Professional Documents
Culture Documents
2008 Weizenecker FFL Magnetic Particle Imaging
2008 Weizenecker FFL Magnetic Particle Imaging
2008 Weizenecker FFL Magnetic Particle Imaging
This article has been downloaded from IOPscience. Please scroll down to see the full text article.
(http://iopscience.iop.org/0022-3727/41/10/105009)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 129.81.226.149
The article was downloaded on 18/07/2013 at 03:43
Abstract
This paper presents a simulation study on the use of a field free line in magnetic particle
imaging. A major improvement in the image quality is demonstrated. The reason for this
image quality improvement is discussed, and routes for the technical implementation of the
effect are sketched.
to the FFL image with only a tenth of the concentration (also higher resolution for the FFL image in the horizontal direction.
symbol
√ α in figure 2). Theoretically, one would expect a factor This is due to the higher selection field gradient in that direction
of 128 ≈ 11.3 applying the rules for the simplest MRI when using the FFL. In general, the regularization in the case
theory. Comparing these two α-images, one can observe a of FFP and FFL acts differently regarding image properties
like resolution and signal to noise. Nevertheless, it is further
possible to equalize the image impression using a spatial low
pass filter on the FFL images.
The factor of about 10 is maintained comparing the images
denoted with β. However, the low concentration images
denoted with γ do not have a comparable quality. The FFL
image has a lower resolution, and the noise background is
higher, too. So, for low concentrations, the improvement in
the sensitivity of the FFL over the FFP decreases. The reduced
number of distinguishable voxels at lower concentrations may
explain this. Due to the lower resolution, the effective number
of voxels in a line is no longer the original 128, but the number
of distinguishable image elements. The resolution of the γ
FFP image is 1.6 mm [3]. Hence, the number of effective
voxels in a line is about
√ 13. Therefore, the expected increase
in sensitivity is only 13 ≈ 3.6. This relative sensitivity is
Figure 1. Schematic setup of the simulated scanner geometry and
the path of the FFL. The ring of 32 small coils produces the rotating validated in image γ̄ (figure 2).
FFL. The two pairs of larger loops (drive field coils) move this FFL
over the field of view. The smaller rectangular coils represent the
recording coils. The drawing is true to scale, and the diameter of the 4. Conclusion and outlook
selection field coil ring is 1 m. On the upper right, the trajectory of
the FFL is illustrated. The grey rosette illustrates the vector of the The results indicate the possibility of a more effective
drive field as a function of time. At the time tx the rosette has
encoding scheme for MPI than presented so far. When
evolved to the ‘position’ x, and the homogeneous drive field B(tx )
(sketched in the centre) has the direction of the connecting line MPI was originally presented [1], the sensitivity of MPI
between the centre of the rosette and the position x. The field was related to the sensitivity of MRI only by comparing
strength is proportional to the length of this line. Superimposing the relative magnetizations. In the view of this paper, this fact
selection field and the drive field, the FFL moves along the drive can be further confirmed, as the encoding efficiency can be
field vector provided that the orientation of the FFL is always
comparable.
perpendicular to the drive field vector. Hence, the FFL scans back
and forth while rotating slowly. This is sketched for three arbitrary However, from a technical point of view, it seems to be
times, t0 , t1 and t2 . a considerable effort to realize a fast scanning with a FFL.
Figure 2. Comparison of the reconstructed images for FFP and FFL encoding. Various concentrations for 30 nm particles are shown in
images of 20 × 20 mm2 size containing 128 × 128 voxels. The acquisition time was assumed to be 40 s. The regularization parameter was
chosen by the best visual impression resulting in a signal-to-noise ratio of about 10 in every image. The images using the FFL resemble the
FFP images, the latter with a roughly 10 times higher concentration. At lower concentration (γ ), this is no longer valid. Here, the increase
of image quality is less, and therefore, an intermediate concentration FFL image is provided (labelled γ̄ ). Images to be compared are
denoted with the same Greek letter.
2
J. Phys. D: Appl. Phys. 41 (2008) 105009 J Weizenecker et al