SYSTEMS AND METHODS OF SEGMENTING MEDICAL IMAGES BASED ON ANATOMICAL LANDMARK-BASED FEATURES Russian patent published in 2019 - IPC A61B6/00 G06K9/46 G06K9/62 

FIELD: medicine.

SUBSTANCE: group of inventions relates to segmentation of medical images, namely to systems and methods for automated segmentation of medical images based on training algorithms using features derived from anatomical landmarks. Method for segmenting medical images implemented on a computer comprises steps of obtaining an image from memory, identifying a reference point, selecting an image point, determining a feature for an image point relative to a reference point and establishing a connection of the image point with the anatomical structure. Method of segmenting medical images implemented on a computer includes steps of extracting an image portion and connecting an image point to an anatomical structure by using a classification model based on the determined feature and the extracted image portion. Non-temporal machine-readable data medium stores computer-executable instructions which, when executed by a computer, cause the computer to perform a training operation on a classification model used for segmentation of medical images. Non-temporal machine-readable data medium stores computer-executable instructions which, when executed by a computer, induce a computer to perform a computer-aided method of segmenting medical images using a classification model. System for training the classification model used for segmentation of medical images comprises a database and an image processor connected to the database in order to access a plurality of training images. Image processor is configured to perform a method of segmenting medical images implemented on a computer. System for segmenting medical images using a classification model comprises an image processor configured to perform a method of segmentation of medical images implemented on a computer.

EFFECT: using the group of inventions enables improving segmentation efficiency on medical images in radiation therapy or in related areas.

20 cl, 9 dwg