The invention relates to the technical field of medical
image processing and computer-
aided diagnosis, and discloses a
nasolacrimal duct obstructive
disease CT image-
aided diagnosis model construction method, which comprises the following steps of: firstly, performing three-dimensional reconstruction and standardized preprocessing on
orbit CT data; constructing a first-stage global diagnosis model, positioning the focus center of gravity and outputting a cause
classification result; intercepting a local region-of-interest
tensor by using the
barycentric coordinates, and constructing a second-level fine diagnosis model based on semi-
supervised learning; training the second-level model through the joint of
supervised learning of the
labeled data and the consistency constraint task of the unlabeled data, and outputting a smoothness grading result; and finally, an auxiliary diagnosis report and a visual image are generated in combination with the cause of
disease and the patency result. According to the method, a coarse-to-fine
cascade architecture and a semi-supervised strategy are adopted, so that the dependence on
labeled data is reduced, and the diagnosis precision and clinical
interpretability of the
nasolacrimal duct microstructure are effectively improved.