MPIGAN: An end-to-end deep based generative framework for high-resolution magnetic particle imaging reconstruction.

BACKGROUND: Magnetic particle imaging (MPI) is a recently developed, non-invasive in vivo imaging technique to map the spatial distribution of superparamagnetic iron oxide nanoparticles (SPIONs) in animal tissues with high sensitivity and speed. It is a challenge to reconstruct images directly from the received signals of MPI device due to the complex physical behavior of the nanoparticles. System matrix and X-space are two commonly used MPI reconstruction methods, where the former is extremely time-consuming and the latter usually produces blurry images. PURPOSE: Currently, we proposed an end-to-end machine learning framework to reconstruct high-resolution MPI images from 1-D voltage signals directly and efficiently. METHODS: The proposed framework, which we termed "MPIGAN", was trained on a large MPI simulation dataset containing 291 597 pairs of high-resolution 2-D phantom images and each image's corresponding voltage signals, so that it was able to accurately capture the nonlinear relationship between the spatial distribution of SPIONs and the received voltage signal, and realized high-resolution MPI image reconstruction. RESULTS: Experiment results showed that, MPIGAN exhibited remarkable abilities in high-resolution MPI image reconstruction. MPIGAN outperformed the traditional methods of system matrix and X-space in recovering the fine-scale structure of magnetic nanoparticles' spatial distribution and achieving enhanced reconstruction performance in both visual effects and quantitative assessments. Moreover, even when the received signals were severely contaminated with noise, MPIGAN could still generate high-quality MPI images. CONCLUSION: Our study provides a promising AI solution for end-to-end, efficient, and high-resolution magnetic particle imaging reconstruction.

Content-Noise Feature Fusion Neural Network for Image Denoising in Magnetic Particle Imaging∗.

Magnetic particle imaging (MPI) is a tomographic imaging method that quantitatively determines the distribution of magnetic nanoparticles (MNPs). However, the performance of MPI is primarily limited by the noise in the receive coil and electronic devices, which causes quantification errors for MPI images. Existing methods cannot efficiently eliminate noise while preserve structural details in MPI images. To address this problem, we propose a Content-Noise Feature Fusion Neural Network equipped with tailored modules of noise learning and content learning. It can simultaneously learn content and noise features of raw MPI images. Experimental results show that the proposed method outperforms the state-of-the-art methods on structural details preservation and image noise reduction of different levels.

An adaptive multi-frame parallel iterative method for accelerating real-time magnetic particle imaging reconstruction.

Objective. Real-time reconstruction of magnetic particle imaging (MPI) shows promising clinical applications. However, prevalent reconstruction methods are mainly based on serial iteration, which causes large delay in real-time reconstruction. In order to achieve lower latency in real-time MPI reconstruction, we propose a parallel method for accelerating the speed of reconstruction methods.Approach. The proposed method, named adaptive multi-frame parallel iterative method (AMPIM), enables the processing of multi-frame signals to multi-frame MPI images in parallel. To facilitate parallel computing, we further propose an acceleration strategy for parallel computation to improve the computational efficiency of our AMPIM.Main results. OpenMPIData was used to evaluate our AMPIM, and the results show that our AMPIM improves the reconstruction frame rate per second of real-time MPI reconstruction by two orders of magnitude compared to prevalent iterative algorithms including the Kaczmarz algorithm, the conjugate gradient normal residual algorithm, and the alternating direction method of multipliers algorithm. The reconstructed image using AMPIM has high contrast-to-noise with reducing artifacts.Significance. The AMPIM can parallelly optimize least squares problems with multiple right-hand sides by exploiting the dimension of the right-hand side. AMPIM has great potential for application in real-time MPI imaging with high imaging frame rate.

The level of ROCK1 and ROCK2 in patients with pulmonary hypertension in plateau area.

Pulmonary hypertension (PH) is defined as the mean pulmonary artery pressure (mPAP) ≥25 mmHg under the sea level in resting state. ROCK1 and ROCK2 can be combined to cause the damage of vascular endothelial function. To explore the differences of ROCK1 and ROCK2 in subjects with pulmonary hypertension or normal pulmonary artery pressure in plateau area, and to further understand the mechanism of Rho/rho-kinase pathway activation for promoting pulmonary hypertension, we collected 64 patients with pulmonary hypertension and 87 normal pulmonary artery healthy controls. All subjects were hospitalized in Cardiology or Respiration Department of Qinghai Provincial Peoples' Hospital from December 2016 to June 2017. The pulmonary artery systolic pressure was measured by Doppler ultrasound, and serum ROCK1 and ROCK2 levels were tested by enzyme linked immunosorbent assay (ELISA). We found that the serum ROCK2 concentration in the pulmonary hypertension group was significantly higher than that in the control group, but serum ROCK1 level had no significant difference. ROCK2 plays a leading role in pulmonary hypertension in the plateau region, so selective ROCK2 inhibitors will be more effective in improving pulmonary hypertension.