Deep learning based digital cell profiles for risk stratification of urine cytology images.

Urine cytology is a test for the detection of high-grade bladder cancer. In clinical practice, the pathologist would manually scan the sample under the microscope to locate atypical and malignant cells. They would assess the morphology of these cells to make a diagnosis. Accurate identification of atypical and malignant cells in urine cytology is a challenging task and is an essential part of identifying different diagnosis with low-risk and high-risk malignancy. Computer-assisted identification of malignancy in urine cytology can be complementary to the clinicians for treatment management and in providing advice for carrying out further tests. In this study, we presented a method for identifying atypical and malignant cells followed by their profiling to predict the risk of diagnosis automatically. For cell detection and classification, we employed two different deep learning-based approaches. Based on the best performing network predictions at the cell level, we identified low-risk and high-risk cases using the count of atypical cells and the total count of atypical and malignant cells. The area under the receiver operating characteristic (ROC) curve shows that a total count of atypical and malignant cells is comparably better at diagnosis as compared to the count of malignant cells only. We obtained area under the ROC curve with the count of malignant cells and the total count of atypical and malignant cells as 0.81 and 0.83, respectively. Our experiments also demonstrate that the digital risk could be a better predictor of the final histopathology-based diagnosis. We also analyzed the variability in annotations at both cell and whole slide image level and also explored the possible inherent rationales behind this variability.

Diagnostic Assessment of Deep Learning Algorithms for Detection of Lymph Node Metastases in Women With Breast Cancer.

Importance: Application of deep learning algorithms to whole-slide pathology images can potentially improve diagnostic accuracy and efficiency. Objective: Assess the performance of automated deep learning algorithms at detecting metastases in hematoxylin and eosin-stained tissue sections of lymph nodes of women with breast cancer and compare it with pathologists' diagnoses in a diagnostic setting. Design, Setting, and Participants: Researcher challenge competition (CAMELYON16) to develop automated solutions for detecting lymph node metastases (November 2015-November 2016). A training data set of whole-slide images from 2 centers in the Netherlands with (n = 110) and without (n = 160) nodal metastases verified by immunohistochemical staining were provided to challenge participants to build algorithms. Algorithm performance was evaluated in an independent test set of 129 whole-slide images (49 with and 80 without metastases). The same test set of corresponding glass slides was also evaluated by a panel of 11 pathologists with time constraint (WTC) from the Netherlands to ascertain likelihood of nodal metastases for each slide in a flexible 2-hour session, simulating routine pathology workflow, and by 1 pathologist without time constraint (WOTC). Exposures: Deep learning algorithms submitted as part of a challenge competition or pathologist interpretation. Main Outcomes and Measures: The presence of specific metastatic foci and the absence vs presence of lymph node metastasis in a slide or image using receiver operating characteristic curve analysis. The 11 pathologists participating in the simulation exercise rated their diagnostic confidence as definitely normal, probably normal, equivocal, probably tumor, or definitely tumor. Results: The area under the receiver operating characteristic curve (AUC) for the algorithms ranged from 0.556 to 0.994. The top-performing algorithm achieved a lesion-level, true-positive fraction comparable with that of the pathologist WOTC (72.4% [95% CI, 64.3%-80.4%]) at a mean of 0.0125 false-positives per normal whole-slide image. For the whole-slide image classification task, the best algorithm (AUC, 0.994 [95% CI, 0.983-0.999]) performed significantly better than the pathologists WTC in a diagnostic simulation (mean AUC, 0.810 [range, 0.738-0.884]; P < .001). The top 5 algorithms had a mean AUC that was comparable with the pathologist interpreting the slides in the absence of time constraints (mean AUC, 0.960 [range, 0.923-0.994] for the top 5 algorithms vs 0.966 [95% CI, 0.927-0.998] for the pathologist WOTC). Conclusions and Relevance: In the setting of a challenge competition, some deep learning algorithms achieved better diagnostic performance than a panel of 11 pathologists participating in a simulation exercise designed to mimic routine pathology workflow; algorithm performance was comparable with an expert pathologist interpreting whole-slide images without time constraints. Whether this approach has clinical utility will require evaluation in a clinical setting.

Deep feature based cross-slide registration.

Registration of multiple sections in a tissue block is an important pre-requisite task before any cross-slide image analysis. Non-rigid registration methods are capable of finding correspondence by locally transforming a moving image. These methods often rely on an initial guess to roughly align an image pair linearly and globally. This is essential to prevent convergence to a non-optimal minimum. We explore a deep feature based registration (DFBR) method which utilises data-driven descriptors to estimate the global transformation. A multi-stage strategy is adopted for improving the quality of registration. A visualisation tool is developed to view registered pairs of WSIs at different magnifications. With the help of this tool, one can apply a transformation on the fly without the need to generate a transformed moving WSI in a pyramidal form. We compare the performance on our dataset of data-driven descriptors with that of hand-crafted descriptors. Our approach can align the images with only small registration errors. The efficacy of our proposed method is evaluated for a subsequent non-rigid registration step. To this end, the first two steps of the ANHIR winner's framework are replaced with DFBR to register image pairs provided by the challenge. The modified framework produce comparable results to those of the challenge winning team.

