Evaluation of apparent diffusion coefficient mappings in amnestic mild cognitive impairment using an image analysis software brain search

Abstract

Background

The apparent diffusion coefficient (ADC) from diffusion-weighted imaging (DWI) can quantify alterations in water diffusivity resulting from microscopic structural changes from amnestic mild cognitive impairment (aMCI) and Alzheimer's disease (AD).

Purpose

To investigate the ADC value for aMCI and AD using Brain Search (BS) software based on anatomical volumes of interest (AVOI).

Material and Methods

In total, 174 aged people were screened, and 25 patients with AD, 26 patients with aMCI, and 18 normal controls (NCs) were recruited. DWI was performed at 1.5 T with a fluid-attenuated inversion recovery (FLAIR), and the independent ADC mapping was generated after imaging acquisition. Ninety regional parcellations were adopted in a Brain Search (BS) based on the automated anatomic labeling atlas. The gray scale intensities (water diffusivity) from the collected ADC mappings were analyzed with BS. The mean value of each anatomical brain region was compared among aMCI, AD, and NC. The statistically significant (P < 0.05) group differences are displayed in color.

Results

During the pathological process of AD, the changes of water diffusivity appeared first in the left hippocampus, then gradually progressed to the bilateral sides and eventually displayed right lateralization. The ADC values from aMCI were obviously elevated compared to the values from the NC group in the left limbic cortex. Between the AD and NC groups, the significantly different brain areas included the bilateral hippocampus, the Cingulum_Mid, the ParaHippocampal_R, and the Temporal and Frontal lobes. There was a negative correlation between the ADC values and the scores from MMSE, MoCA, the Digit test, Raven's IQ, and WAIS IQ. Additionally, the ADC values were positively correlated with the scores from CDR, ADL, and ADAS-Cog.

Conclusion

The water diffusivity for aMCI and AD displays asymmetric anatomical lateralization. The water diffusivity alterations can be analyzed and visualized with our newly designed analytic imaging software, BS, which can be used as a good reference for examining and diagnosing aMCI and AD patients.

Keywords

Apparent diffusion coefficient mapping amnestic mild cognitive impairment anatomical volumes of interest automated anatomic labeling atlas

It is generally accepted that amnestic mild cognitive impairment (aMCI) (1) is an early stage of Alzheimer's disease (AD). Jack et al. (2) proposed a time-dependent order of aMCI and AD biomarkers, including both chemical (β-amyloid peptide, Aβ) and imaging (MRI and PET) biomarkers. In 2007, Dubois et al. (3) revised the diagnostic criteria (4) for aMCI and AD, by focusing on the use of MRIs as important supportive evidence. Therefore, it was of particular importance to explore neuroimaging markers during the pathological process from aMCI to AD (2, 5).

Recently, diffusion-weighted imaging (DWI) has been used to detect biochemical alteration in vivo (5, 6). The apparent diffusion coefficient (ADC) from DWI can quantify the variations in water diffusivity derived from microscopic structural changes. During the past decade, most ADC analysis has been based on regions of interest (ROIs) (5), which were traced by hand. The hippocampal ADC was determined by ROIs on an ADC mapping with coronal T1W anatomic references. Kantarci et al. (5, 7) found that hippocampal ADC values were higher in aMCI and AD patients when compared to normal controls (NCs), and a higher hippocampal diffusivity baseline was associated with a greater risk of aMCI deterioration (5).

Because of the inconvenience of drawing ROIs, more image processing software was developed. Goldszal et al. (8) divided the brain into three regions for analysis: white matter, gray matter, and cerebrospinal fluid. Statistical parametric mapping (SPM) (9) used Voxel-based analyses (VBA) for functional MRI (fMRI) analysis. However, to our knowledge, an image-processing system based on anatomical volumes of interest (AVOI) (10) for ADC mappings has not been previously reported.

In this study, we designed and validated an image-processing system named Brain Search (BS) based on AVOI. We compared classic ROI and our BS method to determine the regional water diffusivity among NCs, aMCI, and AD patients.

Material and Methods

Subjects

During a 2-year span from September 1, 2008 to October 31, 2010, 174 elderly people were screened for eligibility in this study. Among them, 30 subjects were recruited in the preliminary screening, but these data were not used in the final analysis. The exclusion criteria were set as follows: (a) concurrent illnesses or treatments interfering with cognitive function other than AD, such as addictions or psychiatric diseases; (b) a score higher than 4 on the Hachinski Ischemic Scale (11); (c) the presence of structural abnormalities that could cause cognitive impairment, such as acute infarctions, tumors, subdural hematoma, or the presence of leukoaraiosis higher than Fazekas' grade I (12), which were identified by conventional MRI; (d) less than 8 years of education; and (e) not right-handed.

