The application relates to the field of
artificial intelligence medical technology and discloses an early
esophageal cancer screening method and
system based on
artificial intelligence, which comprises the following steps: preparing an obtained esophageal epithelial
cell sample of a subject into a single-layer cytological smear, generating an original
digital pathology image through
optical scanning; extracting independent cells therefrom, calculating local spatial moment parameters of the independent cells; determining the geometric principal axis direction and stretching degree of the independent cells based on the local spatial moment parameters; judging whether the stretching degree exceeds the length-
width ratio benchmark constant of normal esophageal squamous epithelial cells, if yes, applying a local two-dimensional affine transformation to the independent cells along the geometric principal axis direction to offset the
mechanical stretching deformation caused by sampling, and generating calibrated
cell image data; inputting the calibrated
cell image data into an
artificial intelligence classification module, extracting cell
pathological features based on a deep
convolutional neural network, and outputting a screening diagnosis result. The application offsets
cell deformation interference and improves screening accuracy.