| 生理学研究所年報 第27巻 | |
| 研究会報告 |   |
|
|
|
26．神経科学の道具としてのfMRI研究会2005年11月24日−11月25日
Ken'ichi Ueno, R. Allen Waggoner, Keiji
Tanaka, Kang Cheng
(Laboratory for Cognitive Brain Mapping,RIKEN Brain Science Institute)
Pei Sun, Justin Gardner, Mauro Costagli, Kenichi Ueno, R.Allen Waggoner, Keiji Tanaka, Kang Cheng
(Lab. for Cognitive Brain Mapping, RIKEN Brain Science Institute, Wako-shi, Japan)
Manabu Honda1,2,3
(1 Department of Cortical Function Disorders,National Institute of Neuroscience,National Center of Neurology and Psychiatry ; 2Laboratory of Cerebral Integration,National Institute for Physiological Sciences; 3 SORST,JST)
Masaya Misaki, Satoru
Miyauchi, Takashi Abe, and Shigeyuki Kan
(Brain Information Group,National Institute of Information and Communications Technology) 【参加者名】 【概要】 The meeting attracted more than 40 attendees from universities and research institutions. Ten speakers (with one cancellation due to family emergency) presented their recent works on methodological aspects of fMRI and related techniques (MEG and EEG), including parallel imaging, electro-vascular coupling, fMRI-constrained source localization for MEG and EEG measuremnts, and novel stimulation paradigms and analysis methods for fMRI studies. Dr. Matthew Nielson and Dr. Allen R. Waggoner presented their works on parallel imaging. Dr. Nielson examined the effects of parallel acquisition methods on EPI quality and time-series statistics and concluded that due to reduced echo train length using multiple receiving coils, parallel imaging in general helps in reducing geometric distortion. Dr. Waggoner compared two types of parallel imaging reconstruction methods, i.e., SENSE and TSENSE and by making systematic comparisons using data acquired from both auditory and visual experiments concluded that TSENSE is especially useful in the presence of large motions. Simultaneous measurements of EEG and fMRI continued to attract attentions at the meeting. Dr. Jorge Rieta addressed the issue related to the nonlinear electro-vascular coupling, and presented a detailed biophysical model of how electrical and hemodynamic brain signals are generated within a basic cortical unit. The model explicitly addressed several aspects in relation to EEG and fMRI data fusion. Dr. Xiaohong Wan presented their recent works on simultaneous EEG and fMRI measurements both in spontaneous and event-related activities, and outlined a model interpreting the physiological meaning of fMRI by correlating them with the concurrently recorded stimulus-evoked EEG signals and those recorded during the resting state. In a theoretical study, Dr. Yamashita presented a framework for localizing MEG sources based on fMRI activation patterns. The inherent limitation of BOLD fMRI’s spatial resolution has not been successfully addressed. Dr. Ueno presented their recent works investigating point-spread functions (PSF) of BOLD signal in the tissue region of human V1 using spatially localized and size-varied stimuli. A subsequent modelling study indicated that with the measured PSF of ~1.8 mm, BOLD fMRI can be used for resolving human ocular dominance columns with measure ~1.0mm in width, as has been shown previously. Dr. Sun further presented their recent high-resolution works with a novel stimulation paradigm, showing that the tuning to stimulus orientation in human V1 could be directly demonstrated using BOLD fMRI. Some interesting activation results were also presented at the meeting. Dr. Goda presented some detailed retinotopic mapping results in higher order visual areas and clearly showed the existence of multiple retinotopic representations around areas V4, V3A and MT. Dr. Tanabe presented a method for evaluating learning effects within a stimulus-related brain activity when the subjects performed a paired-association task. Finally, using a parametric analysis method, Dr. Misaki presented their recent data regarding the spatial position representation in the human occipito-parieto-frontal regions. In summary, this annual meeting has provided a useful forum for young investigators in the field of functional imaging in Japan to exchange and discuss new ideas. It is strongly felt that the meeting should be continued and a broader range of topics, including neurobiology-related issues, to be covered in the future meetings.
