A computer simulation protocol to assess the accuracy of a Radio Stereometric Analysis (RSA) image processor according to the ISO-5725
Authors:
M. Bontempi,
U. Cardinale,
L. Bragonzoni,
G. M. Marcheggiani Muccioli,
D. Alesi,
B. di Matteo,
M. Marcacci,
S. Zaffagnini
Abstract:
Radio-Stereometric-Analysis and x-ray fluoroscopy are radiological techniques that require dedicated software to process data. The accurate calibration of these software is therefore critical. The aim of this work is to produce a protocol for evaluating the softwares' accuracy according to the ISO-5725. A series of computer simulations of the radiological setup and images were employed. The noise…
▽ More
Radio-Stereometric-Analysis and x-ray fluoroscopy are radiological techniques that require dedicated software to process data. The accurate calibration of these software is therefore critical. The aim of this work is to produce a protocol for evaluating the softwares' accuracy according to the ISO-5725. A series of computer simulations of the radiological setup and images were employed. The noise level of the images was also changed to evaluate the accuracy with different image qualities. The protocol was tested on a custom software developed by the authors. Radiological scene reconstruction accuracy was of (0.092 +- 0.14) mm for tube position, and (0.38 +- 0.31) mm / (2.09 +- 1.39) deg for detectors oriented in a direction other than the source-detector direction. In the source-detector direction the accuracy was of (2.68 +- 3.08) mm for tube position, and of (0.16 +- 0.27) mm / (0.075 +- 1.16) deg for the detectors. These disparate results are widely discussed in the literature. Model positioning and orientation was also highly accurate: (0.22 +- 0.46) mm / (0.26 +- 0.22) deg. Accuracy was not affected by the noise level. The protocol was able to assess the accuracy of the RSA system. It was also useful to detect and fix hidden bugs. It was also useful to detect and resolve hidden bugs in the software, and in optimizing the algorithms.
△ Less
Submitted 17 June, 2022; v1 submitted 6 June, 2020;
originally announced June 2020.
