Figure 1
From: Fluorescence lifetime imaging with a megapixel SPAD camera and neural network lifetime estimation
Principle of time-gated acquisition and the machine learning model. (a) Fluorescence decay is sampled with a number of gates, each shifted by a minimum of 36 ps. Each exposure corresponds to a ‘time bin’, which samples a different part of the fluorescence decay signal. (b) The ANN architecture (see “Methods”) consists of one input layer (IL), one output layer (OL), and a series of hidden layers (HLi, with \(i = 1,2,3\)). Each of these layers consists of a fully-connected dense layer (dark blue) followed by with rectified linear unit (ReLU) activation function (light blue). The input layer is fed with the fluorescence decay signal recorded by a single pixel of the SPAD array.
