29 November 2022 in IEEE Transactions on Biomedical Engineering
Objective: According to the acoustic reciprocity theorem (ART), we propose a system matrix reconstruction algorithm of thermoacoustic imaging for magnetic nanoparticles (MNPs) by a single-pulse magnetic field. Methods: In both cases of inhomogeneous and homogeneous acoustic velocity, we respectively derive the linear equation between the sound pressure detection value and the distribution of MNPs. The image reconstruction problem is converted to an inverse matrix solution by using the truncated singular value decomposition (TSVD) method. Results: In forward problem, the calculated forward results are consistent with the simulated thermoacoustic signal signals. In inverse problem, we build the two-dimensional breast cancer model. The TSVD method based on the ART faithfully reflects the distribution of abnormal tissue labeled by the MNPs. In the experiment, the biological sample injected with the MNPs is used as the imaging target. The reconstructed image well reflects the cross-sectional images of the MNPs area. Conclusion: The TSVD method based on the ART takes into account energy attenuation and inhomogeneous acoustic velocity, and use a non-focused broadband ultrasonic transducer as the receiver to obtain a larger imaging field-of-view (FOV). By comparing the image metrics, we prove that the algorithm is superior to the traditional time reversal method. Significance: The TSVD method based on the ART can better suppress noise, which is expected to reduce the cost by reducing the number of detectors. It is of great significance for future clinical applications.
1 November 2022 in AIP Advances
Under a single-pulse magnetic field (SMF), we proposed a thermoacoustic (TA) concentration imaging method of magnetic nanoparticles (MNPs). This method derived the relationship between the concentration of MNPs and TA signals based on the mechanism of the TA effect. In the simulation, we ideally reconstructed the concentration map of the MNPs from the simulated TA signals. It proves the feasibility of the proposed method. In the experiment, a multi-channel acquisition system was built, and the MNP suspension with the iron concentration of 1 mg/ml was injected into two gaps of the phantom. Then, 128 TA signals were obtained by a multi-channel acquisition system. Finally, the imaging results faithfully reconstructed the cross-sectional image of the MNP area in the phantom. It is demonstrated that the TA concentration imaging of the MNPs under the SMF has the potential to achieve low-concentration MNP imaging in vivo, which provides a promising molecular imaging method for imaging MNP-labeled cancer tissues.
3 November 2025 in IEEE Transactions on Microwave Theory and Techniques
Transcranial imaging is an indispensable method for the diagnosis of cerebral diseases that are major threats to human health. Microwave-induced thermoacoustic tomography (MITAT) is a promising hybrid technique for nonionizing, noninvasive, and time and cost-effective modality for transcranial imaging with a compact hardware system. Nevertheless, the conventional MITAT technique cannot efficiently deal with the acoustic inhomogeneity issue caused by the skull, which leads to low image quality. Although MITAT combined with deep learning (DL) has shown compelling ability in reconstructing high-quality images in a transcranial manner, the requirement for too many training datasets may hinder potential applications. In this work, we propose a new DL-based MITAT modality that leverages a physics-informed neural network (PINN) to improve the image quality of transcranial imaging using much less training data. The PINN is based on the acoustic reciprocity theorem (ART), and the proposed method is named as DL-MITAT-ART. We perform ex vivo 2-D experimental testing employing intact cynomolgus monkey skulls and blood vessel phantoms. The imaging results demonstrate that the proposed DL-MITAT-ART method can faithfully recover the blood vessel phantoms in a transcranial manner applying only 175 training datasets, more than ten times fewer than those for the traditional DL-MITAT methods. This work provides a novel paradigm for PINN-based MITAT technique for transcranial imaging. It is highly meaningful for cerebral disease diagnosis based on MITAT or ultrasonography and microwave imaging applications involving an inhomogeneous environment.
10 February 2023 in Journal of Physics D: Applied Physics
Magneto-thermal-acoustic imaging with magnetic nanoparticles (MNPs) is a novel medical imaging modality based on the thermoacoustic (TA) effect of MNPs. We proposed a new dual-modality imaging strategy that combines the TA and echo signals. Based on the synthetic aperture method, we used TA signals to image cancer tissues labeled with MNPs and applied echo signals to achieve tissue structure imaging. In the simulation, we constructed a breast cancer model and assumed that cancer tissue was labeled with MNPs. Under a limited field of view, the TA and echo signals were collected using a multi-element transducer array. By fusing the imaging results of the TA and echo signals, we obtained a boundary distribution map of the breast cancer model, and accurately located the tumor tissue. In this experiment, we built a multi-channel acquisition system. Pork tissue labeled with MNPs was used as the breast cancer phantom. We collected a total of 120 TA and echo signals at 1 mm intervals using a multi-element transducer array and faithfully imaged the boundary distribution map of the phantom. Compared with the traditional linear back projection method based on focused transducers, we demonstrated that the proposed method not only improves the lateral resolution of the image, but also makes accurate cancer identification easier. This provides a promising medical imaging technology for clinical applications.
1 October 2026 in IEEE Sensors Journal