Automated classification of brain tumor type in whole-slide digital pathology images using local representative tiles

• Published in: Medical image analysis 2016-05, Vol.30, p.60-71
• Main Authors: Barker, Jocelyn, Hoogi, Assaf, Depeursinge, Adrien, Rubin, Daniel L.
• Format: Article
• Language: English
• Subjects: Algorithms; Biopsy - methods; Brain Neoplasms - diagnostic imaging; Brain Neoplasms - pathology; Computer aided diagnosis; Diagnosis, Differential; Digital pathology; Glioma - classification; Glioma - diagnostic imaging; Glioma - pathology; Humans; Image Enhancement - methods; Image Interpretation, Computer-Assisted - methods; Microscopy - methods; Object classification; Pathology - methods; Pattern Recognition, Automated - methods; Reproducibility of Results; Sensitivity and Specificity; Signal Processing, Computer-Assisted; Subtraction Technique
• ISSN: 1361-8415; 1361-8423
• DOI: 10.1016/j.media.2015.12.002

•Rough feature extraction and clustering is used to reveal the diversity of large data.•Independent evaluation of representative regions avoids signal dilution.•Signal based aggregation of data subsets gives strong predictions in noisy data.•A diverse feature set allows for better modeling of pathology images. Computerized analysis of digital pathology images offers the potential of improving clinical care (e.g. automated diagnosis) and catalyzing research (e.g. discovering disease subtypes). There are two key challenges thwarting computerized analysis of digital pathology images: first, whole slide pathology images are massive, making computerized analysis inefficient, and second, diverse tissue regions in whole slide images that are not directly relevant to the disease may mislead computerized diagnosis algorithms. We propose a method to overcome both of these challenges that utilizes a coarse-to-fine analysis of the localized characteristics in pathology images. An initial surveying stage analyzes the diversity of coarse regions in the whole slide image. This includes extraction of spatially localized features of shape, color and texture from tiled regions covering the slide. Dimensionality reduction of the features assesses the image diversity in the tiled regions and clustering creates representative groups. A second stage provides a detailed analysis of a single representative tile from each group. An Elastic Net classifier produces a diagnostic decision value for each representative tile. A weighted voting scheme aggregates the decision values from these tiles to obtain a diagnosis at the whole slide level. We evaluated our method by automatically classifying 302 brain cancer cases into two possible diagnoses (glioblastoma multiforme (N = 182) versus lower grade glioma (N = 120)) with an accuracy of 93.1 % (p