Frequency Structure Consistency Learning Approach for Fast Signal Matrix Measurement in Magnetic Particle Imaging

A core challenge for accelerating high-resolution image reconstruction in magnetic particle imaging (MPI) is to rapidly measure the high-resolution signal matrix (SM). For accurately measuring the high-resolution SM, the number of meticulous measurements at numerous positions in the field-of-view proves time-consuming (measurement of a $37\times 37\times 37$ SM takes about 32 h). To improve reconstructed image quality and shorten SM measurement time, existing methods explore deep learning methods to use time-saving measured low-resolution SM (a $9\times 9\times 9$ SM just takes about 0.5 h), generating high-resolution SM. However, previous methods show poor performance for high-frequency signal recovery in SM, which is crucial for reconstructing high-resolution images. In this work, to achieve a high-resolution SM recovery and shorten its acquisition time, we propose a novel learning approach of frequency structure consistency (FSC) loss and signal component embedding strategy. Our approach guides the model to learn the distribution of the frequency of SM. We evaluate our methods and state-of-the-art methods on the two simulation datasets and four public measured signal matrices in Open MPI Data. The results show that our method outperforms the compared methods in high-frequency structural signal recovery. Additionally, our method can recover a high-resolution SM based on a downsampling factor of 16, less than 15 s, which accelerates the acquisition time over 60 times faster than the measurement-based high-resolution SM with the minimum error (nRMSE = 0.041).