(Requires cookies for changing the display language.)

The main advantage of blood oxygenation level dependent (BOLD) contrast functional magnetic resonance imaging (fMRI) over its predecessor, H2O15 positron emission tomography (PET), is its capability of performing multi-state contrast analysis of a single subject during a single session. Therefore, the experimental paradigm should ideally consist of multiple independent variables. We had shown that, under certain conditions where hemodynamic confounding effects among multiple behavioral states of interest are negligible, multi-state contrast experiments could be readily designed. Such experiments, as exemplified by sequential epoch analysis (SEA) applied to motor control analysis, were effective in identifying the area involved in inhibition of the opposite hand during exclusive usage of a single hand.[1] However, there exist some cases where undefined or unpredictable confounding effects among these variables degrade the performance of SEA. In other words, SEA in its original form has significant limitations to allow for broader range of applications.

Fig.1 Concept of ICS Analysis
Fig.1 Concept of ICS Analysis

Brain activation studies are generally based on the concept that a brain function of interest can be assigned to discrete areas of the cerebral cortex. The concept implies that, in the process of fMRI time series analysis, it is acceptable to consider that areas of activation associated with a given behavioral task are ‘spatially independent.’ This spatial independence represents the principle condition for application of independent component analysis (ICA), a novel higher order statistical method, to fMRI studies. McKeown et al originally showed that transposing the data matrix can indeed realize such spatial-mode ICA.[2,3] Accordingly, we introduced a new method of high-field fMRI time series analysis, which we term independent component - cross correlation - sequential epoch (ICS) analysis[4,5], as a hybrid technique of ICA and a general linear model. The concept of ICS analysis is schematically explained in Fig.1. In ICS analysis, a sequential epoch paradigm is composed to embed the function which correlates to the target multiple behavioral hypothesis. Although any function can be embedded within a sequential epoch paradigm, a box car function is the simplest and common choice. FMRI time series are then subjected to blind separation into independent components by ICA. Subsequently, cross correlation analysis is performed utilizing each embedded function, fi(t), as model function. As a result, fMRI images which represent behavioral correlates given by the chosen function can be obtained as an activation map, the time series of which show significant correlation. Figure 2 shows a representative result of ICS analysis and schematic representation of the anatomic organization of the human primary somatosensory cortex (SMI).

A result of ICS analysis

Fig. 2 A representative result of ICS analysis obtained using hand motion tasks

CS: central sulcus; MI-4a: 4 anterior area of the primary motor cortex; MI-4p: 4 posterior area of the primary motor cortex; 3a: 3a area of the primary sensory cortex; SI, the ‘classical’ primary sensory cortex (Brodmann areas 1, 2 and 3b). To provide better visualization of the three-dimensional (3D) relationship, part of the activation map of the superior slice is placed over its neighboring slice directly below. The schematic gray block (right) shows the 3D spatial position of the SMI and the localization of the independent areas of activation. Should pixel sharing of two independent components occur, the activation located posteriorly within the slice is mapped over that which is located anteriorly. MI-4a and MI-4p are located in the pre-central gyrus, while 3a and SI are located posterior of the central sulcus within the post-central gyrus

With ICS analysis, we succeeded in direct identification of dual representation of the primary motor cortex due to clearly distinguishable four independent subunits. It is noteworthy that the resolution is comparable to that of primate electrophysiology.

Following this success, we have so far applied ICS analysis in various fMRI studies and made important discoveries about human brain function: 1) Asymmetry of parietal lobe activation during piano performance.[6] When musically trained subjects played piano with visually presenting musical scores, activation in the posterior parietal cortex (PPC) was investigated. Quantitative analysis of PPC activation revealed that the number of components in the left PPC was significantly greater than that in the right, regardless of playing with left or right hand. 2) The primary motor area (MI) for diaphragmatic motion.[7] ICS analysis successfully extracted physiological meaningful components with negation of significant confounding effects, using voluntary diaphragmatic motion, and hand-grasping task as an internal control. As a result, the cortical area for diaphragmatic motion were observed in the bilateral primary motor areas anterolateral and anterior to the region activated by the hand-grasping motion in MI. 3) Human primary motor cortex shows hemispheric specialization for speech.[8] We investigated the spatial distribution of neural activities associated with phonation (MIp), silent tongue motion (MIt), and vocalization (MIv) within MI. While MIt showed no significant differences between the two hemispheres, MIp and MIv exhibited significant hemispheric differences. The study demonstrated direct evidence that human MI possesses clearcut hemispheric specialization (Fig.3).

Hemispheric specialization for speech revealed using ICS analysis

Fig.3 Hemispheric specialization for speech revealed using ICS analysis

All components were mapped to MI inferior to the hand motion representational area. Note significant asymmetry in phonation (MIp) and vocalization (MIv) distribution; each forms distinctive clusters, namely inferior (MIp-i, MIv-i) and superior (MIp-s, MIv-s) in the left (dominant) hemisphere.

Because of the immensity of the data set, ICS analysis imposes heavy computational load. To process data that accumulated daily, we utilized a 64-CPU parallel computer (SGI Origin) for these analyses.


  1. Nakada, T., Fujii, Y., Suzuki K., Kwee, I.L. High field (3.0T) functional MRI sequential epoch analysis: an example for motion control analysis. Neurosci Res 32, 355-362, 1998.
  2. McKeown, M.J., Jung, T.-P., Makeig, S., Brown, G.G., Kindermann, S.S., Lee, T.-W., Sejnowski, T.J. Spatially independent activity patterns in functional MRI data during the Stroop color-naming task. Proc Natl Acad Sci USA 95, 803–810, 1998.
  3. McKeown, M.J., Makeig, S., Brown, G.G., Jung, T.-P., Kindermann, S.S., Bell, A.J., Sejnowski, T.J. Analysis of fMRI data by blind separation into independent spatial components. Hum Brain Mapp 6, 160–188, 1998.
  4. Nakada, T., Suzuki, K., Fujii, Y., Matsuzawa, H., Kwee, I.L. Independent component - cross correlation - sequential epoch (ICS) analysis of high field fMRI time series: direct visualization of dual representation of the primary motor cortex in human. Neurosci Res 37, 237–244, 2000.
  5. Suzuki, K., Kiryu, T., Nakada, T. Fast and precise independent component analysis for high field fMRI time series tailored using prior information on spatiotemporal structure. Hum Brain Mapp 15, 54–66, 2001.
  6. Itoh, K., Fujii, Y., Suzuki, K., Nakada, T. Asymmetry of parietal lobe activation during piano performance: a high field functional magnetic resonance imaging study. Neurosci Lett 309(1), 41–4, 2001.
  7. Nakayama, T., Fujii, Y., Suzuki, K., Kanazawa, I., Nakada, T. The primary motor area for voluntary diaphragmatic motion identified by high field fMRI. J Neurol 251, 730–735, 2004.
  8. Terumitsu, M., Fujii, Y., Suzuki, K., Kwee, I.L., Nakada, T. Human primary motor cortex shows hemispheric specialization for speech. Neuroreport 17(11), 1091–1095, 2006

Copyright © 2020 Center for Integrated Human Brain Science, University of Niigata.
Created this page in 0.02 seconds