Algorithm for processing X-ray images using fuzzy logic

Introduction. To improve the diagnostics of knee joint diseases, it is necessary to enhance the quality of processing radiographic images, i.e., to provide experts with more accurate information for pathology analysis. The objective of the study is to demonstrate the capabilities of fuzzy logic in improving the algorithm for determining reference lines and knee flexion angles. This requires a program that analyzes X-ray images. The methods known today, described in scientific and applied literature, are not sufficiently automated. In some cases, orthopedists and surgeons have to manually refine images and adjust lines. This gap is filled by the presented work. The algorithm developed by the author is described. It does not involve human participation and automatically identifies the lines and angles of knee flexion. Based on the result issued by the system, the doctor can, firstly, judge the presence of pathology. Secondly, the information provided by the program allows for more accurate planning, performing operations, and prescribing therapy.Materials and Methods. Images from two X-ray machines operating in Al-Basel Hospital (Latakia, Syria) were used. The Python language was used for the software implementation of the algorithm. The solution was tested on 500 patients at Al-Basel Hospital. The results generated by the new system and previous versions of X-ray image processing programs were compared.Results. An algorithm for constructing reference lines and angles for processing knee joint X-ray images is created, described, and implemented in practice. The capabilities of fuzzy logic in automating double threshold detection when identifying bone boundaries in images are shown. The operation of an improved Gaussian filter designed for processing X-ray images is described. The modified method of knee bone X-ray analysis includes the development of an algorithm for automatic detection of structures and anomalies in knee joints, determination and measurement of anatomical parameters, assessment of the degree of damage, etc. The method for determining the contour boundaries on radiographs combined the Canny detector, the watershed algorithm, and fuzzy logic. The program has been implemented in medical practice and shows 98% accuracy, spending less than 20 seconds to process the image.Discussion and Conclusion. The new system provides high accuracy, acceptable efficiency, and does not require manual correction of images. Experts are now able to identify subtle indicators of disorders. In addition, the new method makes it possible to understand complex cases when several factors are combined, indicating potential pathology. Widespread implementation of the method will improve the quality of medical services in orthopedics. Scientific research in this direction should be continued to expand the set of strategies for the treatment of diseases of the musculoskeletal system. It is necessary to create solutions with absolute accuracy, higher processing efficiency, as well as methods suitable for analyzing other joints.