Fusiform company logo

The profile is currenly unclaimed by the seller. All information is provided by CB Insights.


Founded Year



Series B | Alive

Total Raised


Last Raised

$2.94M | 2 yrs ago

About Fusiform

Fusiform is a software company seeking to revolutionize the fabrication of orthotic devices to make them better and faster.

Fusiform Headquarter Location

916 North Charles St

Baltimore, Maryland, 21201,

United States


Predict your next investment

The CB Insights tech market intelligence platform analyzes millions of data points on venture capital, startups, patents , partnerships and news mentions to help you see tomorrow's opportunities, today.

Latest Fusiform News

COMT-Polymorphisms Modulated Functional Profile of the Fusiform Face Area Contributes to Face-Specific Recognition Ability

Feb 7, 2020

Abstract Previous studies have shown that face-specific recognition ability (FRA) is heritable; however, the neural basis of this heritability is unclear. Candidate gene studies have suggested that the catechol-O-methyltransferase (COMT) rs4680 polymorphism is related to face perception. Here, using a partial least squares (PLS) method, we examined the multivariate association between 12 genotypes of 4 COMT polymorphisms (rs6269-rs4633-rs4818-rs4680) and multimodal MRI phenotypes in the human fusiform face area (FFA), which selectively responds to face stimuli, in 338 Han Chinese adults (mean age 20.45 years; 135 males). The MRI phenotypes included gray matter volume (GMV), resting-state fractional amplitude of low-frequency fluctuations (fALFF), and face-selective blood-oxygen-level-dependent (BOLD) responses (FS). We found that the first COMT-variant component (PLS1) was positively associated with the FS but negatively associated with the fALFF in the FFA. Moreover, participants with the COMT heterozygous-HEA-haplotype showed higher PLS1 FFA-MRI scores, which were positively associated with the FRA in an old/new face recognition task, than those with the COMT homozygous HEA haplotype and HEA non-carriers, suggesting that individuals with an appropriate (intermediate) level of dopamine activity in the FFA might have better FRA. In summary, our study provides empirical evidence for the genetic and neural basis for the heritability of face recognition and informs the formation of neural module functional specificity. Introduction Face recognition is a highly developed skill in humans that shows considerable variability between individuals 1 , 2 , 3 , 4 , 5 . Previous behavioral genetic studies have shown that face recognition ability is heritable 4 , 6 ; however, the neural basis for this heritability is unclear. One candidate cortical region underlying inter-individual variability in face recognition is the human fusiform face area (FFA) especially in the right hemisphere, which shows more significant responses to visual face stimuli than to other non-face stimuli 7 and represents a critical hub in the face network 8 , 9 , 10 . More importantly, features of the FFA are associated with face-specific recognition ability (FRA) 2 , 3 . However, information about the effects of genetic variability on FFA structure and function and their relationship with face recognition ability remain limited. Previous twin studies have demonstrated the higher intra-pair similarity of cortical structure 11 and functional magnetic resonance imaging (fMRI) responses 11 , 12 in the occipitotemporal cortex encompassing the FFA in monozygotic twins than dizygotic twins. Additionally, neural responses to facial expressions are modulated by common genetic variance in a set of regions constituting the face network that include the FFA 13 . These studies suggest that the FFA may be modulated by genetic variations. However, previous studies have not specifically localized the FFA, and it is unclear whether the observed genetic effects are derived from changes in the FFA or other regions. Notably, a previous genome-wide association neuroimaging study explored specific genes contributing to the FFA but failed to identify any significant genome-level correlations between single nucleotide polymorphisms (SNPs) and FFA-responses to facial expressions 14 . One possible reason for this negative result is insufficient statistic power due to a relatively small number of participants (N < 300). Candidate gene studies have demonstrated the association of the catechol-O-methyltransferase (COMT) gene, located on chromosome 22q11.2, with face recognition 15 , 16 , 17 , 18 , 19 . The COMT gene codes for the COMT enzyme, a protein that