**3.3 Results**

#### **3.3.1 Accuracy**

To investigate the differences in the accuracy of angle discrimination for different angle pairs, we calculated the mean accuracy for each pair of reference and comparison angles in different subject groups. The accuracy rate was defined as the number of correct trials divided by the total number of trials for each angle pair. As described above (Fig. 7), the accuracy increased with an increase in the difference between the reference angle and the comparison angle in this experiment. The regression analysis of the mean accuracy yielded

Each subject underwent at least 10 practice trials prior to the start of the experiment. After the training, each pair of angles was presented 10 times in a pseudorandom order. Each

In this study, the 2AFC technique was used to measure the angle discrimination threshold. Subjects were forced to make a choice of what they perceived was the larger of two angles even if they could not detect a difference. The logistic curve is the most common sigmoid curve used extensively in cognitive psychological experiments for measuring thresholds (Voisin et al., 2002a,b). Here, the accuracy data were applied to the following logistic

> <sup>1</sup> *d RA CA Accuracy <sup>e</sup>*

In this equation, *d* represents the unique degree of freedom of the logistic curve, which was adjusted to fit the accuracy data. *RA* and *CA* represent the degree values of the reference

The discrimination threshold was defined as the angle difference at an accuracy rate of 75%. Fig. 7 shows the 2AFC results of one NC subject. The discrimination threshold is indicated where the accuracy line and the 75% line (dashed line) intersect. The discrimination threshold (DT) was computed from the logistic function (2) as follows (X = 75% accuracy):

> <sup>1</sup> 1 ( ) *<sup>X</sup> DT d Ln*

The data were incorporated into logistic functions (1) and (2). The same analyses were

To calculate angle discrimination thresholds, regression analysis with logistic function was performed. Differences in the accuracy and discrimination thresholds of the three subject groups were analysed using separate one-way analysis of variance (ANOVA). The level of significance was fixed at P < 0.05. The Bonferroni test (α = 0.05) was performed to detect the difference between each subject group. Finally, to compare the sensitivity of angle discrimination accuracy and the MMSE score, a receiver operator characteristic (ROC) analysis was used. All analyses were performed using SPSS version 12.0j (SPSS, Tokyo,

To investigate the differences in the accuracy of angle discrimination for different angle pairs, we calculated the mean accuracy for each pair of reference and comparison angles in different subject groups. The accuracy rate was defined as the number of correct trials divided by the total number of trials for each angle pair. As described above (Fig. 7), the accuracy increased with an increase in the difference between the reference angle and the comparison angle in this experiment. The regression analysis of the mean accuracy yielded

*X*

1

(1)

(2)

subject completed 80 angle discrimination trials.

**3.2.4 Data processing and analysis** 

function (1) (adapted from Wu et al., 2010):

and comparison angles, respectively.

applied to all of the data in this experiment.

**3.2.5 Statistics** 

Japan).

**3.3 Results 3.3.1 Accuracy**  significant r2 values of 0.98, 0.98, and 0.67 for the NC, MCI and AD groups, respectively [NC: F(1,6) = 323.95, P < 0.001; MCI: F(1,6) = 473.76, P < 0.001; AD: F(1,6) = 12.16, P = 0.013]. The mean angle discrimination accuracy for the NC (82.1% ± 2.2%), MCI (78.6% ± 1.8%) and AD (67.9% ± 2.5%) groups is shown in Fig. 8(a). We performed a one-way ANOVA on the mean accuracy. The mean accuracy of the angle discrimination differed significantly between the three groups [F(2,34) = 8.01, P = 0.001]. A multiple comparison using the Bonferroni correction (α = 0.05) revealed that the mean accuracy of patients with AD was significantly lower than patients with MCI (P = 0.04) and the NC subjects (P = 0.001). However, the difference in accuracy between patients with MCI and NC subjects was not significant (P = 0.93).

Fig. 7. Calculation method of the angle discrimination threshold. Accuracy of one NC subject is plotted as a function of the angular difference between the comparison (62°–110°) and the reference angle (60°). The solid line represents the logistic curve for threshold calculation. The horizontal dashed line indicates accuracy at 75%. The horizontal axis value of the intersection between the 75% line and logistic curve is defined as the angle discrimination threshold. For this NC subject, the threshold is 10.7°.

