Sparse coding of pathology slides compared to transfer learning with deep neural networks

• Published in: BMC bioinformatics Vol. 19; no. Suppl 18; pp. 489 - 17
• Main Authors: Fischer, Will, Moudgalya, Sanketh S., Cohn, Judith D., Nguyen, Nga T. T., Kenyon, Garrett T.
• Format: Journal Article
• Language: English
• Published: London BioMed Central 21.12.2018; BioMed Central Ltd; Springer Nature B.V; BMC
• Subjects: Algorithms; Artificial intelligence; Artificial neural networks; BASIC BIOLOGICAL SCIENCES; Biochemistry & Molecular Biology; Bioinformatics; Biomedical and Life Sciences; Biotechnology & Applied Microbiology; Cancer; Cancer pathology slides; Care and treatment; Classification; Coding; Computational Biology/Bioinformatics; Computer Appl. in Life Sciences; Deep learning; Deep Learning - trends; Diagnosis; Dictionaries; Error reduction; Feature maps; Gene expression; Histopathology; Humans; Image classification; Image reconstruction; International conferences; Learning algorithms; Life Sciences; Locally Competitive Algorithm; Machine learning; Mathematical & Computational Biology; Medical diagnosis; Medical imaging; Microarrays; Neoplasms - pathology; Neural networks; Neural Networks (Computer); Pathological histology; Pathology; Phase transitions; Representations; Software; Sparse coding; State of the art; TCGA; Transfer learning; Tumors; Unsupervised learning; Locally Competitive Algorithm; Deep learning; Transfer learning; Cancer pathology slides; Sparse coding; TCGA; Unsupervised learning
• ISSN: 1471-2105; 1471-2105
• DOI: 10.1186/s12859-018-2504-8

Background Histopathology images of tumor biopsies present unique challenges for applying machine learning to the diagnosis and treatment of cancer. The pathology slides are high resolution, often exceeding 1GB, have non-uniform dimensions, and often contain multiple tissue slices of varying sizes surrounded by large empty regions. The locations of abnormal or cancerous cells, which may constitute a small portion of any given tissue sample, are not annotated. Cancer image datasets are also extremely imbalanced, with most slides being associated with relatively common cancers. Since deep representations trained on natural photographs are unlikely to be optimal for classifying pathology slide images, which have different spectral ranges and spatial structure, we here describe an approach for learning features and inferring representations of cancer pathology slides based on sparse coding. Results We show that conventional transfer learning using a state-of-the-art deep learning architecture pre-trained on ImageNet (RESNET) and fine tuned for a binary tumor/no-tumor classification task achieved between 85 % and 86 % accuracy. However, when all layers up to the last convolutional layer in RESNET are replaced with a single feature map inferred via a sparse coding using a dictionary optimized for sparse reconstruction of unlabeled pathology slides, classification performance improves to over 93 % , corresponding to a 54 % error reduction. Conclusions We conclude that a feature dictionary optimized for biomedical imagery may in general support better classification performance than does conventional transfer learning using a dictionary pre-trained on natural images.