Rongsheng Lu

Low‐field magnetic resonance imaging (MRI) offers a cost‐effective and accessible alternative to high‐field systems but inherently suffers from low signal‐to‐noise ratio (SNR) and prolonged scan times. Although k ‐space under‐sampling shortens scan time, the resulting incomplete data typically introduces resolution loss and artifacts in the reconstructed images. Deep learning–based MRI reconstruction methods primarily focus on high‐field, high‐quality MRI data and image‐domain reconstruction, with limited attention to the unique characteristics of low‐field data and the relationship between k ‐space and image‐domain representations. Furthermore, existing k ‐space reconstruction approaches often overlook the intrinsic encoding dependencies within k ‐space and lack mechanisms for learnable feature fusion across the two domains. To address the limitations, we propose DUAG, a deep dual‐domain interaction reconstruction framework with adaptive gating fusion for low‐field MRI. The framework employs a cascaded deep architecture with multi‐scale U‐Net structures to achieve hierarchical feature representation and incorporates a hybrid dual‐domain interaction module. Attention mechanisms are introduced to model long‐range dependencies, enabling precise capture of underlying correlations among k ‐space frequency encodings and image‐domain pixels. In addition, an adaptive gating fusion strategy is designed for dynamical weighting and cross‐domain feature fusions. In order to enhance feature reuse and improve the generalization ability of the model. Experiments on public 0.3 T low‐field dataset show that DUAG reaches 42.79 ± 0.75 PSNR and 0.920 ± 0.010 SSIM. To further verify the generalization capability, we conducted real‐world experiments on the laboratory‐collected 0.5 T low‐field MRI scanner data, DUAG achieves superior reconstruction performance of 34.13 ± 1.02 PSNR and 0.889 ± 0.013 SSIM. The proposed framework provides a promising solution for high‐quality low‐field MRI reconstruction, which is expected to promote the deployment of cost‐effective MRI systems in resource‐constrained clinical settings.

Low-cost mobile breast magnetic resonance imaging (MRI) systems are of great significance for promoting early screening of breast diseases. However, the low B₀ magnetic field strength of the mobile MRI system and the B₀ direction perpendicular to the human body usually make the design of breast coils difficult. To address these problems, we propose a composite four-channel breast receive coil that consists of breast sub-coils and axillary tail sub-coils in a 50 mT breast MRI scanner. Breast sub-coils were designed as cap-shaped to reduce signal loss caused by extreme B₁-field inhomogeneity at the breast root. And axillary-tail sub-coils were oriented geometrically orthogonal to the breast sub-coils, compensating for the imaging void at the axillary tail of the breast sub-coil. In order to solve the decoupling problem of the composite four-channel receive coil, a simulation model of ultra-low field decoupling radio-frequency circuit was constructed based on S-parameter analysis. By employing numerical modeling of decoupling components, the challenge of the difficult decoupling of LC components was effectively solved. Phantom imaging results show that the resolution of the 50 mT breast MRI can reach 0.8 mm, with optimal reception and inter-channel decoupling performances. In vivo breast imaging results show that the proposed breast coil can observe the whole breast and cardiac silhouette, without signal loss at the breast root. In summary, the four-channel composite breast receive coil proposed in this paper will effectively overcome the limitations of 50 mT breast MRI, promoting the application of mobile MRI devices in breast disease screening.

The renewed interest in ultra‐low‐field MRI (< 0.1 T), with its advantages of cost‐effectiveness and portability, is expected to drive a paradigm shift from traditional MRI scanners to point‐of‐care imaging systems. Here, we characterized the in vivo T₁ and T₂ relaxation times of healthy breast tissues using quantitative MRI at 50 mT to advance the development of ultra‐low‐field breast MRI methodologies. First, the developed in vivo quantitative mapping protocols were tested and validated using CuSO₄‐doped water phantom experiments on the custom‐built 50‐mT MRI prototype. Then, in vivo quantitative T₁ and T₂ mapping of human breast tissues was measured in healthy female subjects. The T₁ and T₂ values for fat tissue, fibrous tissue, and glandular tissue components were determined using an automatic segmentation method based on the relaxation times. The mean T₁ times were 121.21 ± 14.57, 182.26 ± 20.96, and 238.32 ± 10.71 ms for fat tissue, fibrous tissue, and glandular tissue, respectively. While the mean T₂ times were 111.36 ± 14.28 and 69.44 ± 15.26 ms for fat tissue and fibroglandular tissue, respectively. For healthy breast tissues, the significantly shorter T₁ times at 50 mT are favorable for improving the imaging speed. T₂ times are similar to those seen at conventional clinical field strengths. Additionally, the relatively small difference in T₂ times between fibrous tissue and glandular tissue made it difficult to segment these two tissues independently. This work can provide a valuable reference for future breast tissues diagnostic imaging in ultra‐low‐field MRI scenarios, thereby promoting the wider adoption of portable breast MRI in routine breast disease screening.

Recently, there has been an impetus for developing low-field portable magnetic resonance imaging (MRI) technologies, driving a paradigm shift from conventional MRI scanners to point-of-care (POC) MRI scenarios. To facilitate the miniaturization of portable whole-body MRI, the weight of permanent magnet was reduced to 800 kg by shortening its axial depth, which allowed for the development of a 50 mT portable whole-body MRI prototype. The conventional MRI equipment usually uses Faraday shielded rooms to block electromagnetic interference (EMI) in complex environments, but it is difficult to apply in POC MRI scenarios. Furthermore, compared to the prevailing portable brain MRI scanner, the developed 50 mT short-axis portable whole-body MRI scanner is more susceptible to the EMI in open environments. To address this, a new active EMI rejection strategy featuring reduced hardware configuration and shorter model training time was proposed for unshielded portable whole-body MRI scanner. First, a multi-indicator evaluation method was designed to optimize the deployment of multiple EMI sensing coils. High-quality EMI sensing coils were selected, effectively reducing the hardware requirement by half. Then, the se-CNN model with three-layer convolutional neural network and SENet attention mechanism, was constructed to implement the EMI noise prediction and rejection. For model training, EMI noise data is acquired synchronously from multiple EMI sensing coils and the main MRI receive coil during the imaging data acquisition intervals, with a typical training time of about 2 minutes. The phantom and in-vivo imaging experiments demonstrated the improved image quality and validated the outperformance of the proposed EMI rejection strategy. This EMI rejection method has the potential to turbocharge the deployment of portable whole-body MRI scanners in clinical POC diagnosis and treatment.

Creating an effective gradient coil is essential for enabling portable, high-quality magnetic resonance imaging (MRI). This study introduces a method that incorporates a power consumption optimization function into spherical harmonic analysis to design an efficient gradient coil for a portable Halbach magnet with a low height-to-diameter ratio. The new design approach resulted in significant improvements over traditional methods, with the Y-direction gradient coil achieving an efficiency of 38.1 mT/m/A, reducing inductance and resistance by 43.09% and 24.77%, respectively. The Z-direction gradient coil also showed enhanced efficiency at 18.1 mT/m/A, a 45.97% improvement. Using this gradient coil design, a portable MRI device was constructed by integrating it with a homemade portable Halbach magnet, radio frequency probe, and electronic components from the laboratory. Imaging tests on a water phantom demonstrated the device’s ability to resolve small structures below 1 mm, achieving an image resolution of 62.5 μm × 62.5 μm. Overall, the gradient coil design method presented in this study successfully meets the requirements for portable, high-resolution MRI on-site.