To examine whether the patients with MCI and AD showed any decline in angle discrimination, we further examined the angle discrimination threshold. We performed a one-way ANOVA and a multiple comparison using the Bonferroni correction (α = 0.05) on the mean discrimination threshold. As shown in Fig. 8(b), differences in the mean discrimination thresholds among patients with AD (25.2°±4.2°) or MCI (13.8°±2.7°) and NC subjects (8.7±0.8°) were significant [F(2,34) = 9.45, P < 0.001], with a larger threshold in patients with AD compared to patients with MCI (P = 0.036) and NC (P < 0.001). In addition, the threshold in patients with MCI was also significantly larger compared to NC (P = 0.049). These results indicated that the decline in the ability to discriminate tactile angles in patients with MCI and AD was significant compared to the NC group.

Early Detection of Alzheimer's Disease with Cognitive Neuroscience Methods 51

accuracy and MMSE score of NC and MCI subjects were compared and are presented in Fig. 8. The area under the curve (AUC, is an overall summary of diagnostic accuracy) values for the angle discrimination accuracy was 0.658, and the MMSE score was 0.611. As shown in Fig. 8, we found that the AUC of the angle discrimination accuracy was higher than the

The present study demonstrated that patients with MCI or AD have an impaired ability to discriminate tactile angles compared to age-matched healthy subjects. The current study compared the performance of angle discrimination in three different subject groups (i.e., NC, MCI and AD). The results of this study indicated that there were significant group differences in the ability to discriminate tactile angles (NC > MCI > AD). Both the mean accuracy and threshold of angle discrimination of AD patients were significantly decreased compared to NC individuals and MCI patients. In contrast, although the mean threshold of angle discrimination of patients with MCI was significantly higher than NC individuals, the

All subjects were asked to passively perceive the angles moved under their right index fingers and discriminate the largest of each angle pair, which consisted of a reference angle and a comparison angle. The current angle discrimination task is a commonly used procedure of tactile passive shape discrimination. The sensory feedback, which is critical for shape discrimination by passive touch, is generated by the four mechanoreceptive afferent systems (Johnson, 2001) located in the skin. Moreover, previous studies (Goodwin et al., 1997; Johnson, 2001) have suggested that tactile shape perception can be defined as the sum of the functions of cutaneous mechanoreceptors. However, there are numerous anatomical and morphological changes that develop with age and affect the hand and fingers. The density of mechanoreceptors in the skin was decreased (Wollard, 1936; Bruce, 1980), and the conduction velocity of peripheral nerves was significantly reduced with age (Peters, 2002). A decreased touch sensitivity in elderly individuals can cause many problems (Stevens & Choo, 1996; Vega-Bermudez & Johnson, 2002), including the inability to recognise objects by touch and an impaired ability to detect an object that has come into contact with the skin. Consequently, the ability of normal, older subjects to discriminate angles will be reduced compared to normal, young subjects. For example, the mean threshold of young subjects was 3.7° in our previous angle discrimination study (Wu et al., 2010), and the mean threshold of NC subjects in the present study was 8.7°, which was more than twice the previous value. However, the results of the present study indicated that the older subjects, as well as patients with MCI and AD, were able to complete the angle discrimination task. All subjects in the current experiment were able to perceive the change in size of the angle

However, a significant deficit in angle discrimination was observed in MCI and AD patients in this study. One of the earliest symptoms of AD is impaired working memory (Baddeley et al., 1991; Bäckman & Small, 2007). In addition, previous studies have observed that patients with MCI also show impairments in memory processing compared to healthy aging subjects (Siedenberg et al., 1996; Petersen et al., 1999). In this study, all subjects were instructed to discriminate the larger of two angles by passive touch. To perform this task, the subject had to remember the composing feature of the first angle

mean accuracy of the MCI group was similar to the NC group.

MMSE score.

stimuli.

