Acute Ischemic Stroke (AIS) is a life-threatening condition requiring rapid and accurate detection of vascular occlusions to guide effective treatment, such as endovascular thrombectomy (EVT). Digital Subtraction Angiography (DSA) serves as the gold standard for real-time vascular imaging during EVT, but manual occlusion detection on DSA sequences poses challenges due to anatomical complexity and time constraints. This thesis introduces OccluNet, a spatio-temporal deep learning model that integrates YOLOX, a single-stage object detector, with transformer-based temporal attention mechanisms to automate occlusion detection in DSA sequences. We compared OccluNet with a YOLOv11 baseline trained on either individual DSA frames or minimum intensity projections. Two spatio-temporal variants were explored for OccluNet: pure temporal attention (OccluNet1) and divided space-time attention (OccluNet2). Evaluation on pre-EVT DSA sequences from the MR CLEAN Registry revealed the model’s capability to capture temporally consistent features, with precision and recall of 89.02% and 74.87%, respectively, significantly outperforming the baseline model (p < 0.001, McNemar’s test) and with both attention variants attaining similar performance (p = 0.60, McNemar’s test). These findings highlight the potential of temporal modeling in improving automated occlusion detection and outline key areas for further enhancement.