Context-Aware Convolutional Neural Network for Grading of Colorectal Cancer Histology Images.

Digital histology images are amenable to the application of convolutional neural networks (CNNs) for analysis due to the sheer size of pixel data present in them. CNNs are generally used for representation learning from small image patches (e.g. 224×224 ) extracted from digital histology images due to computational and memory constraints. However, this approach does not incorporate high-resolution contextual information in histology images. We propose a novel way to incorporate a larger context by a context-aware neural network based on images with a dimension of 1792×1792 pixels. The proposed framework first encodes the local representation of a histology image into high dimensional features then aggregates the features by considering their spatial organization to make a final prediction. We evaluated the proposed method on two colorectal cancer datasets for the task of cancer grading. Our method outperformed the traditional patch-based approaches, problem-specific methods, and existing context-based methods. We also presented a comprehensive analysis of different variants of the proposed method.

Using spectral imaging for the analysis of abnormalities for colorectal cancer: When is it helpful?

The spectral imaging technique has been shown to provide more discriminative information than the RGB images and has been proposed for a range of problems. There are many studies demonstrating its potential for the analysis of histopathology images for abnormality detection but there have been discrepancies among previous studies as well. Many multispectral based methods have been proposed for histopathology images but the significance of the use of whole multispectral cube versus a subset of bands or a single band is still arguable. We performed comprehensive analysis using individual bands and different subsets of bands to determine the effectiveness of spectral information for determining the anomaly in colorectal images. Our multispectral colorectal dataset consists of four classes, each represented by infra-red spectrum bands in addition to the visual spectrum bands. We performed our analysis of spectral imaging by stratifying the abnormalities using both spatial and spectral information. For our experiments, we used a combination of texture descriptors with an ensemble classification approach that performed best on our dataset. We applied our method to another dataset and got comparable results with those obtained using the state-of-the-art method and convolutional neural network based method. Moreover, we explored the relationship of the number of bands with the problem complexity and found that higher number of bands is required for a complex task to achieve improved performance. Our results demonstrate a synergy between infra-red and visual spectrum by improving the classification accuracy (by 6%) on incorporating the infra-red representation. We also highlight the importance of how the dataset should be divided into training and testing set for evaluating the histopathology image-based approaches, which has not been considered in previous studies on multispectral histopathology images.

Glandular Morphometrics for Objective Grading of Colorectal Adenocarcinoma Histology Images.

Determining the grade of colon cancer from tissue slides is a routine part of the pathological analysis. In the case of colorectal adenocarcinoma (CRA), grading is partly determined by morphology and degree of formation of glandular structures. Achieving consistency between pathologists is difficult due to the subjective nature of grading assessment. An objective grading using computer algorithms will be more consistent, and will be able to analyse images in more detail. In this paper, we measure the shape of glands with a novel metric that we call the Best Alignment Metric (BAM). We show a strong correlation between a novel measure of glandular shape and grade of the tumour. We used shape specific parameters to perform a two-class classification of images into normal or cancerous tissue and a three-class classification into normal, low grade cancer, and high grade cancer. The task of detecting gland boundaries, which is a prerequisite of shape-based analysis, was carried out using a deep convolutional neural network designed for segmentation of glandular structures. A support vector machine (SVM) classifier was trained using shape features derived from BAM. Through cross-validation, we achieved an accuracy of 97% for the two-class and 91% for three-class classification.

Spatial and spatio-temporal feature extraction from 4D echocardiography images.

BACKGROUND: Ultrasound images are difficult to segment because of their noisy and low contrast nature which makes it challenging to extract the important features. Typical intensity-gradient based approaches are not suitable for these low contrast images while it has been shown that the local phase based technique provides better results than intensity based methods for ultrasound images. The spatial feature extraction methods ignore the continuity in the heart cycle and may also capture spurious features. It is believed that the spurious features (noise) that are not consistent along the frames can be excluded by considering the temporal information. METHODS: In this paper, we present a local phase based 4D (3D+time) feature asymmetry (FA) measure using the monogenic signal. We have investigated the spatio-temporal feature extraction to explore the effect of adding time information in the feature extraction process. RESULTS: To evaluate the impact of time dimension, the results of 4D based feature extraction are compared with the results of 3D based feature extraction which shows the favorable 4D feature extraction results when temporal resolution is good. The paper compares the band-pass filters (difference of Gaussian, Cauchy and Gaussian derivative) in terms of their feature extraction performance. Moreover, the feature extraction is further evaluated quantitatively by left ventricle segmentation using the extracted features. CONCLUSIONS: The results demonstrate that the spatio-temporal feature extraction is promising in frames with good temporal resolution.