A Safe Window Into Brain Pressure

Continuous intracranial pressure (ICP) monitoring is critical in the management of severe neurological conditions, yet current gold-standard methods remain invasive and carry risks such as infection, hemorrhage, and limited monitoring access. This presentation introduces SafeICP, a non-invasive platform that estimates ICP at the bedside using near-infrared optical sensing and speckle contrast optical methods to capture cerebral blood-flow dynamics. The talk focuses on how these optical signals are translated into ICP estimates through machine learning. In particular, we show how deep learning models such as InceptionTime and Multi-Wavelet Decomposition Network learn temporal structure directly from raw blood-flow-index time series and recover absolute ICP values without manual feature engineering.

Speaker Bio: Viacheslav Danilov, PhD, is a scientist based in Barcelona with a decade of experience in artificial intelligence and data science. He has worked at European research institutions including Pompeu Fabra University, Sorbonne University, the Institute of Photonic Sciences, and Politecnico di Milano. His expertise spans healthcare AI, from medical imaging to physiological signal analysis, and he is currently working at the intersection of biophotonics and deep learning within the SafeICP project.