In recent years, abnormalities in spinal intervertebral disc (IVD) disease have significantly increased, affecting a large population worldwide. Accurate detection and segmentation of IVD degeneration are critical for effective diagnosis and treatment planning. However, manual identification is time-consuming, prone to errors, and highly dependent on expert knowledge. This study addresses the challenge of automating IVD disease detection using advanced computer vision and deep learning techniques. Initially, the input data are gathered from publicly available datasets. The image is then pre-processed to remove undesirable noises and improve its quality using the Extended Cascaded Filtering (E-CF) and Improved Contrast Limited Adaptive Histogram Equalization (ICLAHE) techniques. From the pre-processed samples, the IVD regions are segmented with the aid of the Deep Residual Dilated U-Net (DRD-UNet) model. Next, feature extraction takes place to attain the necessary features by utilizing the Hybrid Gabor-Walsh Hadamard Transform (H_GWHT) method. Finally, the IVD from the input MRI images are identified by proposing a new Attention based Multi-Branch Convolutional ResNet-152 (A_MBCResNet) model. In order to enhance the efficiency of the proposed classifier, the parameters are optimally tuned by using the Gauss Chaotic Coati Optimization (GCCO) algorithm. A Python tool is used to implement this proposed work. Thus, the proposed study effectively identifies the IVDs from the given samples, and the performance measure of proposed accuracy is 97.75%.