**3.4 Discussion** 

Fig. 8. Mean accuracy and discrimination threshold of MCI and AD compared to NC. (a) The mean accuracy of the three groups (NC: 82.1% ± 2.2%, MCI: 78.6% ± 1.8%, AD: 67.9% ± 2.5%). (b) The mean discrimination thresholds for the three groups are shown (NC: 8.7° ± 0.8°, MCI: 13.8° ± 2.7°, AD: 25.2° ± 4.2°). Vertical error bars represent standard error of the mean. \*P < 0.05, \*\*P < 0.001.

Fig. 9. Receiver operating characteristic (ROC) curves for the angle discrimination accuracy and MMSE score showing discrimination between the NC and MCI groups. The solid line represents the ROC curve of angle discrimination accuracy, and the dashed line represents the ROC curve of MMSE score. The area under the curve (AUC) of angle discrimination accuracy is larger than that of MMSE score.

#### **3.3.2 Comparison between angle discrimination accuracy and MMSE score**

ROC analysis is a useful tool for evaluating the performance of diagnostic tests (Mendiondo et al., 2003; Zou et al., 2007). We used a ROC analysis to compare the sensitivity of angle discrimination accuracy to that of the MMSE scores. The fundamental measures of diagnostic accuracy are sensitivity (i.e., true positive rate) and specificity (i.e., true negative rate). As described in the Subjects section, all NC and MCI patients were defined by an MMSE score of 27 or greater and the CDR score of 0 and 0.5. In contrast, all AD patients had MMSE scores between 15 and 26. Therefore, only the ROC curves for angle discrimination accuracy and MMSE score of NC and MCI subjects were compared and are presented in Fig. 8. The area under the curve (AUC, is an overall summary of diagnostic accuracy) values for the angle discrimination accuracy was 0.658, and the MMSE score was 0.611. As shown in Fig. 8, we found that the AUC of the angle discrimination accuracy was higher than the MMSE score.

#### **3.4 Discussion**

50 Neuroimaging for Clinicians – Combining Research and Practice

Fig. 8. Mean accuracy and discrimination threshold of MCI and AD compared to NC. (a) The mean accuracy of the three groups (NC: 82.1% ± 2.2%, MCI: 78.6% ± 1.8%, AD: 67.9% ± 2.5%). (b) The mean discrimination thresholds for the three groups are shown (NC: 8.7° ± 0.8°, MCI: 13.8° ± 2.7°, AD: 25.2° ± 4.2°). Vertical error bars represent standard error of the

Fig. 9. Receiver operating characteristic (ROC) curves for the angle discrimination accuracy and MMSE score showing discrimination between the NC and MCI groups. The solid line represents the ROC curve of angle discrimination accuracy, and the dashed line represents the ROC curve of MMSE score. The area under the curve (AUC) of angle discrimination

ROC analysis is a useful tool for evaluating the performance of diagnostic tests (Mendiondo et al., 2003; Zou et al., 2007). We used a ROC analysis to compare the sensitivity of angle discrimination accuracy to that of the MMSE scores. The fundamental measures of diagnostic accuracy are sensitivity (i.e., true positive rate) and specificity (i.e., true negative rate). As described in the Subjects section, all NC and MCI patients were defined by an MMSE score of 27 or greater and the CDR score of 0 and 0.5. In contrast, all AD patients had MMSE scores between 15 and 26. Therefore, only the ROC curves for angle discrimination

**3.3.2 Comparison between angle discrimination accuracy and MMSE score** 

mean. \*P < 0.05, \*\*P < 0.001.

accuracy is larger than that of MMSE score.

The present study demonstrated that patients with MCI or AD have an impaired ability to discriminate tactile angles compared to age-matched healthy subjects. The current study compared the performance of angle discrimination in three different subject groups (i.e., NC, MCI and AD). The results of this study indicated that there were significant group differences in the ability to discriminate tactile angles (NC > MCI > AD). Both the mean accuracy and threshold of angle discrimination of AD patients were significantly decreased compared to NC individuals and MCI patients. In contrast, although the mean threshold of angle discrimination of patients with MCI was significantly higher than NC individuals, the mean accuracy of the MCI group was similar to the NC group.

