Nitroxide probes have been commonly used for biomedical EPR imaging applications; however, image quality has been limited by the number of image projections, as well as the probe linewidth and hyperfine structure. In the current study, we evaluate the use of fast millisecond scan EPR projection acquisition along with a novel reconstruction algorithm optimized for 3D spatial EPR image reconstruction from a high number of noisy projections. This reconstruction method utilizes the raw image projection data and zero gradient spectrum to account for EPR line shape and hyperfine structure of any given paramagnetic probe without the need for deconvolution that is poorly suited for high noise data. Using fast scan EPR imaging with this reconstruction method, we image non-deuterated, deuterated and ¹⁵N substituted nitroxide probes in experimental phantoms of complex geometries. We evaluate the image resolution that can be obtained and the imaging time required. With 16,384 projections acquired over 1 min, and a field gradient of 8 G/cm, with a 250³ voxel 3D matrix, spatial resolutions of up to 100 µm are theoretically possible for a cubical volume of 25 × 25 × 25 mm³. In experiments with a variety of phantoms with mM nitroxide radical probes, resolutions of 600–250 µm were obtained with 1–10 min acquisitions, respectively. The presently obtainable signal sensitivity and noise levels of these acquisitions limited the obtainable resolution. With longer time acquisitions or further improvements in sensitivity and noise reduction, image resolutions approaching 100 µm should be possible.

氮氧化物探针常用于生物医学 EPR 成像应用；然而，图像质量受到图像投影数量以及探针线宽和超精细结构的限制。在当前的研究中，我们评估了快速毫秒扫描 EPR 投影采集的使用以及针对从大量噪声投影中重建 3D 空间 EPR 图像而优化的新型重建算法。这种重建方法利用原始图像投影数据和零梯度谱来解释任何给定顺磁探针的 EPR 线形状和超精细结构，而无需进行不适合高噪声数据的去卷积。使用具有这种重建方法的快速扫描 EPR 成像，我们对非氘化、氘化和¹⁵复杂几何形状的实验体模中的 N 取代氮氧化物探针。我们评估可以获得的图像分辨率和所需的成像时间。通过在 1 分钟内获得 16,384 个投影，以及 8 G/cm 的场梯度和 250 ³体素 3D 矩阵，对于 25 × 25 × 25 mm ³的立方体积，理论上可以实现高达 100 µm 的空间分辨率. 在使用 mM 氮氧自由基探针对各种体模进行的实验中，分别通过 1-10 分钟的采集获得了 600-250 µm 的分辨率。目前可获得的这些采集的信号灵敏度和噪声水平限制了可获得的分辨率。随着更长时间的采集或灵敏度和降噪的进一步改进，接近 100 µm 的图像分辨率应该是可能的。