This work presents the development of a convolutional neural network (CNN) for the classification of Alzheimer’s magnetic resonance imaging (MRI) scans, with the aim of optimizing the early diagnosis process of the disease through the use of digital image processing. Alzheimer’s diagnosis faces a significant challenge due to the need for precise and rapid identification of the different stages of the disease, which can enhance medical care and improve patients' quality of life.

To address this issue, a dataset consisting of MRI images obtained from the Kaggle platform was used, which includes a wide variety of brain images at different stages of the disease. The images were categorized into three classes: "No Alzheimer’s," "Mild," and "Advanced". The methodology involved designing three CNN models with different configurations of convolutional layers, dense layers, and regularization techniques, as well as preprocessing the images by converting them to grayscale and normalizing pixel values. The training process incorporated data augmentation techniques and hyperparameter tuning to improve the model’s accuracy.

Model 3, which had the best configuration, achieved an accuracy of 95.1% and a loss of 0.32, standing out as the most efficient in classifying the images. The results were evaluated using confusion matrices, which demonstrated the model's ability to correctly classify Alzheimer’s images into the three categories.

This innovative approach not only improves the efficiency of Alzheimer’s disease diagnosis but also facilitates the implementation of a medical decision- support system. The successful implementation of this technology represents a significant opportunity to modernize traditional diagnostic imaging methods, contributing to the advancement of precision medicine and the development of artificial intelligence technologies applied to healthcare.