According to the
microwave-induced thermoacoustic transcranial imaging method based on the
physical information neural network provided by the technical scheme of the invention, a matrix P is established by receiving two-dimensional head
simulation ultrasonic signals, the
signal-to-
noise ratio of the two-dimensional head
simulation ultrasonic signals is obtained according to an experiment, and
Gaussian noise is added to obtain an ultrasonic
signal training set; reconstructing the ultrasonic
signal training set by adopting a BP
algorithm to obtain a first input BP image, obtaining a second input BP image, and combining the first input BP image to obtain a final input BP image; performing dimension transformation on the matrix P to obtain a
system matrix G; adding the final input BP image into a
loss function taking a matrix equation of the
microwave-induced
thermoacoustic imaging system as a constraint term to calculate a loss value for training, forming a set according to the final input BP image, an initial sound source, a
system matrix G and a matrix P, dividing the set into a final
training set and a final
verification set according to a preset proportion, and updating hyper-parameters for multiple times, and a final trained
network model is obtained, and an imaging result is obtained.