Search

Stroke is the third leading cause of mortality in many countries. Thermoacoustic imaging has the potential for stroke detection. However, some parameters in thermoacoustic imaging, such as speed of sound, are difficult to be obtained in advance, and a constant velocity value is assumed in the reconstructed algorithm. Thermoacoustic signals propagate through the soft brain tissue and the skull in actual stroke detection. This mismatch between the assumed and actual sound velocity will degrade the imaging quality. In this Letter, we propose a full waveform autofocus inversion method to reconstruct thermoacoustic images for stroke noninvasive and non-ionizing detection. Employing the difference between the simulation forward sensor signals and the measured signals, the approximate speed of sound distribution is updated continuously. The numerical simulation of a real human brain model and the experiment of a real human skull help us to validate the performance of the proposed method in clinical transcranial thermoacoustic detection. [ABSTRACT FROM AUTHOR]

Microwave induced thermoacoustic has attracted broad attention in recent years because of its potential for clinical noninvasive detection. The model matrix in the model-based reconstruction method is often constructed by simulation, assuming that the media is acoustically homogeneous. However, the thermoacoustic signal waves propagate through acoustically heterogeneous media in real imaging scenes and are recorded by sparse sensor array. Thus, there is a perturbation between the actual and pre-designed measurement matrix, called measurement matrix uncertainty. In this Letter, an improved model is proposed to reconstruct the sound pressure distribution from sparse signals with phase distortion caused by variations in the speed of sound between tissues. The measured thermoacoustic signals are employed to construct a series of complex signals, which contain amplitude and phase information. The signal model's sparse constraint of the estimated sound pressure map is combined with the nonzero constraint of the random measurements' phases with respect to the amplitude-only measures. Our simulation and experiment results indicate that the proposed model helps to improve the image quality reconstructed by sparse sampling in acoustically heterogeneous media. [ABSTRACT FROM AUTHOR]