When machine vision meets histology: A comparative evaluation of model architecture for classification of histology sections

Abstract

Classification of histology sections in large cohorts, in terms of distinct regions of microanatomy (e.g., stromal) and histopathology (e.g., tumor, necrosis), enables the quantification of tumor composition, and the construction of predictive models of genomics and clinical outcome. To tackle the large technical variations and biological heterogeneities, which are intrinsic in large cohorts, emerging systems utilize either prior knowledge from pathologists or unsupervised feature learning for invariant representation of the underlying properties in the data. However, to a large degree, the architecture for tissue histology classification remains unexplored and requires urgent systematical investigation. This paper is the first attempt to provide insights into three fundamental questions in tissue histology classification: I. Is unsupervised feature learning preferable to human engineered features? II. Does cellular saliency help? III. Does the sparse feature encoder contribute to recognition? We show that (a) in I, both Cellular Morphometric Feature and features from unsupervised feature learning lead to superior performance when compared to SIFT and [Color, Texture]; (b) in II, cellular saliency incorporation impairs the performance for systems built upon pixel-/patch-level features; and (c) in III, the effect of the sparse feature encoder is correlated with the robustness of features, and the performance can be consistently improved by the multi-stage extension of systems built upon both Cellular Morphmetric Feature and features from unsupervised feature learning. These insights are validated with two cohorts of Glioblastoma Multiforme (GBM) and Kidney Clear Cell Carcinoma (KIRC).

eDoctor: machine learning and the future of medicine

“…Alternatively, a database of images consisting of malignant and nonmalignant specimens could be presented to allow the computer program to determine how to best differentiate between the two categories and then judge the success of the program by presenting it with new slides to verify its accuracy. If successful, the process by which the computer reaches its conclusions is of less importance if it is better at predicting the final result than our current knowledge and rule‐based system .…”

“…This approach has been used in histological specimens and has even been crudely used in ‘reverse image searching’ by using images of specimens and utilizing the search engine to find similar images . Similarly, this process of deriving output variables from previous examples of known input variables allows for regression analysis, a concept commonly used in statistics with the difference being potential iterative improvement in accuracy of prediction by learning to progressively augment the prediction algorithm along with ability to deal with more variables and model complex nonlinear relationships between independent and dependent variables .…”

“…A common application of such a process is to explore complex interrelationships between genetics, biochemistry, histology and disease states within the data. Methods such as logistic regression, principal component analysis and latent class analysis have been used for many years; however, it is the integration with iterative supervised and unsupervised learning techniques that novel applications have emerged .…”

Unsupervised Transfer Learning via Multi-Scale Convolutional Sparse Coding for Biomedical Applications

“…We adopt SPM to construct tissue morphometric context [25], [28]–[31] as the final representation for tissue classification. Let V = [ v 1 , …, v T ] ∊ ℝ SKW×T be the feature set of T feature vectors with dimension SKW .…”

Deep Learning Nuclei Detection in Digitized Histology Images by Superpixels

“…In comparing multiple machine-learning strategies, it was found that the combination of supervised cellular morphology features and predictive sparse decomposition DL features provided the best separation of benign and malignant histology sections. [ 47 ] Wang et al . were able to detect mitosis in breast cancer histopathology images using the combined manually tuned cellular morphology data and convolutional neural net features.…”