（1） Parallel Acquisition Methods in Functional MRIMatthew Nielsen, Osamu Takizawa(Siemens-Asahi Medical Technologies Ltd., Tokyo,
Japan)
We examine the effects of parallel acquisition methods on echo-planar image quality and time-series statistics as typically used in fMRI. In EPI, one train of RF echoes encodes one 2D image. Without parallel acquisition, the required echo train length is equal to the number of pixels along the phase-encoding axis of the image, and within one echo train, the amplitude of each echo decreases exponentially with time constant T2*. To be maximally sensitive to the BOLD effect while having a high SNR, each echo should occur slightly earlier than T2*. However, images are at least 64x64 pixels, and the time needed to form and sample 64 echoes is typically 30-60ms while T2* in human cerebral cortex is 70-80ms in a homogeneous 1.5T field. Such a long acquisition window relative to T2* means (1) that most of the echoes cannot occur at the optimum point in time, and (2) that k-space is sampled with non-uniform sensitivity causing image blurring. In addition, systematic phase errors accumulated with each echo cause distortion and "N/2 ghost" artifacts, making a shorter acquisition desirable. Parallel acquisition methods use multiple RF coils with different, overlapping sensitivity profiles to simultaneously receive multiple samples of each echo. Information about the sensitivity profiles is used to combine data from all coils and thereby reduce the number of echoes required to encode an image. In anatomical images, such methods cause a tolerable reduction in SNR, but this reduction does not immediately apply to fMRI.
（2） SENSE or TSENSE, Which is best for fMRI?R. Allen Waggoner and Mauro Castagli (RIKEN -
Brain Science Institute) Partially Parallel Imaging techniques are having dramatic effects on many areas of MRI. But there are many different methods of acquisition and reconstruction, which works best for fMRI? We compare the traditional SENSE method with TSENSE, a method originally developed for minimizing the motion reconstruction errors in cardiac imaging. Motion effects are also a problem for fMRI, but the magnitude of the motions are not as large as in the case of cardiac imaging. To determine if TSENSE is also beneficial for fMRI, we have collected both SENSE and TSENSE data sets, with various levels of acceleration, during the same experimental session. Both visual and auditory paradigms were used to evaluate the relative performance of SENSE versus TSENSE.
（3） Nonlinear Local Electro-Vascular CouplingRiera JJ, Wan X, Jimenez JC, Kawashima R (Tohoku
University)
In this paper, we present a detailed biophysical model of how electrical and hemodynamic brain signals are generated within a basic cortical unit. The model is obtained from coupling a canonical neuronal mass and an expandable vasculature. In this proposal, we explicitly address several aspects related to electroencephalographic (EEG) and functional magnetic resonance imaging (fMRI) data fusion: a) the impact of the cerebral architecture (at different physical levels) on the observations; b) the physiology involved in electro-vascular coupling; and c) energetic considerations to gain a better understanding of how the glucose budget is used during neuronal activity. The model has three components. The first is the neural mass model of three neuronal sub-populations that responds to incoming excitatory synaptic currents. The second and third components model the generation of measured electrical and hemodynamic signals respectively. In the first part, we describe, in some detail, the biophysical model and establish its face validity using simulations of visually evoked responses under different luminous contrasts and flicker frequencies. Finally, we will use exactly the same model as a forward or observation model to estimate underlying biophysical parameters using real EEG and fMRI data.
（4） Spatial precision of BOLD fMRI: a combined point spread function and modeling studyKen'ichi Ueno, R. Allen Waggoner, Keiji
Tanaka, Kang Cheng The inherent limitation of BOLD fMRI’s spatial resolution has not been sufficiently addressed in relation to the field strength, tissue-vessel compartments and sensory stimuli employed. In this study conducted on a Varian 4 Tesla system, we investigated point spread functions of BOLD signal in the tissue region of human V1 using spatially localized and size-varied stimuli (4°, 2°, 1°, 0.5°, 0.25°and 0.125° in visual angle). For each size, a block of 16s stimulus-on period, in which the stimulus was flashed at 4Hz, and a block of 16s blank period were repeated 14 times. The activation amplitudes for all voxels within the ROI defined by 4° stimulus, calculated as % signal changes between stimulus and blank periods, were then plotted against the geometric distance from the activation center. We found that the activation size decreased gradually as the stimulus size decreased, and the amplitude of the % signal change at the activation center also scaled with the stimulus size. From stimulus size 2° to the smallest size (0.125°), the estimated FWHM of BOLD signal decreased monotonically. The linear fit between FWHMs and stimulus sizes resulted in an intercept of ~1.8mm, which was taken as the estimated point spread. A modeling study in the context of mapping functional architectures using this point spread function was conducted.
