Metrological evaluation of kyphosis in X-ray images: analysis of the Cobb angle and impact of operational variability

Kyphosis is a physiological curvature of the thoracic spine. This curvature assumes values between 20 and 40° Cobb angle in its normal state. The early detection of kyphosis is essential for an effective treatment, since the increase in this curvature can go on to cause a number of related problems. This article presents the evaluation of the effect of the operator factor on the Cobb angle values obtained in X-ray images with different degrees of kyphosis and the evaluation of the measurement uncertainty of this measurand by considering each measurement system used. Three operators, at three levels of experience, measured Cobb angle in nine images (three analogue, three digital and three computerized) of patients with different degrees of kyphosis. Measurements were carried out using a protractor for analogue images, whereas for digital and computerized images a specific software was used. The measurement uncertainty associated with angle was assessed by applying the GUM method. Operator 3 (less experienced) obtained more dispersed Cobb values, which resulted in higher uncertainty. Operator 1 (more experienced) showed better measurement result repeatability and quality. The variability of the readings was the variable that most influenced the final uncertainty of the measurements performed by Operator 3, while the measurement system resolution was that which had greater influence on operators 1 and 2. Operator experience significantly influences Cobb angle values in digitized X-ray images. Failure to consider the associated measurement uncertainty can lead to inconsistent diagnoses, especially in digital images, highlighting the importance of standardized methods and uncertainty evaluation to ensure diagnostic accuracy.