degrades catecholamine neurotransmitters and modulates dopamine metabolisms 20 . Among the COMT polymorphisms, Val158Met (rs4680) has been most reported to play a role in face processing 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 . For example, individuals with the Met/Met (rs4680) genotype are more sensitive to configural changes in faces than those with other genotypes 15 . Healthy homozygous subjects with the Met allele of rs4680 are better at recognizing facial emotions 16 and show a stronger bias to perceive neutral faces as expressions of anger 21 than those with other genotypes. Previous studies also investigated the impact of the COMT gene on the brain activity during face emotional processing 22 , 23 , 24 , 25 and demonstrated that the COMT Met allele seems to contribute to neural substrates of negativity biases 22 , 23 , 24 , and the COMT Val allele is associated with increased amygdala activity in response to fearful/angry facial expressions 25 . It should be noted that several meta-analyses suggest that the effects of COMT Val158Met on cognitive abilities are either very small or nil due to insufficient statistical power, unknown moderators including genetic variations at different loci, publication bias, or effect heterogeneity 26 , 27 , 28 . Therefore, individual studies should pay more attention to the methodological rigor and scientific nature. Accumulating studies have revealed that there are multi-to-multi relationships between genotypes and brain phenotypes linked to learning and cognitive ability 29 , 30 , 31 . Previous studies have indicated that four COMT polymorphisms (rs6269, rs4633, rs4818, and rs4680) have a strong linkage disequilibrium and always function as haplotype blocks to affect dopamine (DA) degrading enzyme activity 32 , 33 . However, little is known about whether certain patterns of the 4 COMT polymorphisms exert significant effects on the FFA and therefore affect the face recognition ability. In this study, we employed a combined genetic and neuroimaging approach and incorporated three strategies to improve statistical power without dramatically increasing the number of participants. First, previous studies focused on a single measure of the FFA (i.e., either gray matter volume (GMV) or blood oxygen level-dependent [BOLD] responses); notably, multiple modals and joint modeling can improve sensitivity for identifying disease-associated biological pathways in common complex disorders 34 , 35 , 36 , 37 ; accordingly, this study used multimodal MRI measures for the FFA (multimodal FFA-MRI data), including GMV, fractional amplitude of low-frequency fluctuations (fALFF) from resting-state fMRI, and BOLD responses specific to faces (face selectivity, FS) from task-state fMRI. Second, the partial least squares (PLS) approach was applied to examine relationships between sets of multivariate measures. This approach has demonstrated utility in studies with relative small samples and problems with multicollinearity 37 , 38 , 39 , 40 . Third, bilateral FFAs are involved in face processing simultaneously 5 , 7 , 8 , notwithstanding some hemispheric lateralization in their size and functional recruitments 41 , 42 . We conducted analyses for the right FFA (rFFA) and the left FFA (lFFA) separately to validate the genetic variants that modulate the FFA. Using these three strategies, we identified profiles of 4 COMT polymorphisms that would be robustly associated with profiles of structural or functional FFA, in a population of Han Chinese individuals. Results Multivariate association between COMT polymorphisms and FFA phenotypes A PLS correspondence analysis (PLSCA) was conducted to examine whether the 12 COMT genotypes contributed to the 3-modal right FFA-MRI phenotype in 338 participants. The PLSCA yielded 3 sets of paired latent variables (LVs) capturing the rFFA-SNP association, ordered by the size of the explained variance (91.51%, 7.0%, and 1.49%). Each component (LV pair) was a linear combination of the weighted COMT SNP scores that most strongly correlated with weighted MRI scores and was orthogonal to each other. The omnibus p value of the model was 0.004 and only the top PLS component (PLS1) was significant (permutation test, p = 0.001). Thus, the PLS1 represented a significant association (r = 0.258, p < 0.001) between a specifically patterned COMT genotypic profile and a specifically patterned rFFA phenotype (middle panel of Fig. 1A ). The power analysis 43 for the PLS model (Supplementary Table  S2 ) indicated a power of 0.713 for the path between the rFFA and the COMT