（5） Direct demonstration of tuning to stimulus orientation in human V1: a high-resolution fMRI study with a novel stimulation paradigmPei Sun, Justin Gardner, Mauro Costagli,
Kenichi Ueno, R.Allen Waggoner, Keiji Tanaka, Kang Cheng Although the preference for stimulus orientations in human visual cortex has been inferred indirectly in a few studies using fMRI, tuning to particular stimulus orientations has not been directly demonstrated using this technique. In an effort towards revealing orientation selectivity and its spatial arrangement in human V1, we have conducted an fMRI study with a novel stimulation paradigm and a differential mapping method. High-resolution (0.625x0.625x3mm) fMRI images were acquired using a 4-segment EPI pulse sequence with a 3-inch surface coil. Two slices (volume TR=2s) were prescribed to cover individual subject’s V1 in the banks of the calcarine sulcus. During the experiment, subjects viewed epochs of 36s full-field square-wave gratings sandwiched between 24s blank displays, and performed a color-change detection task on the central fixation point. Each grating rotated continuously in a clockwise direction for 270 deg (15 deg per TR). The gratings started rotation at either 0 or 90 deg, thereby creating two orientation conditions that are orthogonal to each other continuously, each of which was repeated 12 times. After voxels that responded to these grating stimuli were identified, we performed a voxel-by-voxel analysis and calculated the differential response (difference of averaged estimated hemodynamic response functions to the two orthogonal gratings) for the middle 180 deg. We found that responses of the majority of activated voxels were modulated by the grating’s orientation and individual voxels were sharply tuned to particular orientations. Our results, therefore, provide the first demonstration that orientation selectivity in humans can be directly studied using fMRI.
（6） Retinotopy analysis of higher order visual areas in humansNaokazu Goda（生理学研究所感覚認知情報） 視覚皮質は複数の視覚野から構成されている。このうち，いくつかの視覚野の位置と境界はレチノトピーに基づいて定めることが可能である。このため，fMRI研究において視覚野間の機能比較の際にレチノトピーに基づく視覚野定義がよく利用される。しかしながら，ヒトのV3野より高次の視覚野に関してはレチノトピーについてまだ不明な点も多く，それらの位置と境界をfMRIを用いて定めることは容易ではない。この理由には，高次視覚野ではそもそも明瞭なレチノトピーが存在しないこと，使用する視覚刺激が高次視覚野の賦活に適していないこと，レチノトピー及び視覚野地図がヒトとサルの間で大きく違っているためにサルについての電気生理学的知見を援用できないことなどが考えられる。本発表では，現在行っているヒト高次視覚野領域（特にV4野やV3A野の周辺部）のレチノトピー解析について報告し，測定・解析手法に関する諸問題について考察する。
（7） A method for evaluation of learning effects within a stimulus-related brain activity during paired-association taskHiroki C. Tanabe（生理学研究所心理生理学） To clarify the distribution of the neural substrates and their dynamics during learning, we conducted functional MRI with paired-association learning. We developed a method called whole-brain cross-trial regression analysis to evaluate temporal changes of brain activity. In this seminar, I’ll talk on this method in detail, and then show the data applied with this. In the experiment, subjects had to find pre-defined pairs in a trial and error manner. Each trial consisted of the successive presentation of a pair of stimuli (S1 and S2) with a fixed interval of 1550 ms. Feedback was given by picture whether the response was correct or incorrect to the subjects in each trial. We applied a whole-brain cross-trial regression analysis to investigate the learning effect in the stimulus- or feedback-related neural activity. Briefly, we at first calculated the linear trend (slope) from the 1st to the last trial at each scan point using the least squares method in each and every voxel for each individual. Group inference was then conducted by a one sample t-test using the average of the delay period (5 to 8th scan point) and feedback period (12 to 14th scan point) summary images of each subject, which should represent the recall process for the paired associates and building association, respectively. The results indicate that representation of paired associates might be accomplished by unimodal sensory, polymodal associate, and memory-related areas. The superior temporal sulcus is seemed to be involved in building paired association.
