The present invention performs a first convolution process and max pooling on an input X-ray image, performs a second convolution process and max pooling on max-pooled first channel data, and performs a max-pooled second channel A third convolution process and max pooling are performed on data, a fourth convolution process and max pooling are performed on the max pooled third channel data, and X-rays are based on the max pooled fourth channel data. It relates to a CNN model prediction method and a CNN model apparatus for predicting a disease on an image, predicting a disease on an X-ray image.
展开▼