COMPUTER AIDED CHARACTERIZATION OF DEGENERATIVE DISC DISEASE EMPLOYING DIGITAL IMAGE TEXTURE ANALYSIS AND PATTERN RECOGNITION ALGORITHMS

In this study… The use of texture for the characterization of cervical intervertebral disc degeneration in magnetic resonance images was investigated A computer-based classification system using textural features was proposed for the diagnosis of Degenerative Disc Disease and more specifically for the detection of disc space narrowing Degenerative Disc Disease (1/3) Cervical DDD refers to the degeneration of cervical intervertebral discs In DDD the intervertebral disc loses its hydration and elasticity leading to disc space narrowing and/or disc herniation Degenerative Disc Disease (2/3) DDD is a major cause of neck pain and stiffness In severe cases cervical DDD can cause sciatica and back pain, as well as muscle weakness Degenerative Disc Disease (3/3) Several imaging modalities are used in DDD diagnosis such as: z X-rays z Magnetic Resonance Imaging (MRI) z Computed Tomography- Discography MRI is considered the modality of reference since it is non-invasive, involves no ionizing radiation and provides excellent soft tissue contrast DDD Evaluation Using MR Images DDD diagnosis is based on manual measurements of the intervertebral disc space and comparison to the adjacent levels disc space For the needs of the present study T2-weighted MR saggital images of the cervical spine were reviewed by an experienced orthopaedist and the discs were characterized as normal or degenerated (narrowed) Why texture? In this study a computer based system was proposed for the automatic characterization of intervertebral disc degeneration The system was based on texture analysis of the disc images The selection of texture as a descriptor for DDD was motivated by the physics behind MRI. More specifically MR signal depicts the biochemical composition of the imaged structures and especially the hydrogen nuclei. Because of that, the alterations caused by the degeneration of intervertebral discs are expected to change the texture of MR images