（8） MEG/fMRI spatio-temporal source localization methodOkito Yamashita, Masaaki Sato, Taku Yoshioka To clarify the brain function, it is important to know how a certain task is processed by several brain regions involved. MEG is one of the key tools to answer this question due to its high temporal resolution. MEG, however, does not have enough spatial resolution by itself, therefore the spatial information obtained by the other modalities such as fMRI should be combined. In this talk I will present a source localization method of MEG based on fMRI activities, which is now developing in ATR.
（9） Spontaneous and event related EEG correlates of fMRIXiaohong Wan, Jorge Riera, Ryuta Kawashima In this presentation, we will talk about our works in simultaneous EEG and fMRI recording, both in spontaneous and event-related activities. Currently, simultaneous EEG/fMRI technique is believed to be most promising to explore the spatiotemporal human brain functions, due to their noninvasive feature. However, there are several technical and theoretical issues to be solved prior to implement this advanced technique into cognitive/psychological neurosciences. Firstly, how to get the clean EEG signals from the noisy data during fMRI scanning? Secondly, how to fuse the two different modalities in one framework? In theoretical consideration, combination of the electrical signals of EEG with the hemodynamic responses of BOLD signals definitely enhances us to understand the nature of functional imaging in terms of physiological basis, because the functional imaging signals are not a direct measure of neural activity, rather, are related to the cerebral energy consumption and blood flow. In outline, the first part of our presentation will be focused on the technical issues of data analysis, and the later part will be related to the physiological meaning of fMRI signals implicated by correlation with the concurrent EEG signals in stimulus evoked and resting conditions.
（10） Neural substrates of spontaneous electroencephalographic activity related to the emergence of the hypersonic effectManabu Honda1,2,3 Sounds containing an inaudible high-frequency component (HFC) with non-stationary fluctuation activate deep-lying brain structures and evoke physiological, psychological, and behavioural responses (hypersonic effect). To evaluate the emergence of the hypersonic effect using the electroencephalogram (EEG), we analyzed the regional cerebral blood flow (rCBF) and spontaneous EEG simultaneously measured. First, to examine the whole neuronal network involved in the emergence of the hypersonic effect, principal component analysis was applied to the rCBF data measured from subjects who were exposed to full-range sound including both HFC and low-frequency component (LFC) and LFC alone. The activity of the first principal component of rCBF data including the upper brainstem (midbrain), hypothalamus, thalamus, precuneus, prefrontal cortex, and anterior cingulate gyrus was significantly correlated with the alpha2 component recorded from 7 electrodes in the central and parieto-occipital regions. Vice versa, the averaged alpha2 potential obtained from the above 7 electrodes was significantly correlated with the rCBF of deep-lying brain structures. These results show that the alpha2 potential of spontaneous EEG measured from central and parieto-occipital regions is a plausible index of the deep-lying brain activity reflecting the emergence of the hypersonic effect.
（11） Spatial position representation in the human brain revealed by parametric analysisMasaya Misaki, Satoru Miyauchi, Takashi Abe,
and Shigeyuki Kan Most fMRI studies to date have linked a predefined function to a specific brain region. However how the brain processes the information had not been defined so much from fMRI activation data. To determine how information is processed in a specific region of the brain, we explored which region was activated and how the task parameters modulated the activation. In the experiment, which had an event-related design, we explored the spatial position representation in the human brain while the subject tracked a moving point with their eyes. The fMRI data was analyzed using parametric modulation. We used a non-linear regression analysis using an artificial neural network in which the target spatial position was the regression parameters. Viewpoint- and head-centered spatial positions were used to define the regression parameters. The analyses revealed that the activations of the visual cortices and the dorsal stream areas were systematically modulated depending on the spatial positions of the target. The correspondence between the target position and the activated region was most distinct in the visual cortices, and this correspondence became less distinct from the parietal regions to the premotor regions. Viewpoint- and head-centered representations were intermixed in each region, and it is noticed that head-centered modulations were observed even in the primary visual cortex.
|
||||
