Keywords: Self-supervised learning, pretext task, histopathology, lung adenocarcinoma
TL;DR: A novel pretext task for pathology image classification based on the multi-resolution nature of whole slide images.
Abstract: In computational pathology, fully-supervised convolutional neural networks have been shown to perform well on tasks such as histology segmentation and classification but require large amounts of expert-annotated labels. In this work, we propose a self-supervised learning pretext task that utilizes the multi-resolution nature of whole slide images to reduce labeling effort. Given a pair of image tiles cropped at different magnification levels, our model predicts whether one tile is contained in the other. We hypothesize that this task induces the model to learn to distinguish different structures presented in the images, thus benefiting the downstream classification. The potential of our method was shown in downstream classification of lung adenocarcinoma histologic subtypes using H\&E-images from the National Lung Screening Trial.
Registration: I acknowledge that acceptance of this work at MIDL requires at least one of the authors to register and present the work during the conference.
Authorship: I confirm that I am the author of this work and that it has not been submitted to another publication before.
Paper Type: novel methodological ideas without extensive validation
Primary Subject Area: Learning with Noisy Labels and Limited Data
Secondary Subject Area: Application: Histopathology
Confidentiality And Author Instructions: I read the call for papers and author instructions. I acknowledge that exceeding the page limit and/or altering the latex template can result in desk rejection.