Our end-to-end AI pipeline accelerates stroke diagnosis by transforming raw dynamic CT scans into quantitative perfusion maps (CBF, CBV). To ensure clinical flexibility, the system uses deep learning for segmentation and offers a choice of deconvolution methods (e.g., tSVD, oSVD) for tailored analysis. This provides clinicians with fast, reproducible data to make confident decisions in acute stroke care. 

The Sinomaly project pioneers a new frontier in medical imaging by applying machine learning directly to CT sinograms (the raw data captured by scanners before image reconstruction). While humans cannot interpret sinograms, AI can analyze these projections to detect brain anomalies like intracranial bleeding or acute stroke with incredible speed and efficiency bypassing the reconstruction techniques and sparing time. Our primary goal is to establish a comprehensive database of raw head CT sinogram data. This unique and powerful dataset will serve as the foundation for training robust machine learning models.