Cervical Disc Herniation Detection Using Morphological Image Processing and Neural Network Classification

Intervertebral discs, the rubbery cushions between spinal bones, can bulge or rupture (herniate) beyond their normal boundaries. This can irritate nearby nerves, causing pain, particularly in the upper back with cervical disc herniation. This work explores the use of morphological transformations of image and simulated neural network classification to analyze magnetic resonance imaging (MRI) data for automatic detection of disc herniation. Morphological operations manipulate individual pixels in an image based on the measurement of their surrounding pixels. In this paper, these techniques are employed for herniated regions identified in the magnetic resonance image scans. Specifically, the edge-detected canny image is suitable for identifying the edges of these regions, followed by morphological operations with structuring elements to refine and enhance the image quality. Using non-flat structuring elements such as arbitrary and ball applied to the images. Morphological operations named dilation and erosion are applied for four stages of herniated canny images. A neural network classification method is employed to analyze the processed images at different stages of disc herniation. This allows for the calculation of the effectiveness of each structuring element in accurately identifying herniated discs. The results of the morphological operations could be used to extract quantitative measurements of hernia size, shape, or other relevant features, aiding in treatment planning and monitoring.