 RP specimens. Perhaps the most indicative result is as follows: at a CaP sensitivity of 0.87 the accompanying false positive rates of the system when alternately employing the PPMM and Potts priors are 0.10 and 0.20, respectively.

COMPUTER-AIDED GLEASON GRADING OF PROSTATE CANCER HISTOPATHOLOGICAL IMAGES USING TEXTON FORESTS

The Gleason score is the single most important prognostic indicator for prostate cancer candidates and plays a significant role in treatment planning. Histopathological imaging of prostate tissue samples provides the gold standard for obtaining the Gleason score, but the manual assignment of Gleason grades is a labor-intensive and error-prone process. We have developed a texture classification system for automatic and reproducible Gleason grading. Our system characterizes the texture in images belonging to a tumor grade by clustering extracted filter responses at each pixel into textons (basic texture elements). We have used random forests to cluster the filter responses into textons followed by the spatial pyramid match kernel in conjunction with an SVM classifier. We have demonstrated the efficacy of our system in distinguishing between Gleason grades 3 and 4.