deepPGSegNet: MRI-based pituitary gland segmentation using deep learning.

Introduction: In clinical research on pituitary disorders, pituitary gland (PG) segmentation plays a pivotal role, which impacts the diagnosis and treatment of conditions such as endocrine dysfunctions and visual impairments. Manual segmentation, which is the traditional method, is tedious and susceptible to inter-observer differences. Thus, this study introduces an automated solution, utilizing deep learning, for PG segmentation from magnetic resonance imaging (MRI). Methods: A total of 153 university students were enrolled, and their MRI images were used to build a training dataset and ground truth data through manual segmentation of the PGs. A model was trained employing data augmentation and a three-dimensional U-Net architecture with a five-fold cross-validation. A predefined field of view was applied to highlight the PG region to optimize memory usage. The model's performance was tested on an independent dataset. The model's performance was tested on an independent dataset for evaluating accuracy, precision, recall, and an F1 score. Results and discussion: The model achieved a training accuracy, precision, recall, and an F1 score of 92.7%, 0.87, 0.91, and 0.89, respectively. Moreover, the study explored the relationship between PG morphology and age using the model. The results indicated a significant association between PG volume and midsagittal area with age. These findings suggest that a precise volumetric PG analysis through an automated segmentation can greatly enhance diagnostic accuracy and surveillance of pituitary disorders.

Association between cortical thickness and cognitive ability in very preterm school-age children.

Very preterm children, born before 32 weeks of gestation, are at risk for impaired cognitive function, mediated by several risk factors. Cognitive impairment can be measured by various neurodevelopmental assessments and is closely associated with structural alterations of brain morphometry, such as cortical thickness. However, the association between structural alterations and high-order cognitive function remains unclear. This study aimed to investigate the neurodevelopmental associations between brain structural changes and cognitive abilities in very preterm and full-term children. Cortical thickness was assessed in 37 very preterm and 24 full-term children aged 6 years. Cortical thickness analysis of structural T1-weighted images was performed using Advanced Normalization Tools. Associations between cortical thickness and the Wechsler Intelligence Scale for Children were evaluated by regression analysis based on ordinary least square estimation. Compared with full-term children, very preterm children showed significant differences in cortical thickness, variously associated with cognitive abilities in several brain regions. Perceptual reasoning indices were broadly correlated with cortical thickness in very preterm and full-term children. These findings provide important insights into neurodevelopment and its association with cortical thickness, which may serve as a biomarker in predictive models for neurodevelopmental diagnosis of high-order cognitive function.

Predicting mild cognitive impairments from cognitively normal brains using a novel brain age estimation model based on structural magnetic resonance imaging.

Brain age prediction is a practical method used to quantify brain aging and detect neurodegenerative diseases such as Alzheimer's disease (AD). However, very few studies have considered brain age prediction as a biomarker for the conversion of cognitively normal (CN) to mild cognitive impairment (MCI). In this study, we developed a novel brain age prediction model using brain volume and cortical thickness features. We calculated an acceleration of brain age (ABA) derived from the suggested model to estimate different diagnostic groups (CN, MCI, and AD) and to classify CN to MCI and MCI to AD conversion groups. We observed a strong association between ABA and the 3 diagnostic groups. Additionally, the classification models for CN to MCI conversion and MCI to AD conversion exhibited acceptable and robust performances, with area under the curve values of 0.66 and 0.76, respectively. We believe that our proposed model provides a reliable estimate of brain age for elderly individuals and can identify those at risk of progressing from CN to MCI. This model has great potential to reveal a diagnosis associated with a change in cognitive decline.

DeepParcellation: A novel deep learning method for robust brain magnetic resonance imaging parcellation in older East Asians.

