AI-Powered System for Automatic Determination of Abbreviated Breast Magnetic Resonance Imaging Protocol

Abstract

This invention relates to an artificial intelligence-assisted system and method for automatically determining the abbreviated imaging protocol to be used in breast magnetic resonance imaging (MRI) examinations. The technical problem that the invention aims to solve is that full breast MRI protocols have long imaging and interpretation times, and the sequences to be used in abbreviated protocols are often determined by expert experience or institution-specific preferences. Within the scope of this invention, candidate sequence combinations are automatically generated from the imaging sequences found in the full breast MRI protocol; the imaging time of each combination is calculated, and the relevant MRI images are analyzed using artificial intelligence-based machine learning and / or deep learning models to determine the classification performance. By evaluating the obtained classification performance and imaging time together, the sequence combination that preserves diagnostic information content at an acceptable level and can be performed in a shorter time is selected as the abbreviated breast MRI protocol. Thanks to this invention, the abbreviated breast MRI protocol is being determined in an objective, automated, measurable, and repeatable manner; it reduces imaging and interpretation times, enables more efficient use of equipment and personnel capacity, and allows for the creation of standardized protocols in different healthcare institutions.