Research Of Automatic Diagnosis Algorithm For Knee Osteoarthritis Based On X-ray Image

Knee osteoarthritis is a very common degenerative joint disease in the middle-aged and elderly population.The clinical diagnosis is mainly based on observation of various medical imaging of the knee joint,various indicators are analyzed by medical imaging,and the diagnosis opinions are given in combination with the patient's condition statement.Common knee medical imaging methods include X-ray imaging,magnetic resonance imaging,CT imaging,and ultrasound imaging.X-ray imaging is a knee imaging method in clinical practice which is the most widely used.The main purpose of this paper is to study a complete algorithm to automatically obtain multiple diagnostic indicators of knee osteoarthritis from X-ray images and Complete automatic diagnosis of knee osteoarthritis.The X-ray images in the image set used in this paper are stored in DICOM format,so we firstly need to convert the DICOM image format into an image format that can be recognized by ordinary computers.This paper chooses to convert the DICOM format to BMP image format,and extract other relevant information in the DICOM file.After the image conversion is completed,the paper firstly divides the left and right leg imaging based on the grayscale projection by analyzing the image.Then the longitudinal segmentation of the knee joint interest region is carried out based on the gray projection algorithm and the relaxation principle.Then,based on an adaptive binarization algorithm,the lateral segmentation of the knee joint interest region is completed,and the knee joint interest region is locked.After obtaining the region of interest of the knee joint,this paper firstly analyzes the application effect of the traditional edge detection algorithm based on the improved gradient operator on the extraction algorithm of the knee joint edge,and then through the active shape model algorithm complete the contour extraction of the hard bone edge of the knee.Finally,based on the extraction results of contour edges,the automatic diagnosis of multiple knee arthritis indicators is realized,and compared with the results given by the image set,which proves that the whole algorithm has good adaptability and accuracy.