In brain-related diseases, including Brain Tumours and Alzheimer's, accurate and timely diagnosis is crucial for effective medical intervention. Current state-of-the-art (SOTA) approaches in medical imaging predominantly focus on diagnosing a single brain disease at a time. However, recent research has uncovered intricate connections between various brain diseases, realizing that treating one condition may lead to the development of others. Consequently, there is a growing need for accurate diagnostic systems addressing multiple brain-related diseases. Designing separate models for different diseases, however, can impose substantial overhead. To tackle this challenge, our paper introduces BrainMNet, an innovative neural network architecture explicitly tailored for classifying brain images. The primary objective is to propose a single, robust framework capable of diagnosing a spectrum of brain-related diseases. The paper comprehensively validates BrainMNet's efficacy, specifically in diagnosing Brain tumours and Alzheimer's disease. Remarkably, the proposed model workflow surpasses current SOTA methods, demonstrating a substantial enhancement in accuracy and precision. Furthermore, it maintains a balanced performance across different classes in the Brain tumour and Alzheimer's dataset, emphasizing the versatility of our architecture for precise disease diagnosis. BrainMNet undergoes an ablation study to optimize its choice of the optimal optimizer, and a data growth analysis verifies its performance on small datasets, simulating real-life scenarios where data progressively increases over time. Thus, this paper signifies a significant stride toward a unified solution for diagnosing diverse brain-related diseases.