Dealing with Lack of Training Data for Convolutional Neural Networks: The Case of Digital Pathology

Francesco Ponzio,Gianvito Urgese,Santa Di Cataldo,Elisa Ficarra

doi:10.3390/electronics8030256

Francesco Ponzio, Gianvito Urgese + Show 2 more

Open Access

https://doi.org/10.3390/electronics8030256

Copy DOI

Journal: Electronics	Publication Date: Feb 26, 2019
Citations: 19	License type: CC BY 4.0

Affiliation: Polytechnic University of Turin

Abstract

Thanks to their capability to learn generalizable descriptors directly from images, deep Convolutional Neural Networks (CNNs) seem the ideal solution to most pattern recognition problems. On the other hand, to learn the image representation, CNNs need huge sets of annotated samples that are unfeasible in many every-day scenarios. This is the case, for example, of Computer-Aided Diagnosis (CAD) systems for digital pathology, where additional challenges are posed by the high variability of the cancerous tissue characteristics. In our experiments, state-of-the-art CNNs trained from scratch on histological images were less accurate and less robust to variability than a traditional machine learning framework, highlighting all the issues of fully training deep networks with limited data from real patients. To solve this problem, we designed and compared three transfer learning frameworks, leveraging CNNs pre-trained on non-medical images. This approach obtained very high accuracy, requiring much less computational resource for the training. Our findings demonstrate that transfer learning is a solution to the automated classification of histological samples and solves the problem of designing accurate and computationally-efficient CAD systems with limited training data.

Highlights

Histological image analysis is the gold standard for the primary diagnosis and assessment of c large number of cancers [1]
As a third transfer learning methodology, we extended the fine-tuning to all the blocks of the pre-trained VGG-16, using the weights learned on the ImageNet just for initialization
The purpose of the current study was to investigate the practical use of deep learning, and of convolutional neural networks, for the automatic classification of histopathological images, which is a task characterized by very high intra-class variability

Summary

Introduction

Histological image analysis is the gold standard for the primary diagnosis and assessment of c large number of cancers [1]. On the one hand, automated image analysis is a major improvement on human assessment, which has been majorly affected by inter- and intra-observer variability [3]. It is challenged by the size (in the order of gigapixels), as well as by the high complexity and variability of the histological images. The origin of such variability is three-fold: (i) “biological”, due to different cells (either cancerous or not) and corpuscles of variable appearance normally coexisting in a specimen;

Objectives

Methods

Results

Discussion

Conclusion