A computer simulation protocol to assess the accuracy of a Radio Stereometric Analysis (RSA) image processor according to the ISO-5725

Abstract

Radio-Sterometric-Analysis (RSA) and x-ray fluoroscopy require dedicated software to reconstruct the radiological scene and the position of the objects in space. It is important to have a reliable validation to correctly use these softwares. The two major regulations that deal with the definition of "accuracy" are the ISO-5725 and the GUM 1995. The aim of this work, is to present a protocol for the evaluation of the accuracy of a radio stereometric software in terms of "trueness" and "precision", according to the standard ISO-5725. The protocol consisted in a series of computer simulations of the radiological setup. Each simulation changed the position and orientation of the x-ray sources, detectors and objects. Then, radiological images were generated. The noise level of the images was also changed in order to evaluate the accuracy with different image qualities. Then, the images can be processed with RSA softwares to evaluate their accuracy. The protocol was tested on a custom RSA software developed at the Istituto Ortopedico Rizzoli and the accuracy of the results was evaluated. The radiological scene reconstruction accuracy was found of the order of (0.092 ± 0.14) mm for tube position and (0.38 ± 0.31) mm/(2.09 ± 1.39) deg for detectors in the direction other than the source-detector direction. In that case, in fact, the accuracy is of the order of (2.68 ± 3.08) mm for the tube position and (0.16 ± 0.27) mm/(0.75 ± 1.16) deg for detectors. This fact is an intrinsic limitation of the scene reconstruction technique and it is widely discussed in the literature. The model positioning and orientation evaluations was also very accurate: (0.22 ± 0.46) mm/(0.26 ± 0.22) deg. No differences were highlighted about the noise level. The accuracy remains the same independently from the the noise of the images. The protocol was also useful to detect and fix hidden bugs in the software and optimize the algorithms.

Assessment of paradoxical anterior translation in a CR total knee prosthesis coupling dynamic RSA and FE techniques

“…The validated dynamic RSA method allows to measure with sub-millimetric accuracy [3] (average 0.2 mm, SD ± 0.5 mm for the model position, and 0.3°± 0.2°for the model orientation), according to the ISO − 5725 regulation [30].…”

Over-constrained kinematic of the medial compartment leads to lower clinical outcomes after total knee arthroplasty

“…The inclusion criteria were: (1) age (50-85 years old); (2) severe radiographic osteoarthritis (Kellgren-Lawrence grade 3 and grade 4); (3) patients scheduled for a primary TKA. The exclusion criteria were: (1) previous corrective osteotomy on the afected lower limb; (2) post-traumatic arthritis; (3) severe preoperative varus-valgus deformity (Hip knee ankle angle > 10°); body mass index > 40 kg/m 2 ; (4) rheumatoid arthritis; (5) chronic inlammatory joint diseases; (6) patients with a prepathological abnormal gait (amputated, neuromuscular disorders, poliomyelitis, developmental dysplasia of the hip); (7) severe ankle osteoarthritis (Kellgren-Lawrence > 3); (8) severe hip osteoarthritis (Kellgren-Lawrence > 3); (9) previous total hip or ankle replacement; (10) unwillingness to take part in this study and providing Health Insurance Portability and Accountability Act (HIPAA) authorization; (11) incomplete clinical or kinematical assessment.…”

“…1b). The measurement accuracy of the validated dynamic RSA software is sub-millimetric (0.22 ± 0.46 mm and 0.26° ± 0.2° for the model position and orientation, respectively, according to the ISO-5725 regulation [14]), as evaluated in previous studies [1,2,6]. The operator's repeatability (test-retest reliability) was evaluated through repeated tests under diferent image noise conditions [14].…”

Medial Meniscal Posterior Horn Suturing Influences Tibial Internal-External Rotation in ACL-Reconstructed Knees

“…Each patient performed a series of single-leg squats in a radiographic room equipped with 2 x-ray sources placed so that the beamlines were perpendicular to each other and synchronized to acquire a pair of simultaneous radiographs (8 frames per second). The specifics of the radiographic setup were analogous to those used in previous studies 1,6,13,25 : detector dimensions were 43 Â 43 cm with a pixel matrix of 1440 Â 1440 pixels, and each beamline had the source-to-detector distance set to 180 cm. Each patient performed 3 repetition tasks.…”

“…The workflow had been validated in previous studies in terms of accuracy (0.22 ± 0.46 mm and 0.26 ± 0.2 for the model position and orientation, respectively) and test-retest reliability (mean error, <0.48 mm [95% CI, 0.15-0.80 mm]). 1,5,6,13,15,25 Kinematic data were normalized to the peak knee flexion angle and divided into a descendant phase (from the initial standing position to peak knee flexion) and an ascendant phase (from peak knee flexion to the final standing position). The internal-external and varus-valgus rotation, and the AP and medial-lateral translation were processed.…”