Accurate parcellation of cortical regions is crucial for distinguishing morphometric changes in aged brains, particularly in degenerative brain diseases. Normal aging and neurodegeneration precipitate brain structural changes, leading to distinct tissue contrast and shape in people aged >60 years. Manual parcellation by trained radiologists can yield a highly accurate outline of the brain; however, analyzing large datasets is laborious and expensive. Alternatively, newly-developed computational models can quickly and accurately conduct brain parcellation, although thus far only for the brains of Caucasian individuals. To develop a computational model for the brain parcellation of older East Asians, we trained magnetic resonance images of dimensions 256 × 256 × 256 on 5,035 brains of older East Asians (Gwangju Alzheimer's and Related Dementia) and 2,535 brains of Caucasians. The novel N-way strategy combining three memory reduction techniques inception blocks, dilated convolutions, and attention gates was adopted for our model to overcome the intrinsic memory requirement problem. Our method proved to be compatible with the commonly used parcellation model for Caucasians and showed higher similarity and robust reliability in older aged and East Asian groups. In addition, several brain regions showing the superiority of the parcellation suggest that DeepParcellation has a great potential for applications in neurodegenerative diseases such as Alzheimer's disease.

Involvement of the intrinsic functional network of the red nucleus in complex behavioral processing.

Previous studies suggested the possibility that the red nucleus (RN) is involved in other cognitive functions than motion per se, even though such functions have yet to be clarified. We investigated the activation of RN during several tasks and its intrinsic functional network associated with social cognition and musical practice. The tasks included finger tapping, n-back, and memory recall tasks. Region of interest for RN was identified through those tasks, anatomical information of RN, and a brain atlas. The intrinsic functional network was identified for RN by an analysis of connectivity between RN and other regions typically involved in seven known resting state functional networks with RN used as the seed region. Association of the RN network with a psychological trait of the interpersonal reactivity index and musical training years revealed subnetworks that included empathy related regions or music practice related regions. These social or highly coordinated motor activity represent the most complex functions ever known to involve the RN, adding further evidence for the multifunctional roles of RN. These discoveries may lead to a new direction of investigations to clarify probable novel roles for RN in high-level human behavior.

Effects of Physiological Signal Removal on Resting-State Functional MRI Metrics.

Resting-state fMRIs (rs-fMRIs) have been widely used for investigation of diverse brain functions, including brain cognition. The rs-fMRI has easily elucidated rs-fMRI metrics, such as the fractional amplitude of low-frequency fluctuation (fALFF), regional homogeneity (ReHo), voxel-mirrored homotopic connectivity (VMHC), and degree centrality (DC). To increase the applicability of these metrics, higher reliability is required by reducing confounders that are not related to the functional connectivity signal. Many previous studies already demonstrated the effects of physiological artifact removal from rs-fMRI data, but few have evaluated the effect on rs-fMRI metrics. In this study, we examined the effect of physiological noise correction on the most common rs-fMRI metrics. We calculated the intraclass correlation coefficient of repeated measurements on parcellated brain areas by applying physiological noise correction based on the RETROICOR method. Then, we evaluated the correction effect for five rs-fMRI metrics for the whole brain: FC, fALFF, ReHo, VMHC, and DC. The correction effect depended not only on the brain region, but also on the metric. Among the five metrics, the reliability in terms of the mean value of all ROIs was significantly improved for FC, but it deteriorated for fALFF, with no significant differences for ReHo, VMHC, and DC. Therefore, the decision on whether to perform the physiological correction should be based on the type of metric used.

A 16-Channel Dipole Antenna Array for Human Head Magnetic Resonance Imaging at 10.5 Tesla.

For ultra-high field and frequency (UHF) magnetic resonance imaging (MRI), the associated short wavelengths in biological tissues leads to penetration and homogeneity issues at 10.5 tesla (T) and require antenna transmit arrays for efficiently generated 447 MHz B1+ fields (defined as the transmit radiofrequency (RF) magnetic field generated by RF coils). Previously, we evaluated a 16-channel combined loop + dipole antenna (LD) 10.5 T head array. While the LD array configuration did not achieve the desired B1+ efficiency, it showed an improvement of the specific absorption rate (SAR) efficiency compared to the separate 8-channel loop and separate 8-channel dipole antenna arrays at 10.5 T. Here we compare a 16-channel dipole antenna array with a 16-channel LD array of the same dimensions to evaluate B1+ efficiency, 10 g SAR, and SAR efficiency. The 16-channel dipole antenna array achieved a 24% increase in B1+ efficiency in the electromagnetic simulation and MR experiment compared to the LD array, as measured in the central region of a phantom. Based on the simulation results with a human model, we estimate that a 16-channel dipole antenna array for human brain imaging can increase B1+ efficiency by 15% with similar SAR efficiency compared to a 16-channel LD head array.

