Transfer Learning in Breast Cancer Diagnoses via Ultrasound Imaging

SIMPLE SUMMARY: Transfer learning plays a major role in medical image analyses; however, obtaining adequate training image datasets for machine learning algorithms can be challenging. Although many studies have attempted to employ transfer learning in medical image analyses, thus far, only a few review articles regarding the application of transfer learning to medical image analyses have been published. Moreover, reviews on the application of transfer learning in ultrasound breast imaging are rare. This work reviews previous studies that focused on detecting breast cancer from ultrasound images by using transfer learning, in order to summarize existing methods and identify their advantages and shortcomings. Additionally, this review presents potential future research directions for applying transfer learning in ultrasound imaging for the purposes of breast cancer detection and diagnoses. This review is expected to be significantly helpful in guiding researchers to identify potential improved methods and areas that can be improved through further research on transfer learning-based ultrasound breast imaging. ABSTRACT: Transfer learning is a machine learning approach that reuses a learning method developed for a task as the starting point for a model on a target task. The goal of transfer learning is to improve performance of target learners by transferring the knowledge contained in other (but related) source domains. As a result, the need for large numbers of target-domain data is lowered for constructing target learners. Due to this immense property, transfer learning techniques are frequently used in ultrasound breast cancer image analyses. In this review, we focus on transfer learning methods applied on ultrasound breast image classification and detection from the perspective of transfer learning approaches, pre-processing, pre-training models, and convolutional neural network (CNN) models. Finally, comparison of different works is carried out, and challenges—as well as outlooks—are discussed.