Avaliação dos fatores que afetam a análise de imagens de raio-X para diagnóstico de cifose, escoliose e osteoporose

Abstract

The present study aims to evaluate the effect of the factor Operator on Cobb angle values obtained on X-ray images of patients who were diagnosed with scoliosis, kyphosis and osteoporosis, as well as to propose procedures to evaluate the measurement uncertainty of this measurand taking into account each measuring system used. For this, the Cobb angle was measured by three Operators in analog, digital and computerized images in the diagnosis of scoliosis and kyphosis. Two measurement systems were used to carry out the measurements, the analog transponder (analogue images) and the Kinovea software (digital and computerized images). For the diagnosis of osteoporosis, measurements were made only on digital images using Prodigy Lunar software - BMD and Kinovea. The analysis of the results was carried out by applying analysis of variance (ANOVA) using Statistica R. The measurement uncertainty was estimated following the recommendations of the GUM method proposed in JCGM 101. The results indicated that Operator 1 "Experienced" presented the best repeatability of the results and consequently the lower uncertainty. The images with greater curvature in the vertebral column presented worse repeatability of the average values of Cobb angle and consequently higher values of expanded uncertainty, which demonstrates the difficulty to identify the vertebrae limits by the Operators when the angles are larger. The expanded uncertainty values (95%) associated with the Cobb angle for diagnosis of osteoporosis may be of the same order as the value of the measurand in some cases.