According to the exclusion criteria, 55 subjects were ruled out, while the other 89 subjects were recruited for this MRI study. Quality control (QC) (13) for the ADC mappings was set up as described in the following paragraph. The image data from 20 subjects failed to pass QC. Therefore, in this ADC mapping analysis study, we recruited 69 people including 25 patients with AD (average age 67.0 ± 7.9, 18 men and 7 women), 26 with aMCI (average age 69.4 ± 7.4, 9F/17M) and 18 normal healthy controls (NC) (average age 64.7 ± 8.8, 12 men and 6 women) (Table 1). All patients underwent a general physical and clinical neurological examination. All the patients were informed of this study, and consent forms were obtained. Our institution's ethics committee approved our working protocol. NC subjects were matched to aMCI and AD patients by age, gender, and education.

Table 1
Demographic data of subjects

Probable AD aMCI NC

Enrolling (n) 62 58 54

Recruitment (n) 25 (7F/18M) 26 (9F/17M) 18 (6F/12M)

Age (y) 67.0 ± 7.9 69.4 ± 7.4 64.7 ± 8.8

Education (y) 12.5 ± 3.5 11.5 ± 4.2 13.7 ± 4.9

MMSE Score 14.9 ± 6.4 25.9 ± 3.5 28.9 ± 1.0

MOCA 10.88 ± 5.9 22.2 ± 1.9 27.3 ± 1.0

ADL 34.6 ± 19.2 21.9 ± 2.1 25.3 ± 9.4

Hachinski 2.067 ± 1.5 2.31 ± 1.2 1 ± 1.1

HAM-D 6.533 ± 5.2 3.15 ± 2.8 5.5 ± 5.1

CDR 1.056 ± 0.6 0.55 ± 0.2 0

NPI 0.867 ± 1.8 0 0

ADAS-cog 26.71 ± 11.9 10.8 ± 7.9 8 ± 3.4

Digit-symbol coding 7.8 ± 6.2 20.1 ± 15.7 27 ± 15.1

Digit span (in order) 7.167 ± 2.64 8 ± 1.8 8 ± 2.6

Digit span (backwards) 2.833 ± 0.8 4 ± 1.3 4.3 ± 2.1

Raven's IQ 45.33 ± 5.7 71.2 ± 10.3 92.6 ± 15.4

WAIS 57.6 ± 10.3 83.1 ± 7.2 103.1 ± 5.8

Data are the mean ± SD

MMS = Mini Mental State Examination; MoCA = Montreal cognitive assessment; CDR = Clinical Dementia Rating; ADL =Activities of daily living; ADAS-Cog= Alzheimer's Disease Assessment Scale-cognitive subscale; Raven's IQ = Raven's Progressive Matrices; WAIS= Wechsler Adult Intelligence Scale

Neuropsychological testing

Cognitive testing was performed before the MRI examination. Cognitive status was evaluated by the Mini-Mental State Examination (MMSE) (14) and the Montreal Cognitive Assessment (MoCA) (15). The severity of the dementia was rated based on the Clinical Dementia Rating (CDR) (16). The patient's functional status was evaluated using the activities of daily living (ADL) (4, 17). Memory state was evaluated by the Alzheimer's Disease Assessment Scale-Cognitive subscale (ADAS-Cog): Word recall (WdRc), Word recognition (WdRcg), Remembering instructions (RI), and Orientation (Ori). Language capability was measured by the ADAS-Cog: Naming objects and fingers (Nof), Commands (Com), Spoken language ability (Rlang), Word-finding (WFD), and Comprehension (ComSpk). The attention/executive examination employed ADAS-Cog: Attention (Attn), Constructional praxis (CPrax), and Ideational praxis (IPrax). Abstract reasoning was evaluated by Raven's Progressive Matrices (18). Attention and short-term memory were evaluated by Digit span, and study ability was evaluated by Digit-symbol coding from the Wechsler Adult Intelligence Scale (WAIS) subscores and subtests (19). The Neuropsychiatric Inventory (NPI) (20) was employed to assess the psychopathology of the dementia patients. The Hamilton Depression rating scale score (HAM-D) (21) was used to rate the severity of a patient's major depression.

All tests were administered by an experienced psychologist (Qiu-yun Cao) and two neurologists (Hui Zhao, Lai Qian) and supervised by a clinical neuropsychologist (Yun Xu). At the completion of all of the examinations, a consensus committee meeting was held to evaluate the data from all patients and to make a clinical diagnosis in each case. The consensus committee members include behavioral neurologists (Hui Zhao, Jun Xu, Yun Xu), a neuropsychologist (Qiu-yun Cao), and a radiologist (Bing Zhang).

