A comprehensive computational framework for facilitating the radiological and surgical management of breast cancer

点击翻译按钮获取中文摘要

Diagnosis and treatment have always been the two defining pillars of medicine, with the fidelity of the former being essential to the success of the latter. In particular, in cancer research, diagnosing the disease at its early stages is often associated with better prognosis and prolonged survival. For this reason, considerable efforts have been extended over the last few decades to develop accurate means of early diagnosis of oncological disorders. Among such methods and tools, medical imaging has evolved into a key component of modern health care upon which diagnosis of solid cancers is currently based. That being the case, the main focus of the present research program is to develop advanced, problem-tailored methods of medical image processing that enable clinicians to manage oncological disorders in a more effective way. The fundamental idea of the program stems from the fact that, from a clinical perspective, diagnosis and treatment of solid cancers are two intimately interwoven processes, which involve mutual decision making supported by the same objective information, such as that provided by medical imaging. Consequently, instead of devising separate solutions catering to the needs of either radiology or therapeutics, this program aims at developing a joint computational and visualization platform, which can be shared by both sides to facilitate their interaction towards improved health outcomes.******As a particular application, the present program focuses on breast cancer, which remains the most frequently diagnosed malignancy among Canadian women. Presently, both radiological and surgical management of breast cancer depend on several imaging modalities, among which are X-ray mammography and Magnetic Resonance Imaging (MRI). Based on different physical phenomena, however, the modalities produce images of a distinct and often supplementary nature. For this reason, several clinical applications require the content of such images to be properly “fused”. Such fusion is known to be a difficult problem, effective solutions to which are yet to be found. Moreover, the sublime sensitivity of breast MRI is still counterpoised by its relatively low specificity, which often results in redundant breast biopsies, with their associated phycological and economic burdens. To address this problem calls for the use of cutting-edge imaging technologies, such as diffusion MRI (dMRI). However, having demonstrated a proven ability to discriminate between benign and malignant tumours, dMRI still suffers from a number of artifacts which need to be resolved through further research. Finally, advanced imaging is still to be brought to the operating room, where it is much needed to improve the success of breast surgery through more accurate presurgical planning. A practical realization of this idea based on recent advances in visualization technologies constitutes another goal of this research program.