Jin-Ju Yang

Automated Sulcal Depth Measurement on Cortical Surface Reflecting Geometrical Properties of Sulci

Sulcal depth that is one of the quantitative measures of cerebral cortex has been widely used as an important marker for brain morphological studies. Several studies have employed Euclidean (EUD) or geodesic (GED) algorithms to measure sulcal depth, which have limitations that ignore sulcal geometry in highly convoluted regions and result in under or overestimated depth. In this study, we proposed an automated measurement for sulcal depth on cortical surface reflecting geometrical properties of sulci, which named the adaptive distance transform (ADT). We first defined the volume region of cerebrospinal fluid between the 3D convex hull and the cortical surface, and constructed local coordinates for that restricted region. Dijkstra’s algorithm was then used to compute the shortest paths from the convex hull to the vertices of the cortical surface based on the local coordinates, which may be the most proper approach for defining sulcal depth. We applied our algorithm to both a clinical dataset including patients with mild Alzheimer’s disease (AD) and 25 normal controls and a simulated dataset whose shape was similar to a single sulcus. The mean sulcal depth in the mild AD group was significantly lower than controls (p = 0.007, normal [mean±SD]: 7.29±0.23 mm, AD: 7.11±0.29) and the area under the receiver operating characteristic curve was relatively high, showing the value of 0.818. Results from clinical dataset that were consistent with former studies using EUD or GED demonstrated that ADT was sensitive to cortical atrophy. The robustness against inter-individual variability of ADT was highlighted through simulation dataset. ADT showed a low and constant normalized difference between the depth of the simulated data and the calculated depth, whereas EUD and GED had high and variable differences. We suggest that ADT is more robust than EUD or GED and might be a useful alternative algorithm for measuring sulcal depth.

Dimensionality reduced cortical features and their use in the classification of Alzheimer's disease and mild cognitive impairment.

Features defined on the cortical surface derived from magnetic resonance imaging provide important information to distinguish normal controls from Alzheimers disease (AD) and mild cognitive impairment (MCI). We adopted cortical thickness and sulcal depth, parameterized by three dimensional meshes, as our feature. The cortical feature is high dimensional and direct use of it is problematic in a modern classifier due to small sample size problem. We applied manifold learning to reduce the dimensionality of the feature and then tested the usage of the dimensionality reduced feature with a support vector machine classifier. A leave-one-out cross-validation was adopted for quantifying classifier performance. We chose principal component analysis (PCA) as the manifold learning method. We applied PCA to a region of interest within the cortical surface. Our classification performance was at least on par for the AD/normal and MCI/normal groups and significantly better for the AD/MCI groups compared to recent studies. Our approach was tested using 25 AD, 25 MCI, and 50 normal control patients from the OASIS database.

The relationship between the presence of sulcal pits and intelligence in human brains

Sulcal pits are hypothesized to form early during development and be under tighter genetic control than other regions of the cortex. We investigated the relationship between the presence of sulcal pits and intellectual ability, estimated with the full-scale, verbal, and performance intelligence quotient (IQ), in the brains of 78 healthy young adults. We automatically extracted sulcal pits from magnetic resonance images and developed a method for their automatic labeling. The difference in the number of sulcal pits between high and average IQ groups for each labeled region was statistically analyzed. We found that in the high verbal IQ group a sulcal pit was more frequently present in the left posterior inferior frontal sulcus (70% in the high IQ group vs. 40% in the average IQ group) and the right posterior inferior temporal sulcus (70% vs. 43%), which have been reported to be regions of language function. Greater mean curvature of the deep sulcal areas in these regions was shown for the high verbal IQ group. This provides the complementary morphological information about the presence of more sulcal pits. These findings suggest that factors influencing verbal intelligence may emerge in the language areas early during cortical development and may be under tight genetic control.