polymorphisms, which means that we have the probability of 71.3% to detect a significant association between the rFFA and the COMT polymorphisms, given that the association is really there. The component reliability test showed that the rs4680 AA (Met) genotype (salience = 0.417, BSR = 3.33, FDR corrected p = 0.007, 95%CI [0.17, 0.72]), rs4818 AA genotype (salience = 0.110, BSR = 2.40, FDR corrected p = 0.040, 95%CI [0.02, 0.20]), rs4818 AG genotype (salience = −0.132, BSR = −2.94, FDR corrected p = 0.013, 95%CI [−0.22, −0.04]), rs4633 TT genotype (salience = 0.370, BSR = 3.23, FDR corrected p = 0.007, 95%CI [0.15,0.59]), and rs6269 AG genotype (salience = −0.121, BSR = −2.66, FDR corrected p = 0.024, 95%CI [−0.21, −0.03]) reliably contributed to the PLS1 COMT-SNP component (left panel of Fig. 1A ); the FS (salience = 0.129, BSR = 3.21, FDR corrected p = 0.007, 95%CI [0.05,0.21]) and fALFF (salience = −0.132, BSR = −2.84, FDR corrected p = 0.004, 95%CI [−0.22, −0.04]) reliably contributed to the PLS1 rFFA-MRI component (right panel of Fig. 1A ). Moreover, the genetic regulation of the rFFA presented a pattern that rs4680 and rs4633 exhibited linear trend of modulation (from recessive homozygote to heterozygote and to dominant homozygote) on the PLS1 rFFA-MRI profile, whereas rs6269 and rs4818 showed a nonlinear relationship with the PLS1 rFFA-MRI profile. Figure 1 Partial least squares correspondence analysis (PLSCA) between right FFA-MRI (rFFA-MRI) phenotypes and COMT polymorphisms. (A) The first pair of latent variables (LVs) from the PLSCA. Left: weighted saliences and 95% confidence interval for each COMT genotype; right: weighted saliences (weights) and 95% confidence interval for each modal of the rFFA. The rFFA (cluster in red) was defined as a set of contiguous voxels (233 voxels) that were significantly activated in response to faces vs. objects (Z > 2.58, one-tailed, p < 0.005, uncorrected) in the lateral mid- and posterior fusiform gyrus in the right hemisphere. (B) The first partial least squares component (PLS1) identified a profile (linear combination) of COMT genotypes that were positively correlated with face selectivity (FS), negatively correlated with fALFF, and not correlated with GMV of the rFFA. Finally, because the FFA has been demonstrated to be involved in the process of recognizing familiar and unfamiliar faces 44 , 45 , 46 , 47 , we examined whether there is a difference in modulation of the COMT genotypic profile between old (familiar) face recognition and new (relatively unfamiliar) face recognition. To control the effect of response bias on the accuracy 48 , we calculated the response bias [c = −Z(hit rate) + Z(false alarm rate)] 49 and regressed the bias out of the hit rate and correct rejection rate to obtain the old-face recognition score and new-face recognition score. A significant association was found between the new-face recognition score and the bilateral PLS1 FFA-MRI scores (rFFA: r = 0.162, uncorrected p = 0.003; lFFA: r = 0.149, uncorrected p = 0.006; Fig. 3B ), but not between the old-face recognition score and the bilateral PLS1 FFA-MRI scores (Fig. 3C , Supplementary Table  S3 and Fig. S5 ) when controlling for age, sex, and Raven scores. Moreover, the association between the PLS1 FFA-MRI profile and the new-face recognition score was significantly greater than that between the PLS1 FFA-MRI profile and the old-face recognition score (rFFA: z = 2.70, two-tailed p = 0.007, Cohen’d = 0.33, 95%CI = [0.08, 0.39]; lFFA: z = 2.95. two-tailed p = 0.003, Cohen’d = 0.36, 95%CI = [0.13,0.43]), suggesting that the COMT-modulated FFA-MRI profile was more sensitive to the new-face recognition ability than to the old-face recognition ability. Discussion In the present study, we explored the neural basis of the heritability of face recognition by investigating associations between 12 COMT genotypes and 3-modal FFA-MRI phenotypes. The results revealed that a profile of face-selective responses and resting-state neural fluctuations of the bilateral FFA were modulated by a profile of COMT polymorphisms. Participants with the COMT heterozygous-HEA-haplotype showed higher PLS1 FFA-MRI scores, which represented an increased FS and a decreased fALFF and were associated with a better face-specific recognition ability, than those with the COMT homozygous HEA haplotype and HEA non-carrier, suggesting that participants with an appropriate (intermediate) level of dopamine in the bilateral FFA may have better face-specific recognition ability. Our study provides empirical evidence of genetic