Multi-Racial Normative Data for Lobar and Subcortical Brain Volumes in Old Age: Korean and Caucasian Norms May Be Incompatible With Each Other†.

Brain aging is becoming an increasingly important topic, and the norms of brain structures are essential for diagnosing neurodegenerative diseases. However, previous studies of the aging brain have mostly focused on Caucasians, not East Asians. The aim of this paper was to examine ethnic differences in the aging process of brain structures or to determine to what extent ethnicity affects the normative values of lobar and subcortical volumes in clinically normal elderly and the diagnosis in multi-racial patients with Alzheimer's disease (AD). Lobar and subcortical volumes were measured using FreeSurfer from MRI data of 1,686 normal Koreans (age range 59-89) and 851 Caucasian, non-Hispanic subjects in the ADNI and OASIS datasets. The regression models were designed to predict brain volumes, including ethnicity, age, sex, intracranial volume (ICV), magnetic field strength (MFS), and MRI scanner manufacturers as independent variables. Ethnicity had a significant effect for all lobar (|ß| > 0.20, p < 0.001) and subcortical regions (|ß| > 0.08, p < 0.001) except left pallidus and bilateral ventricles. To demonstrate the validity of the z-score for AD diagnosis, 420 patients and 420 normal controls were selected evenly from the Korean and Caucasian datasets. The four validation groups divided by race and diagnosis were matched on age and sex using a propensity score matching. We analyzed whether and to what extent the ethnicity adjustment improved the diagnostic power of the logistic regression model that was built using the only z-scores of six regions: bilateral temporal cortices, hippocampi, and amygdalae. The performance of the classifier after ethnicity adjustment was significantly improved compared with the classifier before ethnicity adjustment (ΔAUC = 0.10, D = 7.80, p < 0.001; AUC comparison test using bootstrap). Korean AD dementia patients may not be classified by Caucasian norms of brain volumes because the brain regions vulnerable to AD dementia are bigger in normal Korean elderly peoples. Therefore, ethnicity is an essential factor in establishing normative data for regional volumes in brain aging and applying it to the diagnosis of neurodegenerative diseases.

Brain Plasticity Reflects Specialized Cognitive Development Induced by Musical Training.

Learning a musical instrument requires a long period of training and might induce structural and functional changes in the brain. Previous studies have shown brain plasticity resulting from training with a musical instrument. However, these studies did not distinguish the effects on brain plasticity of specific musical instruments as they examined the brain of musicians who had learned a single musical instrument/genre and did not control for confounding factors, such as common or interactive effects involved in music training. To address this research gap, the present work investigated musicians who had experience with both a piano and a wind instrument, for example, flute, trumpet, clarinet etc. By examining the difference between the 2 musical instruments in the same subject, we avoided the effects common to all musical instruments and the confounding factors. Therefore, we identified several high-tier brain areas displaying a brain plasticity specific to each musical instrument. Our findings show that learning a musical instrument might result in the development of high cognitive functions reflecting the skills/abilities unique to the instrument played.

Assessment of olfactory information in the human brain using 7-Tesla functional magnetic resonance imaging.

Olfaction could prove to be an early marker of neurodegenerative diseases, including Alzheimer's and Parkinson's diseases. To use olfaction for disease diagnosis, elucidating the standard olfactory functions in healthy humans is necessary. However, the olfactory function in the human brain is less frequently assessed because of methodological difficulties associated with olfactory-related cerebral areas. Using ultra-high fields (UHF), functional magnetic resonance imaging (fMRI) with high spatial resolution and sensitivity may allow for the measurement of activation in the cerebral areas. This study aimed to apply 7-Tesla fMRI to assess olfactory function in the human brain by exposing individuals to four different odorants for 8 s. We found that olfactory stimulation mainly activated the piriform and orbitofrontal cortex in addition to the amygdala. Among these regions, univariate fMRI analysis indicated that subjective odor intensity significantly correlated with the averaged fMRI signals in the piriform cortex but not with subjective hedonic tone in any region. In contrast, multivariate fMRI analysis showed that subjective hedonic tone could be discriminated from the fMRI response patterns in the posterior orbitofrontal cortex. Thus, the piriform cortex is mainly associated with subjective odor intensity, whereas the posterior orbitofrontal cortex are involved in the discrimination of the subjective hedonic tone of the odorant. UHF-fMRI may be useful for assessing olfactory function in the human brain.