Diagnostic criteria

A clinical diagnosis of aMCI was made according to the criteria of Petersen et al. (22). Patients with aMCI met the following criteria: (a) complaint of memory loss; (b) an MMSE from 25 ∼ 30 and/or a MoCA from 19 ∼ 25; (c) normal activities of daily living (ADL = 20); (d) CDR = 0.5 (that is, a Memory Box score of at least 0.5 without deficits in other cognitive domains); and (e) a Hachinski score ≤4, with NPI = 0 and HAM-D ≤ 12 on a 17-item scale.

The clinical criteria used for a probable AD diagnosis were adopted from the National Institute of Neurological and Communicative Disorders and Stroke/Alzheimer's Disease and Related Disorders Association (NINCDS/ADRDA) criteria (4), namely (2, 3, 23): (a) a gradual and progressive change in memory function reported by patients or informants over more than 6 months; (b) an MMSE score less than 24 and/or a MoCA less than 18; (c) ADL > 20, (d) CDR > 0.5, where mild = 1, middle = 2, and severe = 3; (e) a Hachinski score ≤4, with NPI = / > 0 and HAM-D ≤ 17 on the 17-item scale; and (f) the presence of medial temporal lobe atrophy as a supportive feature, according to the MRI findings.

Normal controls were defined as individuals who (a) had no cognitive complaints, (b) had an MMSE score between 27 ∼ 30 and/or a MoCA of 26 ∼ 30, and (c) had no evidence of any abnormality except atrophy when examined by a conventional MRI.

MRI acquisition

All experiments were conducted with a 1.5-T clinical MR image scanner (Intera Master Gyroscan, Philips Medical Systems, Best, The Netherlands). The conventional sequences were set as follows: T1W (SE, TRA, TR 458 ms, TE 14 ms, T2W TSE, TRA, SAG, TR 4070 ms, TE 110ms), THK/GAP 5 mm/0, slices 18. A high-resolution coronal 3DT1FFE acquisition was performed with TR/TE at 25/4.6 ms, the flip angle at 30 degrees in a plane resolution of 1.0 mm and a slice thickness of 1.5 mm for anatomical reference. The conventional total scan time was approximately 5 min.

The EPI sequence of DWI with fluid attenuation inversion recovery (FLAIR) was set at TR/TI/TE = 6000/1900/95 ms, b = 1000 ms/mm², with an in-plane resolution of 2.0 mm and a slice thickness of 2.5 mm. The partial volume signal from CSF was avoided by suppression by FLAIR DWI (13, 24). The independent ADC mapping was generated using affine registration software (Diffusion postprocessing package, Philips Medical Systems). The pixel value of the gray scale mappings was calculated using the Stejskal and Tanner equation (25) and was expressed in units of 10⁻⁶ mm²/s.

Single ROI DWI-ADC

The ADC value was measured as the signal intensity on the ADC mapping. The bilateral hippocampus (HP) and posterior cingulated gyrus (PC) were traced and utilized in an ROI analysis. The coronal T1-weighted MR image and FLAIR b0 volume were used to guide the location of the ROIs (26). The size of each ROI was approximately 15–30 pixels.

Quality controls (QC) for ADC mappings

The ADC mappings that failed to pass QC were ruled out (6) based on the following standards: (a) an existing susceptibility artifact with geometric warping from magnetic field inhomogeneities; (b) the presence of image warping from eddy currents; and (c) the presence of other image artifacts, such as severe, aperiodic motion artifacts or magnet material artifacts.

Image preprocessing for ADC mapping analysis

Each image preprocessing was performed using SPM8 (27) to convert Dicom raw data into analyzed types (.hdr and .img). Each b0 image that was used as a ‘source image’ was spatially normalized with the EPI template image using the normalization function with standard defaults in SPM8. Then a set of warps was written back to the ADC mapping using the ‘image to write’ function. The normalized ADC images were then imported into the Brain Search (BS) software (written by JGZ and WHG). A regional parcellation method was adopted in BS according to the automated anatomic labeling (28) atlas, which was validated previously by Tzourio-Mazoyer et al. (10). This parcellation method divides each cerebral hemisphere into 45 anatomical brain ROIs. The group difference for the mean value of each anatomical brain region was displayed, and the statistical significance value was set at P < 0.05.

Statistical analysis

BS software provides some basic statistical methods, such as ANOVA, Student's t-test, and regression analysis. SPSS 18 (SPSS Inc., Chicago, IL, USA) was also used to analyze data from the single ROI method.

Group differences among different ages and education levels were analyzed by a group t-test. Gender differences were analyzed by a χ² test. ADC values from patients and the NC group were analyzed by a group t-test and ANOVA, and they were reported as means ± the standard deviation (SD). The ADC values and clinical rating scale scores for subjects were examined using Pearson's correlation coefficient. The effects of the clinical rating scale scores on ADC were tested in a single model using multiple regression analysis. The significance level for these analyses was set at P < 0.05.

