A Novel Multistage Transfer Learning for Ultrasound Breast Cancer Image Classification.

Gelan Ayana,Jin-Woo Jeong,Jinhyung Park,Se-Woon Choe

doi:10.3390/diagnostics12010135

Abstract

Breast cancer diagnosis is one of the many areas that has taken advantage of artificial intelligence to achieve better performance, despite the fact that the availability of a large medical image dataset remains a challenge. Transfer learning (TL) is a phenomenon that enables deep learning algorithms to overcome the issue of shortage of training data in constructing an efficient model by transferring knowledge from a given source task to a target task. However, in most cases, ImageNet (natural images) pre-trained models that do not include medical images, are utilized for transfer learning to medical images. Considering the utilization of microscopic cancer cell line images that can be acquired in large amount, we argue that learning from both natural and medical datasets improves performance in ultrasound breast cancer image classification. The proposed multistage transfer learning (MSTL) algorithm was implemented using three pre-trained models: EfficientNetB2, InceptionV3, and ResNet50 with three optimizers: Adam, Adagrad, and stochastic gradient de-scent (SGD). Dataset sizes of 20,400 cancer cell images, 200 ultrasound images from Mendeley and 400 ultrasound images from the MT-Small-Dataset were used. ResNet50-Adagrad-based MSTL achieved a test accuracy of 99 ± 0.612% on the Mendeley dataset and 98.7 ± 1.1% on the MT-Small-Dataset, averaging over 5-fold cross validation. A p-value of 0.01191 was achieved when comparing MSTL against ImageNet based TL for the Mendeley dataset. The result is a significant improvement in the performance of artificial intelligence methods for ultrasound breast cancer classification compared to state-of-the-art methods and could remarkably improve the early diagnosis of breast cancer in young women.

Highlights

Breast cancer is the most common cancer in women, with approximately 2 million new cases and 685,000 deaths worldwide every year [1]
Utilizing microscopic image data that are of sufficient quantity at the intermediate stage and employing multistage transfer learning (MSTL), we show that it is possible to achieve a performance better than conventional transfer learning (CTL) and state-of-the-art methods for US breast cancer diagnosis
The significance of this study is to show that with the use of MSTL via natural images, which are readily available, and microscopic images that can be acquired in large amounts, a high-performance Convolutional Neural Network (CNN) model can be developed

Summary

Introduction

Breast cancer is the most common cancer in women, with approximately 2 million new cases and 685,000 deaths worldwide every year [1]. Transfer learning (TL), which enables a model pre-trained on natural images (ImageNet) to be harnessed for segmentation, detection, or classification of US breast cancer images, has been applied to develop a relatively better performing deep learning model for breast cancer early diagnosis [11,12,13,14]. In [15], a transfer learning based breast ultrasound image classification deep learning method was proposed. In [19], the authors utilized a transfer learning method whereby an ImageNet pre-trained AlexNet network was used to classify breast ultrasound images. The authors were able to record improved performance using their proposed method These conventional transfer learning (CTL) methods still could not provide the desired performance for clinical applications due to their low test accuracy and the undesired false negatives which arose when the machines were subjected to previously unseen instances [15,20,21]

Methods

Results

Conclusion