Combined effects of physical exercise and education on age-related cortical thinning in cognitively normal individuals

We investigated the association between self-reported physical exercise and cortical thickness in a large sample of cognitively normal individuals. We also determined whether a combination of physical exercise and education had more protective effects on age-related cortical thinning than either parameter alone. A total of 1,842 participants were included in this analysis. Physical exercise was assessed using a questionnaire regarding intensity, frequency, and duration. Cortical thickness was measured using a surface-based method. Longer duration of exercise (≥1 hr/day), but not intensity or frequency, was associated with increased mean cortical thickness globally (P-value = 0.013) and in the frontal regions (P-value = 0.007). In particular, the association of exercise with cortical thinning had regional specificity in the bilateral dorsolateral prefrontal, precuneus, left postcentral, and inferior parietal regions. The combination of higher exercise level and higher education level showed greater global and frontal mean thickness than either parameter alone. Testing for a trend with the combination of high exercise level and high education level confirmed this finding (P-value = 0.001–0.003). Our findings suggest that combined exercise and education have important implications for brain health, especially considering the paucity of known protective factors for age-related cortical thinning.

Prediction for human intelligence using morphometric characteristics of cortical surface: partial least square analysis.

A number of imaging studies have reported neuroanatomical correlates of human intelligence with various morphological characteristics of the cerebral cortex. However, it is not yet clear whether these morphological properties of the cerebral cortex account for human intelligence. We assumed that the complex structure of the cerebral cortex could be explained effectively considering cortical thickness, surface area, sulcal depth and absolute mean curvature together. In 78 young healthy adults (age range: 17-27, male/female: 39/39), we used the full-scale intelligence quotient (FSIQ) and the cortical measurements calculated in native space from each subject to determine how much combining various cortical measures explained human intelligence. Since each cortical measure is thought to be not independent but highly inter-related, we applied partial least square (PLS) regression, which is one of the most promising multivariate analysis approaches, to overcome multicollinearity among cortical measures. Our results showed that 30% of FSIQ was explained by the first latent variable extracted from PLS regression analysis. Although it is difficult to relate the first derived latent variable with specific anatomy, we found that cortical thickness measures had a substantial impact on the PLS model supporting the most significant factor accounting for FSIQ. Our results presented here strongly suggest that the new predictor combining different morphometric properties of complex cortical structure is well suited for predicting human intelligence.

Correlations between Gray Matter and White Matter Degeneration in Pure Alzheimer’s Disease, Pure Subcortical Vascular Dementia, and Mixed Dementia

Alzheimer’s disease dementia (ADD) and subcortical vascular dementia (SVaD) both show cortical thinning and white matter (WM) microstructural changes. We evaluated different patterns of correlation between gray matter (GM) and WM microstructural changes in pure ADD, pure SVaD, and mixed dementia. We enrolled 40 Pittsburgh compound B (PiB) positive ADD patients without WM hyperintensities (pure ADD), 32 PiB negative SVaD patients (pure SVaD), 23 PiB positive SVaD patients (mixed dementia), and 56 normal controls. WM microstructural integrity was quantified using fractional anisotropy (FA), axial diffusivity (DA), and radial diffusivity (DR) values. We used sparse canonical correlation analysis to show correlated regions of cortical thinning and WM microstructural changes. In pure ADD patients, lower FA in the frontoparietal area correlated with cortical thinning in the left inferior parietal lobule and bilateral paracentral lobules. In pure SVaD patients, lower FA and higher DR across extensive WM regions correlated with cortical thinning in bilateral fronto-temporo-parietal regions. In mixed dementia patients, DR and DA changes across extensive WM regions correlated with cortical thinning in the bilateral fronto-temporo-parietal regions. Our findings showed that the relationships between GM and WM degeneration are distinct in pure ADD, pure SVaD, and mixed dementia, suggesting that different pathomechanisms underlie their correlations.

Complementary Characteristics of Correlation Patterns in Morphometric Correlation Networks of Cortical Thickness, Surface Area, and Gray Matter Volume.

Morphometric correlation networks of cortical thickness, surface area, and gray matter volume have statistically different structural topology. However, there is no report directly describing their correlation patterns in view of interregional covariance. Here, we examined the characteristics of the correlation patterns in three morphometric networks of cortical thickness, surface area, and gray matter volume using a Venn diagram concept across 314 normal subjects. We found that over 60% of all nonoverlapping correlation patterns emerged with divergent unique patterns, while there were 10% of all common edges in ipsilateral and homotopic regions among the three morphometric correlation networks. It was also found that the network parameters of the three networks were different. Our findings showed that correlation patterns of the network itself can provide complementary information when compared with network properties. We demonstrate that morphometric correlation networks of distinct structural phenotypes have different correlation patterns and different network properties. This finding implies that the topology of each morphometric correlation network may reflect different aspects of each morphometric descriptor.