modulation of a functionally-specialized cortical region in humans. Previous studies have found that participants with higher face selectivity in the FFA have higher face recognition ability 2 , 3 , and this association is domain-specific 3 . In this study, we revealed that a profile (linear combination) of higher face selectivity BOLD responses, as well as lower resting-state activities in the bilateral FFA, was associated with a higher face recognition (especially the new-face recognition) ability, and the association could not be accounted for by FFA responses to objects such as flowers, suggesting that the face-specific recognition ability might be modulated by a certain covariation pattern of resting-state and task-state activity in the FFA, which was modulated by a certain profile of the COMT polymorphisms. The COMT Val/Met (rs4680) polymorphism has been suggested to play a role in face perception 15 , 16 , 17 , 18 , 19 , 22 . Our study extends previous findings by identifying a neural site where COMT affects face recognition. Specifically, we investigated a broader modulation pattern of COMT genetic effects on the FFA by including multiple SNPs. We found that individuals with the heterozygous HEA haplotype exhibited higher face-selective responses but smaller fALFF. In contrast, individuals with the homozygous HEA haplotype or HEA non-carriers had lower face-selective responses and larger fALFF. Because COMT encodes catechol-O-methyltransferase, a protein that degrades catecholamine neurotransmitters and modulates dopamine metabolism 20 , neural activity (including the face-recognition-task-state and resting-state activities) in the FFA is likely modulated by changes in dopamine. Moreover, we found that an intermediate level of dopamine in the FFA was associated with higher face-selective responses (in the task state) and smaller fALFF (in the resting state), and the profile of higher face-selective responses and lower fALFF in the FFA was associated with better face-specific recognition ability. Thus, an intermediate level of dopamine in the FFA is likely beneficial for behavioral performance in face recognition. That is to say, participants with an appropriate level of dopamine (i.e., the heterozygous HEA haplotype carries) in the FFA possess a functional characteristic profile of higher face-selective responses and lower intrinsic FFA activities, therefore, they might have better face-specific recognition ability. Previous studies have rarely reported the direct impact of COMT polymorphisms on resting-state FFA activity in healthy people. Some studies focusing on the interaction effect between the COMT rs4680 and other dopamine-function related generic variants imply that dopamine-related neurons may affect the baseline brain resting state activity by synchronizing the low-frequency oscillation 50 . Moreover, dopamine could stabilize (keep the balance between facilitating and suppressing) neural signaling in brain networks responsible for behavior-stimuli oriented attention 51 , 52 . Thus, an appropriate level of dopamine may improve the functional segregation-integration of a functional-specialized region by maintaining excitation-inhibition stability of neural signal transduction to ensure that individuals with the HEA heterozygous haplotype have a relatively higher rFFA activation in face recognition tasks (which has been found to be related to a better face-specific recognition ability 2 , 3 ), as well as a relatively lower rFFA activity during resting state (which was inferred from previous research that lower metabolism in the right temporal-parietal cortex is also associated with better cognitive reserve and memory ability in aging people 53 ). This is consistent with previous studies demonstrating associations between dopamine hyper- or hypo-function in cortical or subcortical regions and cognitive dysfunctions 54 , 55 , 56 . Therefore, it would be interesting to examine the level of dopamine in the FFA in individuals with developmental prosopagnosia 10 , 57 , 58 , 59 , who show specific deficits in face recognition with unclear genetic basis 60 and those with mental disorders who might have disturbances in COMT modulation of face processing 16 , 18 , 21 , 23 . Our study found that the 4 COMT-polymorphisms-modulated PLS1 rFFA-MRI profile mainly regulated the recognition ability of new faces. Although previous studies have revealed that it takes more time and is more likely to make mistakes to recognize faces of different identities than faces of same identities 47 , there seems to be no