Event cameras offer unique advantages for facial keypoint alignment under challenging conditions, such as low light and rapid motion, due to their high temporal resolution and robustness to varying illumination. However, existing RGB facial keypoint alignment methods do not perform well on event data, and training solely on event data often leads to suboptimal performance because of its limited spatial information. Moreover, the lack of comprehensive labeled event datasets further hinders progress in this area. To address these issues, we propose a novel framework based on cross-modal fusion attention (CMFA) and self-supervised multi-event representation learning (SSMER) for event-based facial keypoint alignment. Our framework employs CMFA to integrate corresponding RGB data, guiding the model to extract robust facial features from event input images. In parallel, SSMER enables effective feature learning from unlabeled event data, overcoming spatial limitations. Extensive experiments on our real‑event E‑SIE dataset and a synthetic‑event version of the public WFLW‑V benchmark show that our approach consistently surpasses state‑of‑the‑art methods across multiple evaluation metrics. 

Coronary artery lesions in computed tomography coronary angiography (CTCA) are a leading cause of sudden cardiac death, yet labeling is challenging due to the complex 3D structure and small vessel sizes. In this paper, we propose DAAD-CTCA (Dual Augmentation of Image and Feature Domains for Unsupervised Anomaly Detection of Coronary Lesions in CTCA), a novel framework designed to identify coronary artery anomalies without requiring annotated data. DAAD-CTCA introduces a dual augmentation strategy that synthesizes realistic and diverse anomalies by combining image-level and feature-level augmentations. To effectively learn from diverse augmented features, we employ a pretrained backbone suited for small cross-sectional coronary images and integrate a feature selection module that focuses on the informative features. Evaluated on the public ASOCA dataset, our method achieves an AUROC of 99.0% and an AP of 99.8%, outperforming state-of-the-art techniques. These results demonstrate the effectiveness of our approach for unsupervised anomaly detection in coronary artery CTCA images, and also suggest that the framework could serve as a versatile foundation for broader applications in medical imaging.