Method and system for detecting pneumothorax

A pneumothorax detection method performed by a computing device, comprising:

• obtaining predicted pneumothorax information for an input image and a predicted spinal baseline for the input image, from a pneumothorax prediction model trained to predict pneumothorax by using training images tagged with left and right determination labels;
• and
• determining a pneumothorax presence region among left and right regions of a patient by using the predicted pneumothorax information and the predicted spinal baseline,
• wherein the left and right determination labels include region information separated based on the spinal baseline of each training image.

A pneumothorax detection method performed by a computing device operated by at least one processor, comprising:

• receiving training images;
• separating spinal regions from each training image and generating a spinal baseline representative of the spinal regions;
• tagging left and right determination labels based on the spinal baseline in each of the training images;
• generating each training image tagged with the left and right determination labels as training data;
• and
• training a pneumothorax prediction model by using the training data,
• wherein training the pneumothorax prediction model comprises
• training the pneumothorax prediction model for each of a task of predicting pneumothorax, a task of predicting tubes, and a task of predicting the spinal baseline, with respect to at least one training image corresponding to the training data.

A computing device comprising:

• a memory;
• and
• at least one processor that executes instructions of a program loaded in the memory,
• wherein the program comprises instructions for:
• inputting a requested image to a pneumothorax prediction model;
• obtaining predicted pneumothorax information, predicted tube information, and a predicted spinal baseline with respect to the requested image from the pneumothorax prediction model;
• separating a prediction image in which the predicted pneumothorax information and the predicted tube information are displayed into two regions, by using the predicted spinal baseline;
• and
• determining whether the requested image indicates an emergency by comparing regions in which the predicted pneumothorax information and the predicted tube information exist among the two regions.