Lung disease recognition and detection method based on chest CT image

The invention discloses a lung disease identification and detection method based on a chest CT image, and the method comprises the following steps: 1) medical image preprocessing: converting the gray value of an original CT image into an HU value, selecting a window width and a window level, eliminating the sampling difference through an interpolation method, adding a Laplacian convolution layer to sharpen the image, and carrying out the preprocessing of a medical image; outputting a clear chest CT image for the chest CT image lung disease identification and detection method; 2) establishing a chest CT image lung disease identification and detection model: a, adding a transposed convolution layer in a backbone network of Faster-RCNN, and performing information restoration by using transposed convolution; b, providing a transposed convolutional network layer based on an Inception structure, wherein the transposed convolutional network layer is used for solving the information loss condition of a detected objec