difference in the increase of neural response activity of the FFA when recognizing familiar and unfamiliar faces 44 , 47 . Compared with recognizing unfamiliar faces, recognizing familiar faces involves more activity in the anterior and mid-fusiform gyrus and extended face areas 44 , 45 . In our study, the stimuli were not familiar to the subjects, and the social and emotional information involved in the face stimuli was relatively sparse; therefore, the recognition difference between the old face and the new face might be mainly reflected in the familiarity and unfamiliarity of the facial features. We found that the COMT-polymorphisms modulated PLS1 rFFA-MRI profile was sensitive to the recognition of new faces, but not to that of the old faces. The findings may be related to the fact that the novel signals can trigger the dopaminergic metabolic system 55 in the task-related brain areas and enhance perception 61 , 62 . The FFA is sensitive to the static or invariant properties of faces 63 . The posterior superior temporal sulcus (pSTS) is responsible for dynamic features 63 , 64 and sensitive to emotional information 65 , 66 , 67 from faces, and the prefrontal-amygdala circuit is involved in social and emotional regulation in face processing 23 . In fact, face perception as a whole is a network involving multiple brain regions 8 , 9 . Studies have found that the COMT Met allele is related to the modulation of neural substrates of negative face emotional bias 22 , 23 . Therefore, in future studies, it would be interesting to evaluate the role of multiple COMT polymorphisms in the core face network and its relation with face-specific recognition ability and face-emotion recognition. This study has some limitations. First, although we have adopted multiple strategies to improve the statistical power, the findings still need to be validated using data from other cohorts of participants. Second, measurements for the face recognition ability and the general mental ability are limited to the face memory test and the Raven test, respectively. More multivariate measurements could be introduced in the future to improve the reliability of the individual face recognition ability and general mental ability, and further elaborate the specificity of the effect of COMT on the face recognition ability. Third, all participants had the same ethnicity (i.e., Han Chinese); therefore, how the ethnicity of the individual participants affected the results needs to be examined using data from multiple ethnicities in the future. Fourth, the Caucasian children’ faces were used as stimuli to define the FFA in Han Chinese participants. As demonstrated in previous studies, there is an other-race effect in face recognition; therefore, how our findings are affected by stimulus attributes still needs to be investigated in further studies. In conclusion, our study demonstrated that the COMT gene affected both resting- and task-state neural activity in the FFA through rs6269-rs4633-rs4818-rs4680 haplotypes, and the COMT-polymorphisms-modulated FFA was associated with face-specific recognition ability. The present findings provide a basis for the behavioral observation of face-specific recognition heritability and the approach used in this study might be s a feasible method for exploring and validating the neural basis of heritable behavioral performance. In addition, this study may inform the genetic etiology of developmental prosopagnosia and other mental disorders showing abnormities in face processing. Methods Participants Three hundred and sixty-two Han Chinese college students (mean age, 20.47 ± 0.96 years; 144 males) were recruited from Beijing Normal University (BNU) in Beijing, China through campus advertisement. Ethnicity was determined by the ethnic information displayed on the ID card of the participants. All participants had normal or corrected-to-normal vision, and none of them reported a history of neurological or psychiatric disorders. All the experimental protocol and procedures were approved by the Committee for Protecting Human and Animal Subjects of the Faculty of Psychology at Beijing Normal University. All experiments of the study were conducted in accordance with the relevant guidelines and regulations of Beijing Normal University’s Institutional Review Board (Human Subjects Division). All participants gave written informed consent prior to their participation in this study. MRI Data acquisition Functional and structural MRI was performed at the BNU Imaging Center for Brain Research on a Siemens 3 T scanner (MAGENTOM Trio, a