Results

Demographic data of subjects

There was no significant difference (P > 0.05) among NC, aMCI, and AD groups when age, gender, and education level were matched. AD patients had lower scores in MMSE, MOCA, digit-symbol coding, digit span (in order), digit span (backwards), Raven's IQ and WAIS IQ compared with NC and aMCI patients (P < 0.001); but AD patients did have higher scores in ADAS-cog, ADL, HAM-D, CDR and NPI (P < 0.001) (Table 1), confirming that AD patients had more severe recognition impairments.

ADC mapping analysis by the single ROI method

The ADC values from the hippocampus of NC patients that were obtained by Single ROI were 746.2 ± 60.4, and they increased to 803.4 ± 62.6 in aMCI patients and to 884.0 ± 31.4 in AD patients. The ADC values from the posterior cingulums for the three groups were 731.6 ± 45.5, 797.8 ± 65.9 and 857.0 ± 73.3, respectively. The increases were statistically significant (P < 0.05) (Fig. 1).

Fig. 1
The average ADC measurements in bilateral hippocampus (a) and bilateral posterior cingulum (b) between groups

ADC mapping analysis by BS

ADC mappings from the BS analysis displayed the localization, magnitude, and extent of abnormalities (Fig. 2, Table 2). During the pathological process of AD, the changes of water diffusivity appeared first in the left hippocampal and parahippocampal areas, and then gradually progressed to the bilateral sides, eventually displaying right lateralization. The ADC values of aMCI patients were obviously elevated compared to the ADC values of NC patients in the limbic cortex (Hippocampus_L, ParaHippocampal_L, Insula_R), the Thalamus_L, the Angular_R, and the frontal lobe. The ADC values from the AD group were significantly higher than those seen in the aMCI group in the limbic cortex (Hippocampus_R, Bilateral Cingulum_Mid and Temporal_Pole_Sup), and the frontal lobe, but not in the Hippocampus_L. Between AD and NC patients, the brain areas that displayed significant differences included the bilateral hippocampus, the Cingulum_Mid, the ParaHippocampal_R, the Insula_R, the Temporal_Pole_Sup_L, the SupraMarginal_R, Angular_L, and the frontal lobe.

Fig. 2
The figure illustrates a group t-test in ADC mappings by using the BS method to compare NC and aMCI (left), AD and MCI (middle), AD and NC (right). When compared to NC patients, in aMCI patients the water diffusivity was significantly different in the left temporomesial areas (left figure), When the patients with AD were compared to NC patients, the alterations had now additionally progressed to the right temporal lobe (right figure), reaching a seemingly symmetrical bilateral abnormality. The data show that the right temporomesial abnormality develops later than the left and thus appears different when compared with the aMCI (middle figure). Moreover, the color of the statistical figures presented P value, which indicated that the P value in the right hippocampus (P < 0.001), is less than that on the left side (P < 0.01). That result shows the statistical difference is more obvious in the right hippocampus

Table 2
ANOVA analysis in ADC mappings by using BS among groups

aMCI > NC AD > aMCI AD > NC

Limbic cortex Brain regions P Brain regions P Brain regions P

Hippocampus_L 0.03 Hippocampus_R 0.049 Hippocampus_L 0.01

ParaHippocampal_L 0.01 Cingulum_Mid_R 0.01 Hippocampus_R 0.00

Insula_R 0.00 Cingulum_Mid_L 0.01 ParaHippocampal_R 0.01

Insula_R 0.00

Cingulum_Mid_R 0.01

Cingulum_Mid_L 0.01

Temporal lobe Temporal_Pole_Sup_L 0.004 Temporal_Pole_Sup_L 0.00

Temporal_Pole_Sup_R 0.02

Frontal lobe Frontal_Sup_Orb_L 0.02 Frontal_Sup_R 0.03 Frontal_Inf_Tri_R 0.03

Frontal_Sup_Orb_R 0.03 Rolandic_Oper_L 0.03 Rolandic_Oper_L 0.00

Frontal_Mid_Orb_L 0.00 Rolandic_Oper_R 0.01 Rolandic_Oper_R 0.00

Frontal_Sup_Medial_R 0.01 Heschl_L 0.00 Heschl_L 0.00

Frontal_Med_Orb_R 0.00

Parietal lobe Angular_R 0.04 SupraMarginal_R 0.02

Angular_L 0.03

Others Thalamus_L 0.03 Cuneus_R 0.03

Occipital_Inf_R 0.04 Fusiform_L 0.03

The correlation between water diffusivity and clinical rating scales

There were correlations between the averaged ADC values in the bilateral hippocampus (water diffusivity) change and the changes in the clinical rating scales. There was a negative correlation between the hippocampal ADC values of water diffusivity and the scores from MMSE, MoCA, digit-symbol coding, digit span (in order), digit span (backwards), Raven's IQ, and WAIS IQ. Additionally, the hippocampal ADC values of water diffusivity were positively correlated with the scores from CDR, ADL, and ADAS-Cog (Figs. 3 and 4). Two-tailed Pearson correlation coefficients were also used (P < 0.05).

