Content area
Abstract
Scoliosis and spondylolisthesis are spine pathologies affecting 1.5-3% and 13.6% of the population, respectively. These diseases have the potential to further progress. These tridimensional pathologies are mainly studied using two-dimensional geometric indices, which reflect only a fraction of the morphological, biomechanical and biochemical variations of the spine. Clinical interpretations of these pathologies and of their evolution are based on the limited information of spine modifications. The intervertebral disc (IVD) is the soft tissue between adjacent vertebrae that allow the mobility of the spine between the rigid segments. Spine pathologies lead to premature degeneration of the IVD.
In our study, we hypothesize that the MRI signal intensity within clinical T2-weighted images is sensitive to the spine pathology and to its severity. Thus this imaging technique could provide information on the geometrical, biochemical and mechanical properties of the IVD, and facilitate in-vivo follow-up of the evolution of these spine pathologies. This project aims to develop techniques to analyse the tridimensional geometry as well as the Gaussian distribution of the T2-weighted MRI signal within clinical images of the IVD, the annulus fibrosus (AF) and the nucleus pulposus (NP) of patients affected with various spine pathologies. These tools will assess whether or not a specific degeneration of the IVD is caused by the spine pathologies depending on their severity.
In order to analyse automatically the tridimensional MRI signal within the IVD, it is necessary to segment clinical images. A semi-automated method was used in this project and its reproducibility was tested. This method is less time-consuming compared to the commonly used manual methods that are reported in the literature.
MRI signal analysis tools were developed to detect its sensitivity within the IVD to spine pathologies and their severities. These tools allowed a Gaussian and geometric distribution analysis of the MRI signal intensity within the IVD.
A cohort of 79 subjects (32 scoliosis, 32 spondylolisthesis, 15 controls) was studied. A normalization of the signal intensity was done in order to compare images from patients with variable parameters such as the acquisition gain. This study tested two normalizations of the intensity of the signal. The first one was based on the intensity within the cerebrospinal fluid (CSF) as described in the literature. The second one was based on the intensity of the bone tissues, which are known to vary with spine pathologies. A comparison of the results obtained with both normalizations on the total cohort was then performed.
Results of this study demonstrated the reproducibility of this method for image segmentation achieved by a single operator among the patients. Moreover, significant differences were observed within geometric and Gaussian distributions of the MRI signal intensity for T2-weighted clinical images of patients suffering from spine pathologies such as scoliosis and spondylolisthesis.
This study used innovative approaches and for the first time, allowed tridimensional analysis of the MRI signal distribution within the IVD, the Gaussian distribution of the MRI signal within the IVD, the use and comparison of two signal normalizations and a semi-automatic segmenting method. However, there are a few limits in our study such as the variability of the results which was linked to the low resolution of the IVD within clinical images.
We developed new fundamental MRI signal analysis criteria reflecting both morphologic and biochemical variations within the IVD. It allowed the detection of degeneration of the IVD specific to two spine pathologies (Scoliosis/spondylolisthesis). This study can therefore provide a new basis for in-vivo follow-up of spine pathologies in order to analyse their progression. Moreover, the use of other MR image-weighting as well as quantitative MRI images could help to analyse other morphologic, biochemical and mechanical aspects of the IVD affected by spine pathologies.
