The invention relates to a prediction model for non-invasive quantitative evaluation of postoperative concurrent pancreatic
fistula risk before pancreatic resection, and belongs to the technical field of
medical imaging. According to the invention, a
real image obtained in clinical practice is put into an
artificial intelligence end, and ultrasonic multi-mode and preoperative various pancreatic
fistula related indexes are combined; the method also comprises deeply mining image
omics characteristic information in an image by using an AI
algorithm, comprehensively evaluating, screening, weighting and comprehensively calculating the image
omics characteristic information, extracting useful information, and establishing a diagnosis and treatment model for
artificial intelligence prediction of postoperative pancreatic
fistula in combination with various preoperative characteristics of a patient and related
laboratory examination and image examination results. Starting from ultrasonic elastic quantification, the problem that the
pancreas texture cannot be objectively evaluated before a
pancreas resection operation in the past is solved, and non-invasive quantification is conducted on the
pancreas texture; meanwhile, an
artificial intelligence machine learning method is introduced, the problem that a large amount of information cannot be processed in real time during manual
image analysis is solved, and the effectiveness and accuracy of a prediction model are maximized.