**2. Pilot fMRI study on spatial and temporal attention cognition of normal, young subjects**

#### **2.1 Introduction**

38 Neuroimaging for Clinicians – Combining Research and Practice

Previous studies have suggested that both amyloid plaques and neurofibrillary tangles are clearly visible in AD brains using microscopy. The plaques and tangles spread through the cortex in a predictable pattern as AD progresses (Fig. 2). First, during the earliest AD stage, the atrophy of brain cortices occurs in the hippocampus and the surrounding areas, and the changes may begin 20 years or more before diagnosis. Second, during the mild to moderate stages, the damage to the brain cortices is found in the temporal and parietal lobes and parts of the frontal cortex and cingulate gyrus. Finally, during the advanced AD stage, the

Fig. 1. The change of cognitive function deficiency model of aging healthy individuals and

Because the damage to the brain cortices occurs in the hippocampus, the early symptoms are mild memory loss. The first symptoms are often mistaken as related to aging or stress, but these early cognitive deficits can also be symptomatic of the early stages of AD and

Several studies have attempted to identify the preclinical cognitive markers of AD. Bäckman and Small (2007) used four episodic memory tasks on old non-dementia and incident AD subjects to investigate the cognitive deficit changes over three years. These tasks consisted of

Fig. 2. The damaged cortex areas of AD patients during different stages.

degeneration of atrophy is found in the whole brain.

patients with AD.

detected using long-term cognitive tests.

The human brain is a highly efficient information processing system capable of handling a large amount of information rapidly and simultaneously. If we were able to elucidate the sophisticated mechanisms of the brain accurately, we could construct flexible, efficient, humanlike artificial systems. We would also have the capacity to assess the most relevant ways to present information and in the most appropriate manner.

Previously, we studied the human visual and auditory systems, which convey almost all external information, with an emphasis on the parallel processing of visual and auditory data. In human information processing systems, attention plays an important role in selecting and integrating information. Previous studies on attention have proposed various psychological models, which are supported by a variety of psychological and physiological evidence. The neuronal substrate of the human attention system has also been investigated using positron emission tomography (PET) and fMRI to examine visual and auditory attention in humans using audiovisual stimuli. However, the common and unique networks used by the visual and auditory attention systems remains poorly understood. Furthermore, attention to time has not been studied sufficiently compared to space, and little research has compared the differences between the visual and auditory systems regarding spatial and temporal attention.

In this study, we analysed spatial and temporal attention using both visual and auditory stimuli. To evaluate these processes behaviourally, we conducted psychological experiments where we measured the reaction times (RTs) for each task. To reveal the neuronal networks related to these attention systems, we measured the haemodynamics using fMRI.

Early Detection of Alzheimer's Disease with Cognitive Neuroscience Methods 41

Cue stimuli (shown in fig. 3) were used to direct the subjects' attention to a particular target location or onset time. The neutral cue provided neither spatial nor temporal information; the spatial (space) cue directed the subjects' attention to the left or right; and the temporal (time) cue directed their attention to a short or long stimulus onset time. The flow chart of one trial is shown in Fig. 4. The time from the end of stimulus presentation to the onset of the next stimulus was defined as the interval of the stimuli (IOS) and was either 2,200 or 3,700 ms. We recorded the RT, the time from the presentation of a stimulus to a response indicated by a reaction key. The subjects responded to a right stimulus using the middle finger of their right hand and to a left stimulus with the forefinger of their right hand. The

We used a Philips 1.5 Tesla Magnetom Vision whole-body MRI system to measure the brain activation using a head coil. The imaging area consisted of 32 functional gradient-echo planar imaging (EPI) axial slices (voxel size 3×3×4 mm3, TR=3,000 ms, TE=50 ms, FA=90°, 64×64 matrix) that were used to obtain T2\*-weighted fMRI images in the axial plane. The EPI imaged the entire cortex. For each task, we obtained 124 functional volumes. Before the EPI scan, a T2-weighted volume was acquired for anatomical alignment (TR=3,500 ms, TE=100 ms, FA=90°, 256×256 matrix, voxel size=0.75×0.75×4 mm3). The T2 image acquisition used

Reaction times were used as the behavioural data. The RT data during the fMRI experiment were analysed using repeated measures analysis of variance (ANOVA; SPSS 12.0j for Windows). For each task, 60 RTs were acquired from each subject. We used the average of the RT data for the ANOVA, except for error trials (all subjects reacted with an accuracy above 90%). Therefore, we had 16 RT data for each task. Six tasks were presented in this experiment, and we compared the visual and auditory tasks separately. Between the modalities (visual and auditory), we compared VS and AS, VT and AT, and VN and AN. For the functional images, we used MRIcro to change the DICOM files into MRIimg and MRIhdr files. In each task, the functional images of the first four volumes were not used for the data analysis. The DICOM files from the 5th through 124th scan were exported as MRIima and MRIhdr files. In addition, the DICOM files for the T2 images were exported as

The functional images were analysed using statistical parametric mapping (SPM5, Wellcome Department of Cognitive Neurology, London, UK). The functional images from each task were realigned using the first scan as a reference. The T2-weighted anatomical images were co-registered to the first scan in the functional images. Then, the co-registered T2-weighted anatomical images were normalised to standard T2 template images as defined by the Montreal Neurological Institute. Finally, these spatially normalised functional images were

Statistical analyses identified the brain areas shared by visual (VS, VT) and auditory (AS, AT) attention and the brain areas that were selectively engaged by each task. To eliminate the brain activation caused by finger motion, we told the subjects to click the reaction key ten times during every rest. As a control task, we used VN for the visual attention task and

subjects performed 60 trials under each condition.

the same slices as the functional image acquisition.

smoothed using an isotopic Gaussian kernel of 8 mm.

**2.2.3 fMRI scanning** 

**2.2.4 Data analysis** 

MRIimg and MRIhdr files.

AN for the auditory attention task.