All subjects were asked to passively perceive the angles moved under their right index fingers and discriminate the largest of each angle pair, which consisted of a reference angle and a comparison angle. The current angle discrimination task is a commonly used procedure of tactile passive shape discrimination. The sensory feedback, which is critical for shape discrimination by passive touch, is generated by the four mechanoreceptive afferent systems (Johnson, 2001) located in the skin. Moreover, previous studies (Goodwin et al., 1997; Johnson, 2001) have suggested that tactile shape perception can be defined as the sum of the functions of cutaneous mechanoreceptors. However, there are numerous anatomical and morphological changes that develop with age and affect the hand and fingers. The density of mechanoreceptors in the skin was decreased (Wollard, 1936; Bruce, 1980), and the conduction velocity of peripheral nerves was significantly reduced with age (Peters, 2002). A decreased touch sensitivity in elderly individuals can cause many problems (Stevens & Choo, 1996; Vega-Bermudez & Johnson, 2002), including the inability to recognise objects by touch and an impaired ability to detect an object that has come into contact with the skin. Consequently, the ability of normal, older subjects to discriminate angles will be reduced compared to normal, young subjects. For example, the mean threshold of young subjects was 3.7° in our previous angle discrimination study (Wu et al., 2010), and the mean threshold of NC subjects in the present study was 8.7°, which was more than twice the previous value. However, the results of the present study indicated that the older subjects, as well as patients with MCI and AD, were able to complete the angle discrimination task. All subjects in the current experiment were able to perceive the change in size of the angle stimuli.

However, a significant deficit in angle discrimination was observed in MCI and AD patients in this study. One of the earliest symptoms of AD is impaired working memory (Baddeley et al., 1991; Bäckman & Small, 2007). In addition, previous studies have observed that patients with MCI also show impairments in memory processing compared to healthy aging subjects (Siedenberg et al., 1996; Petersen et al., 1999). In this study, all subjects were instructed to discriminate the larger of two angles by passive touch. To perform this task, the subject had to remember the composing feature of the first angle

Early Detection of Alzheimer's Disease with Cognitive Neuroscience Methods 53

discrimination between the MCI patients and the NC group using the present tactile discrimination system. These findings may improve the sensitivity of the previous mental

Audiovisual spatial and temporal orienting attention studies were examined in our study to further our understanding of neuroimaging studies for early detection of dementia. By using another method to compare the cognitive ability of tactile angle discrimination, we initially found that at the early stages of Alzheimer's disease, the behavioural cognition was decreased. These basic data that obtained from our studies, we consider that they are able to apply for a clinical diagnosis method of dementia early detection after enough confirm

A portion of this study was supported by a Grant-in-Aid for Scientific Research (B) 21404002, Japan and the AA Science Platform Program of the Japan Society for the

Anthony, J.C.; Le Resche, L.; Niaz, U.; Von Korff, M.R. & Folstein, M.F. (1982). Limits of the

Bäckman, L. & Small, B.J. (2007). Cognitive deficits in preclinical Alzheimer's disease and

Baddeley, A.D.; Bressi, S.; Della, S.S.; Logie, R. & Spinnler, H. (1991). The decline of

Bai, O.; Huang, D.; Lin, P.; Wu, J.; Chen, X. & Fei, D. (2010). An event-related study for

Blatow, M.; Nennig, E.; Durst, A.; Sartor, K. & Stippich, C. (2007). fMRI reflects

Bodegård, A.; Geyer, S.; Grefkes, C.; Zilles, K. & Roland, P.E. (2001). Hierarchical Processing

Bruce, M.F. (1980). The relation of tactile thresholds to histology in the fingers of elderly

Cai, C.; Kochiyama, T.; Osaka, K. & Wu, J. (2007). Lexical/semantic processing in dorsal left

