<p>Scanning trajectories are essential and important components of trajectory-based scanning and imaging systems such as laser scanning confocal microscopes, atomic force microscopes (AFM), laser scanning systems for aerial surveying (LiDAR), micro-electromechanical systems (MEMS), medical imaging and 2D/3D printing. Previous study has demonstrated that Sinusoidal Lissajous is the optimal scanning trajectory and proposed that increasing the scanning repetition could further enhance image quality. However, it is challenging to pinpoint the essential elements needed to enhance the quality of the reconstructed images since there is currently no comprehensive analysis of how the scanning trajectory resolution and scanning time affect the reconstructed image quality. The purpose of this work is to look into the influence of scanning resolution and scanning time on the quality of magnetic particle imaging (MPI)-reconstructed images across a variety of scanning trajectories. This study offers a comprehensive analysis of how scanning repetition and scanning time affect the reconstructed images’ quality at the individual pixel in the field of view (FOV) as well as throughout the entire FOV. The impacts of the scanning time were investigated both before and after image reconstruction. In the pre-reconstruction phase, the minimum and maximum distances to the closest neighboring points in the FOV and their distribution in different regions of the FOV were analyzed in order to investigate the density, homogeneity, and time spent on each pixel. Here, we demonstrated that the image resolution for any scanning trajectory is scale invariance, meaning that for a fixed frequency ratio <InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="41598_2025_89296_Article_IEq1.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="25" /> </InlineMediaObject> <EquationSource Format="TEX">\(\:{N}_{P}\)</EquationSource> </InlineEquation>, the ratio of pixel size to image FOV size remains constant. Peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), normalized root mean square error (NRMSE), and normalized sum of squared error (NSSE) are used to evaluate and compare the reconstructed images’ quality in the post-reconstruction stage. It is found that the Sinusoidal Lissajous scanning trajectory is the best in terms of accuracy, structural similarity, and signal-to-noise. We showed that only the image quality for bidirectional Cartesian and sinusoidal Lissajous trajectories are sensitive to scanning time, particularly at the pixel scale. Furthermore, contrary to expectations, we discovered that optimizing the scanning repetition did not enhance the MPI-reconstructed image quality. Nevertheless, the image quality would be enhanced by extending the scanning duration through decreasing the scanning frequency. The optimal scanning frequency is half of the frequently used 25 kHz for the chosen <InlineEquation ID="IEq2"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="41598_2025_89296_Article_IEq2.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="25" /> </InlineMediaObject> <EquationSource Format="TEX">\(\:{N}_{P}\)</EquationSource> </InlineEquation> ratio of 100.</p>

A comprehensive analysis of the impacts of Image Resolution and Scanning Times on the quality of MPI-reconstructed images

期刊论文
出版模式：

开放获取
发表日期：

2025年2月14日

查看全文

Ton Duc Do,
Azamat Mukhatov,
Aizhan Tolebay,
Tuan-Anh Le,
Tri T. Pham

摘要

Scanning trajectories are essential and important components of trajectory-based scanning and imaging systems such as laser scanning confocal microscopes, atomic force microscopes (AFM), laser scanning systems for aerial surveying (LiDAR), micro-electromechanical systems (MEMS), medical imaging and 2D/3D printing. Previous study has demonstrated that Sinusoidal Lissajous is the optimal scanning trajectory and proposed that increasing the scanning repetition could further enhance image quality. However, it is challenging to pinpoint the essential elements needed to enhance the quality of the reconstructed images since there is currently no comprehensive analysis of how the scanning trajectory resolution and scanning time affect the reconstructed image quality. The purpose of this work is to look into the influence of scanning resolution and scanning time on the quality of magnetic particle imaging (MPI)-reconstructed images across a variety of scanning trajectories. This study offers a comprehensive analysis of how scanning repetition and scanning time affect the reconstructed images’ quality at the individual pixel in the field of view (FOV) as well as throughout the entire FOV. The impacts of the scanning time were investigated both before and after image reconstruction. In the pre-reconstruction phase, the minimum and maximum distances to the closest neighboring points in the FOV and their distribution in different regions of the FOV were analyzed in order to investigate the density, homogeneity, and time spent on each pixel. Here, we demonstrated that the image resolution for any scanning trajectory is scale invariance, meaning that for a fixed frequency ratio \(\:{N}_{P}\) , the ratio of pixel size to image FOV size remains constant. Peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), normalized root mean square error (NRMSE), and normalized sum of squared error (NSSE) are used to evaluate and compare the reconstructed images’ quality in the post-reconstruction stage. It is found that the Sinusoidal Lissajous scanning trajectory is the best in terms of accuracy, structural similarity, and signal-to-noise. We showed that only the image quality for bidirectional Cartesian and sinusoidal Lissajous trajectories are sensitive to scanning time, particularly at the pixel scale. Furthermore, contrary to expectations, we discovered that optimizing the scanning repetition did not enhance the MPI-reconstructed image quality. Nevertheless, the image quality would be enhanced by extending the scanning duration through decreasing the scanning frequency. The optimal scanning frequency is half of the frequently used 25 kHz for the chosen \(\:{N}_{P}\) ratio of 100.