Work

Template-based and model-based decoding of movie clip identities from brain hemodynamics with high-density diffuse optical tomography