Artificial Intelligence for Diagnosis and Image Synthesis in Breast Cancer

The outline of this thesis is divided into three parts. Part 1 describes the influence and application of radiomics and artificial intelligence in breast imaging. Chapter 2 introduces the progress, application and challenge of radiomics and AI in breast cancer diagnosis in recent years, as well as the impact and significance of AI on future breast cancer research. In breast cancer research, most medical information, and particularly radiology results, are stored in free text format and the potential of this data remains untapped. Therefore, in Chapter 3, RadioLogical repOmics driven model incorporatinG medIcal token Cognition (RadioLOGIC) is proposed to extract repomics features from unstructured electronic health records, and to assess human health and predict pathological outcome via transfer learning. Accurately determining the molecular subtypes of breast cancer is important for the prognosis of breast cancer patients and can guide treatment selection. Therefore, in Chapter 4, we develop a multi-modal deep learning-based model for predicting the molecular subtypes of breast cancer.

Multi-parameter MRI information, which will be described in Part 2, has been shown to improve radiologist performance in lesion classification, as well as improving the performance of artificial intelligence models in various tasks. To overcome the less durable acquisition of full-spectrum MRI sequences, in Chapter 5 and Chapter 6, we develop different AI models to synthesize specific MRI images, including T1-weighted, T2-weighted and DCE-MRI, and compare the effectiveness of image synthesis with existing methods.

Part 3 provides a general discussion regarding the studies in this thesis.