Tim system) with a 12-channel phased-array head coil. Resting-state fMRI was acquired using a T2*-weighted gradient-echo echo-planar-imaging (GRE-EPI) sequence (repetition time [TR] = 2000 ms, echo time [TE] = 30 ms, flip angle = 90°, number of slices = 33, voxel size = 3.125 × 3.125 × 3.6 mm3). Resting-state scanning lasted for 8 min and consisted of 240 contiguous volumes. Participants were instructed to relax and remain still with their eyes closed during resting state scanning 9 . A GRE-EPI sequence was also used to acquire task-state fMRI (TR = 2000 ms, TE = 30 ms, number of slices = 30, voxel size = 3.125 × 3.125 × 4.8 mm3). A high-resolution structural T1-weighted image was acquired with a 3D magnetization-prepared rapid acquisition gradient echo (MP-RAGE) sequence (TR = 2530 ms, TE = 3.39 ms, inversion time = 1100 ms, flip angle = 7°, matrix = 256 × 256, number of slices = 128, voxel size = 1 × 1 × 1.33 mm3). Earplugs were used to attenuate scanner noise, and a foam pillow and extendable padded head clamps were used to restrain head motion 3 . None of the participants were excluded due to excessive head motion (>2 mm in translation or 2° in rotation) during MRI scanning. MRI Data processing Structural MRI Voxel-based morphometry (VBM) was performed using the SPM8 (Statistical Parametric Mapping, Wellcome Department of Imaging Neuroscience, London, UK). First, unified segmentation in SPM8 was used to segment T1-weighted anatomical images into gray matter (GM), white matter (WM), and cerebral spinal fluid (CSF). Second, the Diffeomorphic Anatomical Registration through Exponential Lie algebra (DARTEL) registration method 68 was used to build a study-specific GM template and normalize individual GM images into the template. Third, GM voxel values were modulated by multiplying the Jacobian determinants derived from the normalization procedure in order to preserve the tissue volume of each structure. Fourth, modulated GM images were smoothed using an 8-mm full width at half maximum (FWHM) isotropic Gaussian kernel. Finally, the modulated images were masked using absolute masking with a threshold of 0.2 to exclude noisy voxels. Masked-modulated GM images were used for further statistical analyses 3 , 69 . Task-state fMRI A dynamic face localizer was used to define the face-selective area and derive FS 3 , 63 . Movie clips of close-up faces of 7 Caucasian children (filmed when children were dancing or playing), objects (moving toys), scenes (moving view of a suburb, canyons, or tunnels), and scrambled objects (constructed by scrambling each frame of the object movie clips) were included to examine category selectivity in the ventral visual cortex 63 . Participants were instructed to passively view movie clips during scanning. Because Adults’ FFA have been effectively identified by children’ faces in previous studies 3 , 5 , 63 , 65 , 67 and the FFA is sensitive to the invariant properties of faces, but not to the dynamic or emotional information of faces 65 , 66 , 67 , 70 , we expected that the dynamic children’ faces would accurately and unbiasedly localize the FFA in adults. In total, participants completed three 198-s runs. Each run consisted of 2 block sets intermixed with 3 18-s rest blocks at the beginning, middle, and end of the run. Each block set contained four 18-s (including six 3-s movie clips) blocks corresponding to 4 stimulus categories. The order of category blocks in each run was palindromic and randomized across runs (Fig. 4A ). Figure 4 (A) Illustration of scanning sequence for the functional MRI and localization of dynamic face, object, scene, and scramble. (B) Example stimuli and trial types in the old/new recognition task. Participants studied a single image (either a face or a flower), and they were then shown a series of individual images of the corresponding type and asked to indicate which of the images had been shown in the study segment. Functional images were analyzed with the FMRI Expert Analysis Tool version 5.98, part of FMRIB’s Software Library ( www.fmrib.ox.ac.uk/fsl ). Preprocessing included head-motion correction, brain extraction, spatial smoothing (Gaussian kernel; FWHM = 6 mm), grand-mean intensity normalization, and high-pass temporal filtering (120-s cutoff). A first-level analysis was performed separately on each run for each participant using FMRIB’s Improved Linear Model with a local autocorrelation correction. A boxcar kernel was convolved with a gamma hemodynamic response function and its temporal derivative was used to model BOLD signal. Six parameters from