Alzheimer’s disease (AD) is a neurodegenerative condition damaging brain cells, leading to memory loss, cognitive decline, and forgetfulness, collectively known as dementia. Currently, there is no definitive treatment for AD, although medications may help slow its progression. Therefore, early detection is paramount to impede AD from reaching advanced stages. Distinguishing between healthy nerve cells and abnormal tissue in MRI images poses a significant challenge for medical professionals, requiring considerable expertise and time. Artificial intelligence (AI) techniques offer substantial support in early AD detection through the analysis of MRI images; the traditional diagnostic approaches for Alzheimer’s disease frequently identify the condition only after significant neurodegeneration and cognitive decline, posing a critical limitation. This study focuses on developing a Convolutional Neural Network (CNN) utilizing InceptionV3 as the foundational model to differentiate between two categories: Non-dementia and Mild Dementia. The MRI images, sourced from the open-access OASIS image dataset, encompass three dimensions: Axial, Coronal, and Sagittal. Upon incorporating additional functional layers, the proposed model demonstrated remarkable performance. For the sagittal dataset, the area under the curve (AUC) achieved a value of 0.9998 with an accuracy (ACC) of 0.9935. For the axial dataset, the AUC was 0.9890 with an ACC of 0.9687. Finally, for the coronal dataset, the AUC reached 0.9944, accompanied by an ACC of 0.9677. These metrics indicate mentioned classification algorithm outperforms existing methods in terms of accuracy and effectiveness.