Cerebral artery segmentation based on magnetization-prepared two rapid acquisition gradient echo multi-contrast images in 7 Tesla magnetic resonance imaging.

Cerebral artery segmentation plays an important role in the direct visualization of the human brain to obtain vascular system information. On ultra-high field magnetic resonance imaging, cerebral arteries appearing hyperintense on T1 weighted (T1w) images could be segmented from brain tissues such as gray and white matter. In this study, we propose an automated method to segment the cerebral arteries using multi-contrast images including T1w images of a magnetization-prepared two rapid acquisition gradient echo (MP2RAGE) sequence at 7 T. The proposed method, termed MP2rase-CA (MP2rage based RApid SEgmentation Cerebral Artery), employed a seed-based region-growing strategy and Frangi filtering as well as our brain tissue segmentation (MP2rase Brain Tissue). Time-of-flight (TOF) magnetic resonance angiography (MRA) images were obtained as a reference to evaluate the MP2rase-CA. We successfully performed vessel segmentations, from T1w MP2RAGE images, which mostly overlapped with the segmentations of large cerebral arteries from the TOF-MRA. We also investigated the effect of the large cerebral arteries on spatial transformation of anatomical images to standard coordinate space using vessel segmentation by MP2rase-CA. As a result, the T1w image without the cerebral arteries by MP2rase-CA showed better agreement with the standard atlas compared with the T1w image containing the arteries. In addition, voxel-based morphology showed significant differences between T1w images with and without cerebral arteries in brain areas nearby large arteries. Thus, because MP2rase-CA using MP2RAGE images can obtain brain tissue anatomical information as well as relatively large cerebral artery information without need for additional structure acquisition, it is useful and time saving for functional and structural studies.

Measurement of ultra-fast signal progression related to face processing by 7T fMRI.

Given that the brain is a dynamic system, the temporal characteristics of brain function are important. Previous functional magnetic resonance imaging (fMRI) studies have attempted to overcome the limitations of temporal resolution to investigate dynamic states of brain activity. However, finding an fMRI method with sufficient temporal resolution to keep up with the progress of neuronal signals in the brain is challenging. This study aimed to detect between-hemisphere signal progression, occurring on a timescale of tens of milliseconds, in the ventral brain regions involved in face processing. To this end, we devised an inter-stimulus interval (ISI) stimulation scheme and used a 7T MRI system to obtain fMRI signals with a high signal-to-noise ratio. We conducted two experiments: one to measure signal suppression depending on the ISI and another to measure the relationship between the amount of suppression and the ISI. These two experiments enabled us to measure the signal transfer time from a brain region in the ventral visual stream to its counterpart in the opposite hemisphere through the corpus callosum. These findings demonstrate the feasibility of using fMRI to measure ultra-fast signals (tens of milliseconds) and could facilitate the elucidation of further aspects of dynamic brain function.

Brain tissue segmentation based on MP2RAGE multi-contrast images in 7 T MRI.

We proposed a method for segmentation of brain tissues-gray matter, white matter, and cerebrospinal fluid-using multi-contrast images, including a T1 map and a uniform T1-weighted image, from a magnetization-prepared 2 rapid acquisition gradient echoes (MP2RAGE) sequence at 7 Tesla. The proposed method was evaluated with respect to the processing time and the similarity of the segmented masks of brain tissues with those obtained using FSL, FreeSurfer, and SPM12. The processing time of the proposed method (28 ± 0 s) was significantly shorter than those of FSL and SPM12 (444 ± 4 s and 159 ± 2 s for FSL and SPM12, respectively). In the similarity assessment, the tissue mask of the brain obtained by the proposed method showed higher consistency with those obtained using FSL than with those obtained using SPM12. The proposed method misclassified the subcortical structures and large vessels since it is based on the intensities of multi-contrast images obtained using MP2RAGE, which uses a similar segmentation approach as FSL but is not based on a template image or a parcellated brain atlas, which are used for FreeSurfer and SPM12, respectively. However, the proposed method showed good segmentation in the cerebellum and white matter in the medial part of the brain in comparison with the other methods. Thus, because the proposed method using different contrast images of MP2RAGE sequence showed the shortest processing time and similar segmentation ability as the other methods, it may be useful for both neuroimaging research and clinical diagnosis.