Corbetta, M.; Miezin, F.; Shulman, G. & Petersen, S.E. (1993). A PET study of visuospatial

of Tactile Shape in the Human Brain. *Neuron*, Vol.31, pp.317–328

people. *Journal Neurol Neurosurg Psychiatry*, Vol.43, pp.730–734

inferior frontal gyrus, *NeuroReport*, Vol. 18, No.11, pp. 1147-1151

attention. The Journal of Neuroscience, Vol.13, pp.1202–1226

patients. *Pgychol Med*, Vol.12, pp. 397-408

*& Behavior*, Vol.92, pp.80–86

*Computational Biology*, Vol.2, p.1-10,

pp.2521–2542

pp.927–936

Mini-Mental State as screening test for demenia and delirium among hospital

vascular dementia: Patterns of findings from the Kungsholmen Project. *Physiology* 

working memory in Alzheimer's disease: A longitudinal study. *Brain*, Vol.114,

dynamic analysis of corticomuscular connectivity, *Biomedical Engineering and* 

functional connectivity of human somatosensory cortex. *NeuroImage* Vol.37,

tests for AD diagnosis and treatment.

**4. Conclusion** 

experiments.

**5. Acknowledgment** 

Promotion of Science.

**6. References** 

and then compare it with the second angle to make a judgment. The working memory contributed to the performance of the somatosensory discrimination (Bodegård et al., 2001; Kitada et al., 2006). Therefore, the present study suggests that the impaired working memory of MCI and AD patients is one factor that contributes to the decline of angle discrimination performance.

Moreover, we found that the mean accuracy of AD patients was significantly lower than the accuracy of the MCI and NC groups, and the mean threshold of AD patients was also reduced compared to the other two groups. Specifically, the mean threshold of AD patients was almost double the threshold of the MCI patients. However, we also found that there was a significant difference in the mean threshold between the MCI patients and the NC group, whereas the mean accuracy of the MCI patients and the NC group remained unchanged. There are two possible reasons to explain this phenomenon. First, AD is a neurodegenerative brain disease. Unlike patients with MCI, AD patients have more severe working memory impairments (Blatow et al., 2005). Second, the isolated memory impairment found in patients with MCI is more severe than the impairment observed in healthy aging individuals, whereas other cognitive functions remain normal. In contrast, AD patients have further deficits in spatial learning and memory and planning and problem solving (Kalman et al., 1995; Förstl & Kurz, 1999). These profound cognitive impairments of AD patients may explain the more severe deficits in angle discrimination found in patients with AD.

In addition, the tactile spatial discrimination procedure activates a diverse cerebral network (Bodegård et al., 2001; Kitada et al., 2006; Wu et al., 2010). The results from these neuroimaging studies also support our findings. For example, the intraparietal sulcus (located on the lateral surface of the parietal lobe) is engaged in multisensory spatial processing during the classification of grating and shape, and it has been shown that the intraparietal sulcus is a high-class area for computations and elaborates shape reconstructions (Bodegård et al., 2001). Neuroimaging studies (Delbeuck et al., 2003; Dickerson & Sperling, 2009; Huang et al., 2010) have demonstrated that abnormalities in the frontal, temporal, and parietal cortices contribute to the functional deficits in AD patients. Consequently, our results suggest that both the impairment of working memory and spatial discrimination of AD patients contribute to the lowest angle that is discernible compared to the MCI patients and the NC group.

The MMSE is a brief mental status examination designed to quantify the cognitive status in adults (Folstein et al., 1975). Recently, MMSE has been commonly used to test for complaints of memory problems or when a diagnosis of dementia is being considered. We plotted the ROC curves for the angle discrimination accuracy and MMSE score. We found that used the angle discrimination accuracy was better anbe to differentiate the MCI patients from older individuals than the MMSE score because the MMSE also has limitations. For example, previous studies (Anthony et al., 1982; Galasko et al., 1990) have suggested that the sensitivity of the MMSE has been rated at approximately 80%. Thus, the MMSE score may not represent the cognitive function deficits of all individuals. In contrast, we specifically focused on the difference in tactile angle discrimination in MCI and AD patients compared to the NC group. Although the present angle discrimination experiment examined working memory, spatial discrimination and problem-solving processes, there were limitations to this study. Despite these limitations, we have found a significant decline in tactile angle discrimination between the MCI patients and the NC group using the present tactile discrimination system. These findings may improve the sensitivity of the previous mental tests for AD diagnosis and treatment.