Fig. 3
The figure shows a cross-correlation of the averaged ADC value in the bilateral hippocampus and the scores of MoCA, WAIS. There had been a negative correlation between the water diffusivity and MoCA or WAIS IQ scores. The effects of clinical rating scales on the averaged ADC values in the bilateral hippocampus were analyzed by a multiple regression analysis (a step-wise model) of all the subjects. WAIS and MoCA scores affected ADC values significantly

Fig. 4
Cross-correlation of an inter-group analysis is shown. There had been a negative correlation between the averaged ADC value (water diffusivity) in the bilateral hippocampus and the scores from MMSE, digit-symbol coding, digit span (in order), digit span (backwards), Raven's IQ. Concurrently, water diffusivity had a positive correlation with the scores of CDR, ADL, ADAS-Cog

The effects of clinical rating scales on the averaged ADC values in the bilateral hippocampus were analyzed by a multiple regression analysis (a step-wise model) of all the subjects. WAIS and MoCA scores affected ADC values significantly (Table 3).

Table 3
ADC measurements (mean± SD) and coefficients by multiple regression analysis

ADC Groups HOV ANOVA NC-aMCI NC-AD aMCI-AD

NC (18) aMCI (26) AD (25) Levene Sig. F Sig. Sig. Sig. Sig.

HP 746.2 ± 60.4 803.4 ± 62.6 884.0 ± 31. 3.73 0.029 37.17 0.00 0.01 0.00 0.00

Rating scales Unstandardized coefficients Standardized coefficients t Sig.

WAIS −2.07 −0.55 −4.42 0

MoCA −2.61 −0.25 −2.05 0.04

The unit of the ADC value was 10^−6mm2/s

Statistical significance for P < 0.05

Levene Statistical analysis was used in the Test of Homogeneity of Variances (abbreviated as HOV)

There were three significant test in the corresponding groups, including NC to aMCI, NC to AD and aMCI to AD in the above Multiple Comparisons by LSD and post hoc tests

The averaged ADC values were measured in the bilateral hippocampus (abbreviated as HP)

The effects of clinical rating scales on the averaged ADC values in the bilateral hippocampus were analyzed by multiple regression analysis (a step-wise model) of all the subjects

WAIS and MoCA scores affected ADC values significantly

Discussion

The goal of this study was to investigate changes in the ADC value in aMCI and AD patients and to validate the image analysis software developed by our team. We obtained the following results: (a) The water diffusivity in aMCI and AD patients displays asymmetric anatomical lateralization. During the pathological process of AD, the changes of water diffusivity appeared first in left hippocampal and parahippocampal areas, and then gradually progressed to the bilateral sides, eventually displaying right lateralization; and (b) The water diffusivity alterations can be analyzed and visualized with our newly designed BS imaging analytic software, which can be used as a good reference in the examination and diagnosis of aMCI and AD patients.

The loss of neuron cells, axons, and dendrites in AD patients contributes to the expansion of the extracellular space, leading to a loss of restrictive barriers and acceleration of water diffusivity (5). Most aMCI cases showed early AD pathologic features in the limbic cortex (5, 29). Therefore, increased diffusivity may be sensitive to those early pathologic features before aMCI progresses to AD. We found that the ADC values in the hippocampus and the limbic system were higher in aMCI and AD patients than in NC patients, using both the single ROI method and the BS method based on AVOI.

FMRIs (30) indicated that limbic cortex, including the hippocampus and the parahippocampal gyrus, played an important role in memory encoding and the recognition of scenes. In our study, when compared to NC patients, the water diffusivity was significantly different in the left hippocampal areas in aMCI patients. When the AD patients were compared to NC patients, the alterations had progressed also to the right hippocampus, reaching a seemingly symmetrical bilateral abnormality. The data showed that the right medial temporal lobe abnormality developed later than the left. These results were in accordance with the theory of asymmetrical anatomical lateralization. Shapleske (31) advocated that most right-handers, and more than half of all left-handers, showed leftward asymmetry in the planum temporale. Moreover, Wang (32) found that hippocampal connectivity in control subjects displayed rightward asymmetry, which was diminished in AD patients. During the deterioration from aMCI to AD, the left medial temporal is first involved because of the leftward anatomical lateralization. Then, the deterioration arrives at the bilateral medial temporal. Finally, in the late stage of AD, hippocampal diffusivity extends to rightward lateralization, indicating the involvement of right limbic system. This observation was in line with the previous theory that aMCI, as a preclinical Alzheimer disease, tended to form mild neurofibrillary tangles, amyloid plaques, and atrophy first in the hippocampus (33).

