The Peculiar Effect of Scanning Trajectory Pattern Density on MPI-Reconstructed Image Quality

Abstract

Scanning trajectory is a key component of many point scanning systems, including light detection and ranging (LiDAR), laser scanning confocal microscopes (LSCM), and Micro Electro Mechanical System (MEMS) scanners. This study investigates how different scanning trajectories affect the quality of MPI-reconstructed images, with an emphasis on scanning parameters, particularly scanning duration, which includes both scanning repeat and scanning time. The impacts of sampling frequency, scanning repeat, and scanning time on the quality of MPI-reconstructed images were examined across a wide range of scanning trajectories and pattern densities. Peak Signal to Noise Ratio (PSNR), Structural Similarity Index Measure (SSIM), Normalized Root Mean Square Error (NRMSE), and Normalized Sum of Squared Errors (NSSE) were the four metrics used to assess the quality of MPI-reconstructed images. The optimal sampling frequency is 6 × 106 Hz or higher, as evidenced by the convergence of all four measures to the same values. At low scanning density ( Np < 80 ), increasing scanning repeats or times does not improve image quality. However, increasing scanning repeats or scanning times would help enhance image quality when employing a high density scan with Np ≥ 80 . This study illustrates the peculiar effect of different scanning trajectories on the quality of MPI-reconstructed images at low and high scanning densities. Our simulations reveal how trajectory density and scan time impact image quality and reconstruction accuracy, providing a foundation for how long to scan or how many repetitions to rescan the sample to improve image quality.