ABSTRACT
The aim of this study was to verify the relationship between quantitative T2 relaxation measurements of lumbar intervertebral discs (IVDs) and spinopelvic parameters in patients with chronic low back pain. The study was approved by the Clinical Hospital of the Ribeirao Preto Medical School (USP) Ethics Committee, and written consent was obtained from all patients. A total of 455 IVDs from 91 consecutive patients with chronic low back pain were included in this prospective study. All subjects were assessed using the Oswestry Disability Index and visual analogue scale questionnaires and were confirmed to have no other spine diseases except disc degeneration. Spinopelvic parameters including the pelvic incidence (PI), pelvic tilt (PT), sacral slope (SS), sagittal vertical axis (SVA), global tilt (GT), T1 pelvic angle (TPA), lumbar lordosis (LL), thoracic kyphosis (TK), pelvic incidence minus lumbar lordosis mismatch (PI-LL), and lack of lumbar lordosis (LLL) were measured. The study group was categorized according to the Roussouly classification. Sagittal T2 maps were acquired to extract the IVD relaxation times, and the complete manual segmentation of IVDs at all levels was performed using Display® software. Lumbar IVD T2 relaxation times showed significant correlation with PT (P<0.01), GT (P<0.01), TPA (P<0.01), PI-LL (P=0.01), and LLL (P=0.01). No difference was noted between Roussouly subtypes regarding T2 relaxation times at any disc level. Data from questionnaires showed no correlation with T2 relaxation times. Global tilt and T1 pelvic angle were correlated with IVD composition changes (T2 relaxometry). There was no correlation between clinical symptoms and IVD T2 relaxation times.
ABSTRACT
Este trabalho descreve o desenvolvimento de um método que classifica de forma semi-automática a degeneração de discos intervertebrais lombares em imagens de ressonância magnética ponderadas em T2. O conjunto de imagens consiste de 210 discos extraídos de exames de 94 indivíduos (20 a 80 anos). A classificação é feita por uma rede neural do tipo perceptron multicamada com 6 entradas, 15 neurônios na camada intermediária e 1 saída. Os resultados obtidos mostraram uma taxa média de acerto de 81,42%, com erro padrão de 9,11%.
This article describes the development of a method that classifies semi-automatic degeneration of lumbar intervertebral discs in magnetic resonance T2-weighted images. The dataset consists of images of 210 discs obtained from94 individuals (20 to 80 year old). An artificial neural network of the multilayer perceptron with 6 inputs, 15 neuronsin the hidden layer and 1 output, was used to check the efficiency of this study. Obtained an average rate of sucess of81.42%, with a standard error of 9.11%.