Machine learning-based thoracic vertebrae detection and rib numbering of CT/MRI images

摘要

The present invention relates to a method for detecting the thoracic spine and specifying the order of the ribs in a CT/MRI image using machine learning. Specifically, the cervical spine / thoracic spine / automatically using machine learning to reduce discomfort and unnecessary waste of time for radiologists. The present invention relates to a method of detecting the thoracic spine in CT/MRI images and specifying the order of the ribs using machine learning that can help identify the number of ribs in the lesioned rib by automatically assigning the lumbar spine and automatically assigning the rib order. The present invention relates to a first step of receiving a CT / MRI image, a second step of performing object detection of a spine using an R-CNN model from the first step of the CT / MRI image, and a second step. The acquired spine image is continuously input into an image classification machine learning algorithm to specify the region of interest (ROI) of the spine image and classify it as a third grade, and in the third step, the spine is graded in three stages. The fourth step for discriminating the cervical vertebrae, thoracic, and lumbar vertebrae, the fifth step for designating the order of the ribs in consideration of grading from the highest image of the thoracic vertebrae identified in the fourth step, and the order of the spine and ribs in the CT/MRI image It provides a method for detecting thoracic vertebrae in a CT/MRI image using machine learning and specifying a rib sequence number using machine learning, characterized in that it includes a sixth step.