<p>Recent attractive work in the medical imaging industry has been on spinal cord deformation predictive and classification systems. The chief demerits behind this study are the average abnormal region findings in the MRI data. Hence, the traditional neural algorithms and optimization approaches were not sufficient to predict the spine cord disorder with a high exactness score. Considering these issues, the present work has built a novel Spider ZFNet Classification System (SZCS) with the combination of a deep network and an optimal heuristic model. Compared to other traditional models, it has tuned features with the support of the spider's best solution in the ZFNet dense layer. This dense layer regulation by the spider model has helped to earn the finest outcome. The data obtained for valuing the introduced novel approach is lumbar spine cord degenerative data from Kaggle. In addition, the prime process such as filtering, feature selection, prediction, and classification, was performed with the support of the finest solution of the spider algorithm. Here, the incorporation of spider algorithm is the best solution to provide proper deformation region prediction outcomes from the trained MRI data. In addition, the introduced novel solution has scored the lowest misclassification score at 5% and high classification accuracy at 95%, which is the best solution compared to other traditional models.</p>