Neural predictors of cognitive improvement by multi-strategic memory training based on metamemory in older adults with subjective memory complaints

Previous studies have indicated that memory training may help older people improve cognition. However, evidence regarding who will benefit from such memory trainings has not been fully discovered yet. Understanding the clinical and neural inter-individual differences for predicting cognitive improvement is important for maximizing the training efficacy of memory-training programs. The purpose of this study was to find the individual characteristics and brain morphological characteristics that predict cognitive improvement after a multi-strategic memory training based on metamemory concept. Among a total of 49 older adults, 39 participated in the memory-training program and 10 did not. All of them underwent brain MRIs at the entry of the training and received the neuropsychological tests twice, before and after the training. Stepwise regression analysis showed that lower years of education predicted cognitive improvement in the training group. In MRI, thinner cortices of precuneus, cuneus and posterior cingulate gyrus and higher white matter anisotropy of the splenium of corpus callosum predicted cognitive improvement in the training group. Old age, lower education level and individual differences in cortical thickness and white matter microstructure of the episodic memory network may predict outcomes following multi-strategic training.

Improved explanation of human intelligence using cortical features with second order moments and regression

BACKGROUND Cortical features derived from magnetic resonance imaging (MRI) provide important information to account for human intelligence. Cortical thickness, surface area, sulcal depth, and mean curvature were considered to explain human intelligence. One region of interest (ROI) of a cortical structure consisting of thousands of vertices contained thousands of measurements, and typically, one mean value (first order moment), was used to represent a chosen ROI, which led to a potentially significant loss of information. METHODS We proposed a technological improvement to account for human intelligence in which a second moment (variance) in addition to the mean value was adopted to represent a chosen ROI, so that the loss of information would be less severe. Two computed moments for the chosen ROIs were analyzed with partial least squares regression (PLSR). Cortical features for 78 adults were measured and analyzed in conjunction with the full-scale intelligence quotient (FSIQ). RESULTS Our results showed that 45% of the variance of the FSIQ could be explained using the combination of four cortical features using two moments per chosen ROI. Our results showed improvement over using a mean value for each ROI, which explained 37% of the variance of FSIQ using the same set of cortical measurements. DISCUSSION Our results suggest that using additional second order moments is potentially better than using mean values of chosen ROIs for regression analysis to account for human intelligence.

Decreased hemoglobin levels, cerebral small-vessel disease, and cortical atrophy: among cognitively normal elderly women and men.

BACKGROUND Decreased hemoglobin levels increase the risk of developing dementia among the elderly. However, the underlying mechanisms that link decreased hemoglobin levels to incident dementia still remain unclear, possibly due to the fact that few studies have reported on the relationship between low hemoglobin levels and neuroimaging markers. We, therefore, investigated the relationships between decreased hemoglobin levels, cerebral small-vessel disease (CSVD), and cortical atrophy in cognitively healthy women and men. METHODS Cognitively normal women (n = 1,022) and men (n = 1,018) who underwent medical check-ups and magnetic resonance imaging (MRI) were enrolled at a health promotion center. We measured hemoglobin levels, white matter hyperintensities (WMH) scales, lacunes, and microbleeds. Cortical thickness was automatically measured using surface based methods. Multivariate regression analyses were performed after controlling for possible confounders. RESULTS Decreased hemoglobin levels were not associated with the presence of WMH, lacunes, or microbleeds in women and men. Among women, decreased hemoglobin levels were associated with decreased cortical thickness in the frontal (Estimates, 95% confidence interval, -0.007, (-0.013, -0.001)), temporal (-0.010, (-0.018, -0.002)), parietal (-0.009, (-0.015, -0.003)), and occipital regions (-0.011, (-0.019, -0.003)). Among men, however, no associations were observed between hemoglobin levels and cortical thickness. CONCLUSION Our findings suggested that decreased hemoglobin levels affected cortical atrophy, but not increased CSVD, among women, although the association is modest. Given the paucity of modifiable risk factors for age-related cognitive decline, our results have important public health implications.