Investigation of Training Multiple Instance Learning with Instance Sampling

Aliasghar Tarkhan, Trung Kien Nguyen, Noah Simon, Jian Dai

Keywords: Attention, computational pathology, deep learning, multiple instance learning, prostate cancer, sampling, transfer learning, weekly supervised learning

TL;DR: Training Multiple Instance Learning with Instance Sampling

Abstract: One challenge of training deep neural networks with gigapixel whole-slide images (WSIs) in computational pathology is the lack of annotation at pixel level or region (instance) level due to the high cost and time-consuming labeling effort. Multiple instance learning (MIL) as a typical weakly supervised learning method aimed to resolve this challenge by using only the slide-level label without the need for pixel or region labels. Not all instances are predictive of the outcome. The attention-based MIL method leverages this fact to enhance the performance by weighting the instances based on their contribution in predicting the outcome. A WSI typically contains hundreds of thousands of image regions. Training a deep neural network with thousands of image regions (patches) per slide is computationally expensive, and it needs a lot of time for convergence. One way to alleviate this issue is to sample a subset of instances from the available instances within each bag for training. While the benefit of sampling strategies for decreasing computing time might be evident, there is a lack of effort to investigate their performances. This paper investigates different sampling strategies from both computing time and performance points of view. We empirically show how these sampling strategies substantially reduce computation time. Moreover, random sampling can even improve performance if we carefully choose the number of selected instances.

Registration: I acknowledge that publication of this at MIDL and in the proceedings 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: both

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.