Aberrant Neural Activation Underlying Idiom Comprehension in Korean Children with High Functioning Autism Spectrum Disorder.

PURPOSE: Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by social communication impairments and repetitive behaviors or restricted interests. Impaired pragmatic language comprehension is a universal feature in individuals with ASD. However, the underlying neural basis of pragmatic language is poorly understood. In the present study, we examined neural activation patterns associated with impaired pragmatic language comprehension in ASD, compared to typically developing children (TDC). MATERIALS AND METHODS: Functional magnetic resonance imaging (fMRI) was applied to 15 children with ASD and 18 TDC using the Korean pragmatic language task. RESULTS: Children with ASD were less accurate than TDC at comprehending idioms, particularly when they were required to interpret idioms with mismatched images (mismatched condition). Children with ASD also showed different patterns of neural activity than TDC in all three conditions (neutral, matched, and mismatched). Specifically, children with ASD showed decreased activation in the right inferior frontal gyrus (IFG) (Brodmann area 47) in the mismatched condition, compared with TDC (IFG; t(31)=3.17, p<0.001). CONCLUSION: These results suggest that children with ASD face difficulties in comprehending pragmatic expressions and apply different pragmatic language processes at the neural level.

Estimation of vocational aptitudes using functional brain networks.

The success of human life in modern society is highly dependent on occupation. Therefore, it is very important for people to identify and develop a career plan that best suits their aptitude. Traditional test batteries for vocational aptitudes are not oriented to measure developmental changes in job suitability because repeated measurements can introduce bias as the content of the test batteries is learned. In this study, we attempted to objectively assess vocational aptitudes by measuring functional brain networks and identified functional brain networks that intrinsically represented vocational aptitudes for 19 job divisions in a General Aptitude Test Battery. In addition, we derived classifiers based on these networks to predict the aptitudes of our test participants for each job division. Our results suggest that the measurement of brain function can indeed yield an objective evaluation of vocational aptitudes; this technique will enable a person to follow changes in one's job suitability with additional training or learning, paving a new way to advise people on career development.

A Set of Functional Brain Networks for the Comprehensive Evaluation of Human Characteristics.

Many human characteristics must be evaluated to comprehensively understand an individual, and measurements of the corresponding cognition/behavior are required. Brain imaging by functional MRI (fMRI) has been widely used to examine brain function related to human cognition/behavior. However, few aspects of cognition/behavior of individuals or experimental groups can be examined through task-based fMRI. Recently, resting state fMRI (rs-fMRI) signals have been shown to represent functional infrastructure in the brain that is highly involved in processing information related to cognition/behavior. Using rs-fMRI may allow diverse information about the brain through a single MRI scan to be obtained, as rs-fMRI does not require stimulus tasks. In this study, we attempted to identify a set of functional networks representing cognition/behavior that are related to a wide variety of human characteristics and to evaluate these characteristics using rs-fMRI data. If possible, these findings would support the potential of rs-fMRI to provide diverse information about the brain. We used resting-state fMRI and a set of 130 psychometric parameters that cover most human characteristics, including those related to intelligence and emotional quotients and social ability/skill. We identified 163 brain regions by VBM analysis using regression analysis with 130 psychometric parameters. Next, using a 163 × 163 correlation matrix, we identified functional networks related to 111 of the 130 psychometric parameters. Finally, we made an 8-class support vector machine classifiers corresponding to these 111 functional networks. Our results demonstrate that rs-fMRI signals contain intrinsic information about brain function related to cognition/behaviors and that this set of 111 networks/classifiers can be used to comprehensively evaluate human characteristics.

Steady-state and dynamic network modes for perceptual expectation.