In addition to hippocampus and parahippocampal gyrus, our BS software can effectively analyze other brain regions, including the insular cortex, the medial prefrontal regions, the Rolandic operculum, the thalamus and the cingulum. Recently, insula has attracted increasing attention for its role in body representation and subjective emotional experience (34). However, the medial prefrontal lobe seems to function in network in its role in retrospective memory (35). In our study, the results from water diffusivity of the limbic system correlated significantly with clinical rating scales, which measured different impairments of the cortex, including language, attention/execution, abstract reasoning, and study ability. Indefrer's (36) study provided evidence that the Rolandic operculum was involved in both sentence-level and local (phrase-level) syntactic encoding. In agreement with the lower scores in the language tests of AD patients, which indicate difficulty in word-finding, we found that ADC values increased in the bilateral Rolandic operculum. We also detected increased ADC values in the left thalamus in aMCI patients and in the cingulum in AD patients, which represented the memory impairment in the Papez circuit (37).

Our data are in agreement with prior studies (26) that show that the order of the changes in the ADC value in different brain regions followed the pathological progression of AD patients. For example, AD patients had a progression of pathology from the medial temporal lobes to the temporoparietal cortex. Our measurements by BS followed exactly the same pattern of water diffusivity changes, which fits well with the Braak (38) neurofibrillary pathological staging scheme in AD.

In contrast to SPM, a method based on the VBA method, BS is based on the AVOI that utilizes the AAL atlas as a template (10). The advantage of using AVOI is that it parcellated each cerebral cortex into 45 anatomical brain ROIs to reduce type II statistical error. Such errors were inevitable in VBA. Furthermore, BS results are reported anatomically with AAL labeling, and it is convenient to link the ADC value changes to the affected brain regions.

A limitation of this study is the lack of pathologic confirmation in all subjects. Patients who meet the clinical criteria for AD often have some degree of vascular dementia pathology. Although we excluded subjects who had a score higher than 4 on the Hachinski Ischemic Scale and those who had the presence of leukoaraiosis higher than a Fazekas' grade I, we acknowledge that some of our patients likely have mixed pathologies. We will follow-up with all cases to verify our hypothesis.

In conclusion, the examination of water diffusivity patterns in aMCI and AD patients by MRI analysis is a powerful way to assess prodromal AD progression. As a useful tool to analyze water diffusivity, our newly developed BS method can visualize global and regional brain diffusivity, thereby potentially improving the diagnosis and follow-up examinations of aMCI and AD patients.

	Probable AD	aMCI	NC
Enrolling (n)	62	58	54
Recruitment (n)	25 (7F/18M)	26 (9F/17M)	18 (6F/12M)
Age (y)	67.0 ± 7.9	69.4 ± 7.4	64.7 ± 8.8
Education (y)	12.5 ± 3.5	11.5 ± 4.2	13.7 ± 4.9
MMSE Score	14.9 ± 6.4	25.9 ± 3.5	28.9 ± 1.0
MOCA	10.88 ± 5.9	22.2 ± 1.9	27.3 ± 1.0
ADL	34.6 ± 19.2	21.9 ± 2.1	25.3 ± 9.4
Hachinski	2.067 ± 1.5	2.31 ± 1.2	1 ± 1.1
HAM-D	6.533 ± 5.2	3.15 ± 2.8	5.5 ± 5.1
CDR	1.056 ± 0.6	0.55 ± 0.2	0
NPI	0.867 ± 1.8	0	0
ADAS-cog	26.71 ± 11.9	10.8 ± 7.9	8 ± 3.4
Digit-symbol coding	7.8 ± 6.2	20.1 ± 15.7	27 ± 15.1
Digit span (in order)	7.167 ± 2.64	8 ± 1.8	8 ± 2.6
Digit span (backwards)	2.833 ± 0.8	4 ± 1.3	4.3 ± 2.1
Raven's IQ	45.33 ± 5.7	71.2 ± 10.3	92.6 ± 15.4
WAIS	57.6 ± 10.3	83.1 ± 7.2	103.1 ± 5.8

