Grant: Alliance Grants

Breast cancer is the second most common cancer in women and will affect 1 in 9 Canadian women in their lifetime. On average, 78 Canadian women will be diagnosed with breast cancer every day and 15 will die from it. Mammographic screening allows for early management and a better prognosis. However, it has been shown in Quebec that in 43.7% of screening mammography examinations, the positioning of the breasts is inadequate, requiring a repeated examination. This results in additional radiation exposure for the patient, anxiety due to delayed results and an increased workload for technologists and radiologists. The goal of this project is to develop and validate a first automatic system for assessing the quality of breast positioning at the time of the mammography examination, which would objectively indicate to the technologists whether it is necessary to repeat the examination before the patient leaves the clinic. The system would indicate whether the examination is of sufficient quality and if not, would provide specific explanations as to which criteria were not met. To do this, we will implement novel deep learning approaches for image and text analysis, and leverage unique databases provided by our partners and collaborators. Such a solution, integrated in mammography acquisition platforms, would improve performance in terms of breast cancer screening.