Method of classifying lesion of chest x-ray radiograph based on data normalization and local patch and apparatus thereof

A method of classifying a lesion of a chest x-ray radiograph, the method comprising:

• segmenting an input chest x-ray radiograph into an organ area by using a first neural network based on a pre-learned segmentation model;
• generating local patches for the segmented organ area;
• and
• classifying a lesion in the input chest x-ray radiograph by using a second neural network based on a pre-learned classification model for the generated local patches.

An apparatus for classifying a lesion of a chest x-ray radiograph, the apparatus comprising:

• a normalization unit configured to convert an input chest x-ray radiograph into a normalized image;
• a segmentation unit configured to segment the converted normalized image into an organ area by using a first neural network based on a pre-learned segmentation model;
• a generation unit configured to generate local patches for the segmented organ area;
• and
• a classification unit configured to classify a lesion in the input chest x-ray radiograph by using a second neural network based on a pre-learned classification model for the generated local patches.

A method of classifying a lesion of a chest x-ray radiograph, the method comprising:

• converting an input chest x-ray radiograph into a normalized image;
• segmenting the converted normalized image into an organ area by using a first neural network based on a pre-learned segmentation model;
• generating local patches for the segmented organ area;
• and
• classifying a lesion in the input chest x-ray radiograph by using a second neural network based on a pre-learned classification model for the generated local patches and classifying a most classified lesion among lesion classification results for each of the generated local patches as a final lesion.

A method of quantifying lesion severity based on a vision transformer, the method comprising:

• receiving an input chest x-ray radiograph;
• extracting a feature map from the received input chest x-ray radiograph by using a pre-learned neural network;
• classifying a lesion from the input chest x-ray radiograph by using the vision transformer based on the extracted feature map;
• and
• quantifying severity of the lesion based on the extracted feature map and the classified lesion.