Analysis and Improvement of Raimondi’s Table for Measurement of Axial Vertebral Rotation

Abstract

Axial vertebral rotation (AVR) is essential in studying idiopathic scoliosis. The Raimondi method is one of the techniques used to measure AVR on two-dimensional medical images. This method employs a set of data ordered in a table known as Raimondi’s table (RT) that was obtained experimentally. This research aims to analyze and eliminate the systematic errors in the data consigned in RT. We programmed a simplified software in C++ to automate the calculations. We applied a previously published equation and checked the values obtained against the original values of the RT to correct the AVR data. The error distribution in the set of 1530 values in the RT is Me (mean error) = 0.43 mm, SD (standard deviation) = 0.33 mm, Max (maximum value) = 2.11 mm, Q3 (third quartile) = 0.61 mm, median = 0.35 mm, Q1 (first quartile) = 0.18 mm, and Min (minimum value) = 0. The distribution of the error due to those errors is Me = 2.57º, SD = 3.8º, Max = 26º, Q3 = 2.53º, median = 1.18º, Q1 = 0.62º, and Min = 0º. The error in the RT values can produce a significant AVR measurement error (greater than 25% of the actual value) when the measured rotation is slight. The relative error decreases as the measured rotation becomes more outstanding. We can conclude that improving the data consigned in the RT allows us to measure the AVR on medical images of scoliosis with a lower bias error when using Raimondi’s method.