motion-correction were also included in the model as regressors of no interest to account for the effect of head movement. A second-level fixed-effect analysis was conducted to combine all runs. Specifically, the parameter (i.e., beta) image from the first-level analysis was first aligned to the individual’s structural images using FMRIB’s linear image registration tool with 6 degrees of freedom and then warped to the MNI152 template using FMRIB’s nonlinear image registration tool with default parameters. The spatially normalized parameter images (resampled to 2-mm isotropic voxels) were then summarized across all runs using a fixed-effect model 3 , 9 , 69 . In this study, statistical images from the contrast between faces and objects were used to define the right fusiform face area and derive its face selectivity (FS), which was calculated as the average z-score (faces vs. objects) across all voxels in the defined face-selective region. Resting-state fMRI Resting-state functional images were preprocessed with a procedure similar to task-state fMRI. Motion correction parameters, mean signals of the CSF and the WM, and the first derivatives of these signals were regressed out to remove the signal fluctuations caused by head motion, cardiac cycle and respiration. Then, the fractional amplitude of low-frequency fluctuations (fALFF), which is defined as the fractional sum of the amplitudes within the low frequency range divided by the sum of amplitude across the entire frequency range (0–0.25 Hz) 71 , was calculated for each GM voxel. Participant-level voxel-wise fALFF maps were further standardized by subtracting the mean whole-brain voxel-wise fALFF from the participant-level voxel-wise fALFF and dividing it by the standard deviation. Finally, the standardized fALFF was normalized to the MNI152 space with the same normalization procedure as that used in the task-state fMRI analysis. Identification of fusiform face area A threshold of Z > 2.58 (one-tailed p < 0.005, uncorrected) was used to define the bilateral FFA based on the random-effect group-level Z-statistic image for faces vs. objects 3 . Specifically, all contiguous voxels that surpassed the threshold within the posterior fusiform area (defined by the anatomical automatic labeling [AAL] template; http://www.gin.cnrs.fr/AAL ) were defined as the FFA (right panel of Figs. 1A and S4 ). Genotyping Genomic DNA was extracted from peripheral blood samples of each subject using QuickGene-Mini80 equipment and the QuickGene DNA whole blood kit S (Fujifilm). Sixty-four cognition-related candidate SNPs (including the four COMT polymorphisms of rs6269, rs4633, rs4818, and rs4680 examined in this study; please refer to Supplementary Table  S1 for details of other SNPs) were automatically genotyped using a customized 64 TaqManOpenArray GT Kit (Applied Biosystems; Foster City, CA, USA). Quality control was performed using PLINK software version 1.07 72 . All the 4 COMT SNPs met the following criteria: genotyping call rate >0.95, minor allele frequency (MAF) > 0.05, and Hardy–Weinberg equilibrium (HWE) P > 0.05. Participants were excluded from analyses if they had a missing genotype rate of >25% (i.e., the participants who had one of the four COMT polymorphisms missing). Twenty-four participants were excluded, and the remaining 338 participants (mean age, 20.45 ± 0.96 years; 135 males) were included in the data analyses. Allele frequencies in this sample were very similar to those in the Han Chinese sample from the HapMap dataset (HapMap Data Release 27 Phase II + III). Face recognition task We used an old/new cognitive memory task to measure the face-specific recognition ability (FRA). There was a face block (containing grayscale face pictures of adult Chinese faces with the external contours removed) and a flower block (containing grayscale pictures of common flowers) in this task (face and flower stimuli were shown in Fig. 4B ) that were counterbalanced across participants. Each block consisted of one study segment and one test segment. In the study segment, each stimulus (picture) was presented twice (1 second each time), with an inter-stimulus interval of 0.5 seconds. Then, in the test segment, half of the studied images were randomly intermixed with new images from the same category. For each stimulus picture, participants were instructed to indicate whether the image had been shown in the study segment. For a participant, a face recognition accuracy score was calculated as the average proportion of hits and correct rejections for faces, and an object (flower) recognition