Figure 4.

(a) High-throughput magnetic particle spectrometry is enabled by the lack of tuned circuit elements in the AWR. In a single automated acquisition (500 ms total time), we discretely sample 100 drive frequencies from 16 kHz to 115 kHz at 25 mT. The data from this single acquisition is shown as a stack plot of Resovist fourier spectra for 100 discrete drive frequencies (after background correction and removal of out-of-band signal). (b) Analysis of the spectra in Fig. 3a shows the expected steeper slope of harmonic decay with higher drive frequencies. This matches the findings of prior work²⁹. A steeper slope implies a poorer modulation transfer function response leading to poorer spatial resolution. (c) From a single automated acquisition, 33 unique sets of drive parameters were tested (n = 3) on 125 μg of Resovist. The optimal drive waveform (gray arrow) with best resolution is with a 17 kHz, 8.5 mT amplitude waveform. The FWHM resolution (mm) assumes a 3.5 T/m gradient. (d) The same dataset from part c is plotted for both spatial resolution and signal strength. While the lowest amplitude and frequency (gray arrow) gives the best spatial resolution, this is at a significant cost of almost 10-fold lower signal strength which has implications for MPI sensitivity. The high-throughput and denser sampling of frequency uniquely allows the AWR to better optimize for both spatial resolution and signal strength. This reveals the 100 kHz, 8.5 mT amplitude waveform (black arrow) which shows almost as good resolution improvement as the gray arrow while having no loss in signal strength, therefore having better overall MPI performance than the waveform obtained from simply optimizing for one parameter.