Timely and accurate interpretation of medical images is a cornerstone of modern clinical diagnosis. In this thesis, we focus on developing a medical image classification system using deep learning techniques, specifically convolutional neural networks (CNNs). Our goal is to build a model capable of detecting degenerative spinal conditions from magnetic resonance imaging (MRI) scans, using the RSNA 2024 dataset. The model is trained on thousands of annotated images, covering various lumbar spine levels and classifying conditions such as spinal stenosis, disc degeneration, and nerve compression. The objective is to combine high diagnostic accuracy with rapid processing time, ultimately contributing to more efficient clinical workflows. This work presents the architecture we implemented, along with the preprocessing, training, and evaluation strategies used. We also discuss challenges in the field, including data imbalance and integration with existing healthcare systems. Through this project, we aim to take a step forward in enhancing diagnostic support with the power of artificial intelligence.