Intervertebral disc degeneration is related to chronic back pain and functional incapacity. Magnetic Resonance Imaging (MRI) is the modality of choice for diagnosing this condition, providing both morphological and biochemical information for the disc tissue. In clinical practice, grading schemes based on qualitative descriptions of disc image features such as the signal intensity and disc height are commonly used for disc degeneration severity evaluation. However, these grading schemes have a limited number of degeneration severity classes which impairs the detection of small changes. Additionally, this grading is susceptible to inter and intra observer variabilities.udTo deal with these issues, this study introduces a system for the automated quantification and computer aided diagnosis of disc degeneration severity from spine MRI. The proposed system consists of a segmentation method, a quantification process, and a classification scheme. An atlas-based segmentation approach, combining prior anatomical knowledge provided by means of a probabilistic disc atlas with fuzzy clustering techniques, was designed for extracting the disc region from the images. In the quantification process, texture and shape descriptors are calculated from the segmented disc region aiming to capture structural and biochemical alterations of the tissue related to degeneration. Finally, the classification scheme exploits this information for differentiating between degeneration severity grades. The system is tested on a case sample of 255 discs from conventional T2-weighted MR images acquired by a 3 Tesla scanner.udResults indicate that the atlas-based method provides accurate disc segmentation, texture descriptors measuring intensity inhomogeneity can serve the quantification of degeneration severity, and the computer aided diagnosis scheme achieves high agreement to clinical diagnosis.udConcluding, the proposed system could be a valuable tool in hands of physicians to support clinical diagnosis of disc degeneration, track the evolution of disease progress and monitor the response to treatment in a simple, precise and repeatable manner.
展开▼