	aMCI > NC	AD > aMCI	AD > NC
Limbic cortex	Brain regions	P	Brain regions	P	Brain regions	P
	Hippocampus_L	0.03	Hippocampus_R	0.049	Hippocampus_L	0.01
	ParaHippocampal_L	0.01	Cingulum_Mid_R	0.01	Hippocampus_R	0.00
	Insula_R	0.00	Cingulum_Mid_L	0.01	ParaHippocampal_R	0.01
					Insula_R	0.00
					Cingulum_Mid_R	0.01
					Cingulum_Mid_L	0.01
Temporal lobe			Temporal_Pole_Sup_L	0.004	Temporal_Pole_Sup_L	0.00
			Temporal_Pole_Sup_R	0.02
Frontal lobe	Frontal_Sup_Orb_L	0.02	Frontal_Sup_R	0.03	Frontal_Inf_Tri_R	0.03
	Frontal_Sup_Orb_R	0.03	Rolandic_Oper_L	0.03	Rolandic_Oper_L	0.00
	Frontal_Mid_Orb_L	0.00	Rolandic_Oper_R	0.01	Rolandic_Oper_R	0.00
	Frontal_Sup_Medial_R	0.01	Heschl_L	0.00	Heschl_L	0.00
	Frontal_Med_Orb_R	0.00
Parietal lobe	Angular_R	0.04			SupraMarginal_R	0.02
					Angular_L	0.03
Others	Thalamus_L	0.03	Cuneus_R	0.03
	Occipital_Inf_R	0.04	Fusiform_L	0.03

ADC	Groups	HOV	ANOVA	NC-aMCI	NC-AD	aMCI-AD
HP	746.2 ± 60.4	803.4 ± 62.6	884.0 ± 31.	3.73	0.029	37.17	0.00	0.01	0.00	0.00
Rating scales				Unstandardized coefficients	Standardized coefficients	t	Sig.
WAIS				−2.07	−0.55	−4.42	0
MoCA				−2.61	−0.25	−2.05	0.04

Footnotes

ACKNOWLEDGMENTS

This work was supported by Jiangsu Province innovation foundation for a PhD candidate (CX09B_015Z,10284, BZ), a key project grant by the Medical Science and technology development Foundation, Nanjing Department of Health (ZKX10006, BZ), and the Science and technology project of Jiangsu Province health department (H200740, BZ). Additional support was gained from the National Natural Science Foundation of China (30971010,30670739, XY), the Doctoral Program Foundation of the Ministry of Education of China (20060284044, XY), the Out-standing Researcher Program (RC2007006, XY), and the National Natural Science Foundation (BK2009037, XY) of Jiangsu Province of China. Funding was supplied from the State Key Laboratory of Pharmaceutical Biotechnology (KF-GN-200901, XY) and the 973 Program from the Ministry of Science and Technology of China (2009CB21906, XY). We would like to thank Marilyn White, Zhen-yu Yin, and Lei Huang for suggesting changes.

Conflict of interest: None.

References

Petersen

, Smith

, Waring

, Mild cognitive impairment: clinical characterization and outcome. Arch Neurol 1999;56:303–8

Jack

Jr. , Knopman

, Jagust

, Hypothetical model of dynamic biomarkers of the Alzheimer's pathological cascade. Lancet Neurol 2010;9:119–28

Dubois

, Feldman

, Jacova

, Research criteria for the diagnosis of Alzheimer's disease: revising the NINCDS-ADRDA criteria. Lancet Neurol 2007;6:734–46

McKhann

, Drachman

, Folstein

, Clinical diagnosis of Alzheimer's disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer's Disease. Neurology 1984;34:939–44

Kantarci

, Petersen

, Boeve

, DWI predicts future progression to Alzheimer disease in amnestic mild cognitive impairment. Neurology 2005;64:902–4

Johansen-Berg

, Behrens

TEJ

. Diffusion MRI From quantitative measurement to in-vivo neuroanatomy. Elsevier Inc 2009;43–6

Kantarci

, Avula

, Senjem

, Dementia with Lewy bodies and Alzheimer disease: neurodegenerative patterns characterized by DTI. Neurology 2010;74:1814–21

Goldszal

, Davatzikos

, Pham

, An image-processing system for qualitative and quantitative volumetric analysis of brain images. J Comput Assist Tomogr 1998;22:827–37

Fein

, Landman

, Tran

, Statistical parametric mapping of brain morphology: sensitivity is dramatically increased by using brain-extracted images as inputs. NeuroImage 2006;30:1187–95

10.

Tzourio-Mazoyer

, Landeau

, Papathanassiou

, Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. NeuroImage 2002;15:273–89

11.

Hachinski

, Iliff

, Zilhka

, Cerebral blood flow in dementia. Arch Neurol 1975;32:632–7

12.

Heo

, Lee

, Kon

, White matter hyperintensities and cognitive dysfunction in Alzheimer disease. J Geriatr Psychiatry Neurol 2009;22:207–12

13.