Perceptual expectation can attenuate repetition suppression, the stimulus-induced neuronal response generated by repeated stimulation, suggesting that repetition suppression is a top-down modulatory phenomenon. However, it is still unclear which high-level brain areas are involved and how they interact with low-level brain areas. Further, the temporal range over which perceptual expectation can effectively attenuate repetition suppression effects remains unclear. To elucidate the details of this top-down modulatory process, we used two short and long inter-stimulus intervals for a perceptual expectation paradigm of paired stimulation. We found that top-down modulation enhanced the response to the unexpected stimulus when repetition suppression was weak and that the effect disappeared at 1,000 ms prior to stimulus exposure. The high-level areas involved in this process included the left inferior frontal gyrus (IFG_L) and left parietal lobule (IPL_L). We also found two systems providing modulatory input to the right fusiform face area (FFA_R): one from IFG_L and the other from IPL_L. Most importantly, we identified two states of networks through which perceptual expectation modulates sensory responses: one is a dynamic state and the other is a steady state. Our results provide the first functional magnetic resonance imaging (fMRI) evidence of temporally nested networks in brain processing.

Structural and functional plasticity specific to musical training with wind instruments.

Numerous neuroimaging studies have shown structural and functional changes resulting from musical training. Among these studies, changes in primary sensory areas are mostly related to motor functions. In this study, we looked for some similar functional and structural changes in other functional modalities, such as somatosensory function, by examining the effects of musical training with wind instruments. We found significant changes in two aspects of neuroplasticity, cortical thickness, and resting-state neuronal networks. A group of subjects with several years of continuous musical training and who are currently playing in university wind ensembles showed differences in cortical thickness in lip- and tongue-related brain areas vs. non-music playing subjects. Cortical thickness in lip-related brain areas was significantly thicker and that in tongue-related areas was significantly thinner in the music playing group compared with that in the non-music playing group. Association analysis of lip-related areas in the music playing group showed that the increase in cortical thickness was caused by musical training. In addition, seed-based correlation analysis showed differential activation in the precentral gyrus and supplementary motor areas (SMA) between the music and non-music playing groups. These results suggest that high-intensity training with specific musical instruments could induce structural changes in related anatomical areas and could also generate a new functional neuronal network in the brain.

Abnormal brain activity in social reward learning in children with autism spectrum disorder: an fMRI study.

PURPOSE: We aimed to determine whether Autism Spectrum Disorder (ASD) would show neural abnormality of the social reward system using functional MRI (fMRI). MATERIALS AND METHODS: 27 ASDs and 12 typically developing controls (TDCs) participated in this study. The social reward task was developed, and all participants performed the task during fMRI scanning. RESULTS: ASDs and TDCs with a social reward learning effect were selected on the basis of behavior data. We found significant differences in brain activation between the ASDs and TDCs showing a social reward learning effect. Compared with the TDCs, the ASDs showed reduced activity in the right dorsolateral prefrontal cortex, right orbitofrontal cortex, right parietal lobe, and occipital lobe; however, they showed increased activity in the right parahippocampal gyrus and superior temporal gyrus. CONCLUSION: These findings suggest that there might be neural abnormality of the social reward learning system of ASDs. Although this study has several potential limitations, it presents novel findings in the different neural mechanisms of social reward learning in children with ASD and a possible useful biomarker of high-functioning ASDs.

Abnormal activation of the social brain network in children with autism spectrum disorder: an FMRI study.

OBJECTIVE: The aim of this study is to investigate abnormal findings of social brain network in Korean children with autism spectrum disorder (ASD) compared with typically developing children (TDC). METHODS: Functional magnetic resonance imaging (fMRI) was performed to examine brain activations during the processing of emotional faces (happy, fearful, and neutral) in 17 children with ASD, 24 TDC. RESULTS: When emotional face stimuli were given to children with ASD, various areas of the social brain relevant to social cognition showed reduced activation. Specifically, ASD children exhibited less activation in the right amygdala (AMY), right superior temporal sulcus (STS) and right inferior frontal gyrus (IFG) than TDC group when fearful faces were shown. Activation of left insular cortex and right IFG in response to happy faces was less in the ASD group. Similar findings were also found in left superior insular gyrus and right insula in case of neutral stimulation. CONCLUSION: These findings suggest that children with ASD have different processing of social and emotional experience at the neural level. In other words, the deficit of social cognition in ASD could be explained by the deterioration of the capacity for visual analysis of emotional faces, the subsequent inner imitation through mirror neuron system (MNS), and the ability to transmit it to the limbic system and to process the transmitted emotion.