Handheld microscope examination stands as the main diagnostic tool for malaria even though it needs skilled professionals while being time-consuming. Deep learning technologies in computer-aided diagnosis systems have become increasingly popular. Current methods exist to detect malaria parasites in high-resolution images yet small parasites remain difficult to detect accurately. State-of-the-art deep learning models identify malaria parasites with difficulty due to spatial information degradation alongside high false positive detection which leads to both diagnostic precision and operational efficiency decline. This study proposes a deep learning framework involving a specialized Convolutional Neural Network structure with a lightweight MobileNetV2 pre-trained component has been developed for malaria-infected cell classification. The system employs three elements: Residual Blocks together with Receptive Field Block (RFB) and Convolutional Block Attention Module (CBAM) for better feature learning across spatial domains and channels. This model achieves higher accuracy recall and precision plus lower false positives by detecting both parasite and uninfected cell imagery which will lead to improved predictions. The model presents diagnostic accuracy enhancement capabilities while delivering valuable prognostic insights that lead to successful clinical and resource-saving applications.