ResViTM-Net: A novel hybrid deep learning model for automated Tuberculosis (TB) detection. It cleverly combines local features from CNNs, global context from ViTs, and patient clinical prior information. Achieved 96.0% overall accuracy and significantly faster training speed on three public datasets.