Breast cancer is the most common malignancy among women worldwide. Conventional methods face challenges in accurately detecting tumors. To address these limitations, this paper presents BreastScanAI, a computational tool that combines 3D volumetric visualization with deep learning to enhance breast MRI interpretation. The system was developed using the Advanced-MRI-Breast-Lesions dataset. It features a Unity-based desktop application for real-time 3D rendering and anatomical exploration, incorporating a hybrid neural network with a ResNet-18 backbone for slice-level feature extraction and a bidirectional LSTM with attention for sequence modeling, enabling classification as benign or malignant. BreastScanAI achieved an accuracy of 74%, a 92% recall for malignant lesions, and an AUC of 0.79. The platform offers an interface with adjustable rendering settings and integrates a Flask-based backend for remote inference and automated natural-language. This paper demonstrates the potential of integrating interactive visualization with AI analysis to enhance diagnostic support.