accuracy score was calculated as the average proportion of hits and correct rejections for flowers. The FRA was calculated as the normalized residual of the face recognition score after regressing out the flower recognition score 3 . D’ (sensitivity index) 48 was also calculated for face recognition and flower recognition. Statistical analyses The PLS correspondence analysis (PLSCA) was conducted using the R Package ExPosition ( https://cran.r-project.org/web/packages/ExPosition/ ). PLS is particularly suitable for identifying associations between 2 sets of variables, especially when variables of each set are highly interdependent or multi-collinear 38 , 39 , 40 , 73 , 74 . The PLS approach maximizes the covariance between a set of variables (a linear combination of X measures) and another set of variables (a linear combination of Y measures). More specifically, the cross-correlation matrix between measures of SNPs and measures of brain-MRI was first computed. Then, a set of paired latent variables (LVs), which were uncorrelated (orthogonal) with each other and ordered by the explained covariance of brain and SNP measures, were generated via singular value decomposition. LVs are linearly weighted by the raw variables, and these weights or saliences measure the contribution of each raw variable to the LV 73 . Finally, the significance of the variance of singular values was tested by comparison with the distribution of variance arising from random permutation tests (1000 repetitions), and a bootstrap procedure (500 times) was used to test the reliability (i.e, bootstrap ratio, BSR) of each variable in the significant PLS component 40 , 73 . SNPs-FFA association analysis The SNPs-FFA multivariate association analysis was used to examine whether specific patterns of COMT polymorphisms contributed to the multimodal FFA-MRI phenotypes. First, GM, fALFF, and FS within the right FFA and the left FFA ROI were extracted for each participant. Then, each SNP was coded in accordance with genotypic (codominant) models 37 , 75 : each SNP was coded into 3 genotypes (e.g., AA, AG, or GG) as categorical variables, and each variable was weighted according to the information it provided. The weight of a variable was defined as the inverse of its relative frequency because a rare variable provides more information than does a frequent variable 37 , 75 . This fully categorical coding scheme allows us to detects linear and nonlinear relationships within (e.g., genotypic) and between datasets 37 . Thus, treating each genotype as a level of a categorical variable is suitable for genetic association studies, especially when the inheritance pattern, direction, and/or size of the effect are unknown 37 , 76 . Accordingly, a matrix of measures of 12 COMT genotypes estimated in the 338 participants and a matrix of measures of 3-modal FFA-MRI phenotype estimated in the 338 participants were entered into the PLS analysis. Haplotype analysis A haplotype block analysis 77 was conducted to investigate how SNPs in strong linkage disequilibrium work as blocks to modulate FFA phenotypes. Haplotype blocks of SNPs affecting the FFA phenotype were assessed and illustrated using Haploview software version 4.2 ( https://sourceforge.net/projects/haploview/ ) with the solid spine of the linkage disequilibrium method and parameter 0.80. Associations of haplotypes with the FFA were examined using R software version 3.5.2 with generalized linear models. References 1. Crouch, D. J. M. et al. Genetics of the human face: Identification of large-effect single gene variants. Proc. Natl Acad. Sci. USA 115, E676–e685, https://doi.org/10.1073/pnas.1708207114 (2018). Additional information Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Supplementary information Rights and permissions Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ .

Fusiform Web Traffic

Page Views per User (PVPU)
Page Views per Million (PVPM)
Reach per Million (RPM)
CBI Logo

Fusiform Rank

Discover the right solution for your team

The CB Insights tech market intelligence platform analyzes millions of data points on vendors, products, partnerships, and patents to help your team find their next technology solution.

Request a demo

CBI websites generally use certain cookies to enable better interactions with our sites and services. Use of these cookies, which may be stored on your device, permits us to improve and customize your experience. You can read more about your cookie choices at our privacy policy here. By continuing to use this site you are consenting to these choices.