SAT0069 The Validation about The Measurement of Joint Space Distance by Using Super Resolution Image and Curve Fitting Methods

BackgroundCartilage loss is irreversible progress, so physicians have to find the detection of joint space narrowing as early as possible. It is stated that physical disability in rheumatoid arthritis (RA) is associated with joint space narrowing rather than erosion [1], so early detection of progressive joint damage is the most important. But in DICOM which is the international standard for medical images, 1 pixel on radiographic images is 0.10 to 0.15 mm, so the detection of the change which is lower than that is impossible.To solve this problem, we previously developed new method measuring the joint space distance with the X-rays image which performed super-resolution image processing [2]. However, we did not validate the measured value by our methods.ObjectivesTo validate the measurement of joint space distance by using super resolution image and curve fitting methods. We also evaluated whether the focus-film distance affected the measured value.MethodsWe producted phantom model (PM) of joint space that expressed metacarpophalangeal joint space 1mm by using 3D printer.The super-resolution images of PM with radiographs by using total variation regularization and shock filter were prepared (Fig. 1). We measured the distance of joint space by three methods, 1) conventional measurement; distance between two points, 2) measurement by using normal vector from the top of joint surface of phalanx bone, 3) measurement using joint space area calculated by integration method. Each method processed on MATLAB®. 2), 3) methods were processed by curve fitting method which create parabolic curve by choosing some points of joint surface.We measured the distance ten times to adjust the error of measurement.We took radiographs with the focus-film distance under conditions of 90cm, 100cm, 110 cm, respectively.ResultsBetween the two points method, normal line method and integration method, the mean (standard deviation) of measured value under focus-film distance 90 cm was 1.033 (0.018), 0.964 (0.017), 0.939 (0.009), respectively. Under focus-film distance 100 cm, it was 0.963 (0.031), 0.969 (0.032) ,0.943 (0.025), respectively. Under focus-film distance 110 cm, it was 1.019 (0.026), 0.964 (0.018), 0.943 (0.015) , respectively. There were no significant differences with measured value between three methods under each focus-film distance. There were also no significant differences with each focus-film distance (Fig. 2).ConclusionsThe measured value was not affected by a diff