Medical Diagnostic by Artificial Intelligence applied to LIBS Elemental Microscopy

The overall goal of the DIAg-EM project is to use laser-induced breakdown spectroscopy (LIBS) elemental-microscopy with artificial intelligence processing to identify, localize and quantify pathogen elements present in selected human lung biopsies. We are proposing to develop a deep learning method able to perform, in real time, a non-supervised elemental identification, and to generate quantitative results over large areas of complex human specimens. A prospective clinical study will be initiated to collect samples from different hospitals in France, and from patients with various type of lung pathologies, caused by occupational exposures to inhaled mineral particles, metals, and dust. We expect to bring major inputs in the daily diagnoses of lung occupational and environmental diseases, and to produce useful data about past- or present exposure to agents that would help clinicians to better understand the disease of a given patient, therefore reinforcing his clinical management.