Leergaard

T. B

, White

N. S

, de Crespigny

, Quantitative histological validation of diffusion MRI fiber orientation distributions in the rat brain. PLoS One 2011:5:85–95

14.

Folstein

, Folstein

, McHugh

. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189–98

15.

Nasreddine

, Phillips

, Bedirian

, The Montreal Cognitive Assessment, MoCA: a brief screening tool for mild cognitive impairment. J Am Geriatr Soc 2005;53:695–9

16.

Morris

, Ernesto

, Schafer

, Clinical dementia rating training and reliability in multicenter studies: the Alzheimer's Disease Cooperative Study experience. Neurology 1997;48:1508–10

17.

Roper

, Logan

, Tierney

. The Roper-Logan-Tierney Model of Nursing: Based on Activities of Living. Elsevier Health Sciences 2000

18.

Raven

. Mental tests used in genetic studies: The performance of related individuals on tests mainly educative and mainly reproductive. MSc Thesis 1936

19.

Lichtenberger

, Kaufman

. Assessing adolescent adult intelligence. John Wiley & Sons 2002;50–5

20.

Cummings

. The Neuropsychiatric Inventory: assessing psychopathology in dementia patients. Neurology 1997;48:10–6

21.

Hedlund

, Viewig

. The Hamilton rating scale for depression: a comprehensive review. Journal of Operational Psychiatry 1979;10:149–65

22.

Petersen

. Mild cognitive impairment as a diagnostic entity. J Intern Med 2004;256:183–94

23.

Petersen

, Aisen

, Beckett

, Alzheimer's Disease Neuroimaging Initiative (ADNI): clinical characterization. Neurology 2010;74:201–9

24.

Kwong

, McKinstry

, Chien

, CSF-suppressed quantitative single-shot diffusion imaging. Magn Reson Med 1991;21:157–63

25.

Stejskal

, Tanner

. Spin diffusion measurements: spin-echo in presence of a time dependent field gradient. J Chem Phys 1965;42:288–92

26.

Kantarci

, Jack

Jr. , Xu

, Mild cognitive impairment and Alzheimer disease: regional diffusivity of water. Radiology 2001;219:101–7

27.

Statistical Parametric Mapping. Wellcome Department of Cognitive Neurology http://wwwfilionuclacuk/spm/

28.

Prins

, van Straaten

, van Dijk

, Measuring progression of cerebral white matter lesions on MRI: visual rating and volumetrics. Neurology 2004;62:1533–9

29.

Carpenter

, Xiong

, Porensky

, Reaction to a dementia diagnosis in individuals with Alzheimer's disease and mild cognitive impairment. J Am Geriatr Soc 2008;56:405–12

30.

Michels

, Mehnert

, Boy

, The somatosensory representation of the human clitoris: an fMRI study. NeuroImage 2010;49:177–84

31.

Shapleske

, Rossell

, Woodruff

, The planum temporale: a systematic, quantitative review of its structural, functional and clinical significance. Brain Res Brain Res Rev 1999;29:26–49

32.

Wang

, Zang

, He

, Changes in hippocampal connectivity in the early stages of Alzheimer's disease: evidence from resting state fMRI. NeuroImage 2006;31:496–504

33.

Fox

, Warrington

, Freeborough

, Presymptomatic hippocampal atrophy in Alzheimer's disease. A longitudinal MRI study. Brain 1996;119:2001–7

34.

Kawasaki

, Kaufman

, Damasio

, Single-neuron responses to emotional visual stimuli recorded in human ventral prefrontal cortex. Nat Neurosci 2001;4:15–6

35.

Kesner

. Retrospective and prospective coding of information: role of the medial prefrontal cortex. Exp Brain Res 1989;74:163–7

36.

Indefrey

, Brown

, Hellwig

, A neural correlate of syntactic encoding during speech production. Proc Natl Acad Sci U S A 2001;98:5933–6

37.

Azzaroni

, Parmeggiani

. Papez circuit and regulation of the hippocampal theta rhythm in the cat. Electroencephalogr Clin Neurophysiol 1968;25:84–5

38.

Braak

, Braak

. Evolution of the neuropathology of Alzheimer's disease. Acta Neurol Scand Suppl 1996;165:3–12

ADC	Groups			HOV		ANOVA		NC-aMCI	NC-AD	aMCI-AD
	NC (18)	aMCI (26)	AD (25)	Levene	Sig.	F	Sig.	Sig.	Sig.	Sig.
HP	746.2 ± 60.4	803.4 ± 62.6	884.0 ± 31.	3.73	0.029	37.17	0.00	0.01	0.00	0.00
Rating scales				Unstandardized coefficients	Standardized coefficients	t	Sig.
WAIS				−2.07	−0.55	−4.42	0
MoCA				−2.61	−0.25	−2.05	0.04