Daisuke Kokuryo1, Chika Sato2, Takashi Itahashi3, Shigeyoshi Saito4, Hiroyuki Ueda1, Etsuko Kumamoto1,5, Toshiya Kaihara1, Nobutada Fujii1, Noriaki Yahata2,6, and Ichio Aoki2,6
1Graduate School of System Informatics, Kobe University, Kobe, Japan, 2National Institute of Radiological Sciences, National Institutes for Quantum and Radiological Science and Technology, Chiba, Japan, 3Medical Institute of Developmental Disabilities Research, Showa University, Tokyo, Japan, 4Graduate School of Medicine, Osaka University, Suita, Japan, 5Information Science and Technology Center, Kobe University, Kobe, Japan, 6Institute for Quantum Life Science, National Institutes for Quantum and Radiological Science and Technology, Chiba, Japan
Synopsis
Establishing
data harmonization in fMRI “big data” is necessary to ensure the reliability
and reproducibility of fMRI research. Our group recently developed two distinct
phantoms: a brain-shaped phantom to correct field heterogeneity and a functional
image-specific phantom to calibrate shape distortion and signal irregularity. A
preliminary experiment confirmed a shape distortion and signal irregularity caused
by field inhomogeneity when using EPI signals, and these were corrected using
an affine transformation. Therefore, our developed phantoms have the potential
to contribute to achieving fMRI data standardization.
Introduction
Functional
MRI (fMRI) is a major research tool for investigating the neurophysiological status
of the brain in both healthy and diseased conditions. In clinical and
pre-clinical fMRI research, a great deal of fMRI data are required to achieve reliability
and reproducibility, and it is necessary to acquire fMRI data from multiple
sites to collect a large amount of data from many volunteers, patients and
animal models. However,
no method has yet been established to harmonize the fMRI data acquired from
multiple imaging sites and scanners, which are characterized by shape
distortion and signal irregularity caused by field inhomogeneity and living bodies,
such as human and animals, when using an EPI sequence and by differing imaging
protocols and parameters. In clinical
research, some groups have used travelling human volunteers and patients to standardize
and harmonize studies of fMRI data that were acquired from multiple imaging1-4.
By contrast, our group focuses on developing special phantoms and compensation
algorithms to standardize
and harmonize fMRI data acquired from multiple imaging sites. Herein,
we describe the development of specific phantoms to correct the shape
distortion and signal irregularity caused by field inhomogeneity and by living
body when using an EPI sequence.Methods
Phantom
development: Two types
of distinct phantoms were developed for fMRI data harmonization: a brain-shaped
phantom to correct field heterogeneity and a functional image-specific phantom to
calibrate the shape distortion and signal irregularity caused by living bodies
when using an EPI sequence (Fig. 1). The brain-shaped phantom aims to calibrate and
correct the shape distortion and signal irregularity caused by field heterogeneity
of each scanner, and the functional image-specific phantom aims to compensate for
the shape distortion and signal irregularity caused by the living subject (whether
animal or human) when using an EPI sequence. In the brain-shaped phantom, the size
and shape of the phantom were designed by referring to the size and shape of
the target, such as marmoset and mouse, and the phantom was constructed of
acrylic resin by a 3D printer. The size and shape of the phantom were adjusted
by the size of the coils. A hollow area was
created in the brain-shaped phantom, and a copper sulfate aqueous solution was introduced
through inlets to evaluate and correct field inhomogeneity (Fig. 1a). To detect
the image position, grooves were made on the outer side of the hollow area as
position markers (Fig. 1b). In the functional image-specific phantom, the
multiple tube-type phantoms with distilled water were used to compensate for
the shape distortion and signal irregularity caused by the living subject, such
as a marmoset or mouse, because these phantoms can be measured simultaneously with
the target living body in fMRI experiments. The phantoms can be installed around
the brain-shaped phantom and the target living body5. When functional
image-specific phantoms were used with the living body, the positions of the
phantoms were used to correct the shape distortion caused by the living body when
using an EPI sequence, and the phantoms’ signals were used to compensate for the
signal irregularity.
Correction
of shape distortion: As a
preliminary trial for the standardization of fMRI data, an affine transformation
compensated for the shape distortion of fMRI data caused by the field
inhomogeneity. In this proceeding, the MR images acquired using an EPI sequence
were set as fMRI data and spin-echo T2-weighted images were set as standard
images. The evaluation criterion for transformation was normalized mutual
information.
MR
experiments: To
evaluate the image quality and preliminary registration using the affine
transformation, spin-echo T2-weighted images as a standard image and EPI images
as a functional image were acquired using a 7.0 Tesla preclinical MR scanner (Bruker
BioSpec 70/20, Bruker BioSpin, Ettlingen, Germany) and an 8-channel phased
array coil. The
imaging parameters were as follows for the spin-echo T2-weighted images: TR/TE
= 2975/33 ms; FOV = 76.8 x 76.8 mm2; matrix size = 128 x 128; slice thickness =
1.0 mm; number of slices = 27; number of acquisitions = 1. The imaging
parameters for the EPI images were TR/TE = 2000/18 ms; FOV = 41.4 x 41.4 mm2; matrix
size = 69 x 69; slice thickness = 1.0 mm; number of slices = 27; number of
acquisitions = 1.Resutls and discussion
Figure
2 presents
the MR images acquired
using spin-echo and EPI sequences. The EPI image had a serious shape distortion
and signal irregularity in comparison to the spin-echo T2-weighted image. This
result clearly indicates that the EPI images are easily affected by field
inhomogeneity, demanding a correction procedure. Figure 3 shows the changes to
the values of the normalized mutual information before and after registration.
The values of the normalized mutual information after registration were always
higher than those before registration. The example images after registration, as
shown Figure 4, were more similar to the images before registration. Thus, the
correction using an affine transformation effectively corrected the influence
of field inhomogeneity. In future, we will improve the registration technique
for correcting signal disparities caused by field inhomogeneity and living bodies
using an EPI sequence.Conclusion
Our
developed phantoms may potentially contribute to the standardization of fMRI
data.Acknowledgements
This research was supported by AMED under Grant Number JP19dm0307026.References
[1] A. Yamashita, et al: PLoS Biology, 17(4):
e3000042, 2019.
[2] C. Hawco, et al: Psychiatry Res Neuroimaging, 282, 134-142,
2018.
[3] O. Mogan, et al: ISMRM 2019, 237, 2019.
[4] W. T. Clarke, et al:
ISMRM 2019, 2812, 2019.
[5] D. Kokuryo, et al: Phys. Med. Biol., 55 (14):
4119-4130, 2010.