We developed a data-adaptive, self-supervised deep learning method combining sparse coding (SC) + visual transformer (ViT) for denoising pediatric cardiac photon-counting CT. Preserves anatomy, adapts to noise, generalizes well—even to preclinical data.
Clark, D. P., Cao, J. Y., & Badea, C. T. (2025). Denoising pediatric cardiac photon-counting CT data with sparse coding and data-adaptive, self-supervised deep learning. Medical Physics. https://doi.org/10.1002/mp.17918
