The invention discloses a
cardiac magnetic resonance cardiac function evaluation method based on multi-
task learning, and belongs to the technical field of medical
image processing. The method comprises the following steps: labeling, enhancing and preprocessing
cardiac magnetic resonance image data; training a multi-
task learning model by using the preprocessed data, supervising a multi-task training process by introducing regional
geometric consistency loss, and improving the accuracy and generalization performance of the model; inputting a to-be-evaluated heart magnetic
resonance image into the trained multi-
task learning model to obtain left and right ventricular intima segmentation results in a heart anatomical structure, a left ventricular myocardial segmentation result and a right ventricular
insertion point and a left ventricular central point in a heart key point position; further calculating the
ventricular volume to obtain left and right ventricular ejection fractions; calculating the thickness of each segment of the
left ventricular myocardium; and outputting a cardiac function
evaluation result. According to the method, the accuracy and generalization performance of the model are improved, and the problems that a traditional method is low in precision and long in analysis time consumption are solved.