The hippocampal formation forms a circuit of cytoarchitectonically distinct subregions, and substantial evidence suggests each region makes unique computational contributions that support spatial and episodic memory. With aging, hippocampal subfields undergo unique neurobiological alterations, and primate in vivo work making use of both MR imaging and chronic neural recording devices has important links to changes seen in nonprimate animal models with aging (Thome et al., 2016; Yassa et al., 2011a; Yassa et al., 2010). While MRI offers a noninvasive way to study the hippocampal subfields, identifying hippocampal subregions without using post mortem histology is a challenge. When different research labs attempted to identify the hippocampal subregions using a single subject’s MRI, researchers showed significant disagreement in where to label different subregions (Yushkevich et al., 2015a). Alternatively, chronic neural recording devices offer an invasive solution to studying hippocampal subfields. However, it is currently not clear whether the mechanical trauma and foreign body response produced by neural recording devices disrupts neural circuits critical for behavior. Here, my colleagues and I address these issues with in vivo primate research. Chapter I provides a general introduction to the hippocampal circuits and changes observed in aging. Chapter II presents novel methods for construction of a histology-driven MRI atlas of nonhuman primate hippocampus that addresses accurate identification of hippocampal subfields in MR images. Chapter III presents empirical work that examines whether chronic neural recording devices targeted at the hippocampus affect recognition memory. Finally, Chapter IV provides a general discussion of both works in the context of the broader literature.

Human brain structure can be measured across the lifecourse (“in vivo”) with magnetic resonance imaging (MRI). MRI data are often used to create “atlases” and statistical models of brain structure across the lifecourse. These methods may define how brain structure changes through life and support diagnoses of increasingly common, yet still fatal, age-related neurodegenerative diseases. As diseases such as Alzheimer’s (AD) cast an ever growing shadow over our ageing population, it is vitally important to robustly define changes which are normal for age and those which are pathological. This work therefore assessed existing MR brain image data, atlases, and statistical models. These assessments led me to propose novel methods for accurately defining the distributions and boundaries of normal ageing and pathological brain structure. A systematic review found that there were fewer than 100 appropriately tested normal subjects aged ≥60 years openly available worldwide. These subjects did not have the range of MRI sequences required to effectively characterise the features of brain ageing. The majority of brain image atlases identified in this review were found to contain data from few or no subjects aged ≥60 years and were in a limited range of MRI sequences. All of these atlases were created with parametric (mean-based) statistics that require the assumptions of equal variance and Gaussian distributions. When these assumptions are not met, mean-based atlases and models may not well represent the distributions and boundaries of brain structure. I tested these assumptions and found that they were not met in whole brain, subregional, and voxel-based models of ~580 subjects from across the lifecourse (0- 90 years). I then implemented novel whole brain, subregional, and voxel-based statistics, e.g. percentile rank atlases and nonparametric effect size estimates. The equivalent parametric statistics led to errors in classification and inflated effects by up to 45% in normal ageing-AD comparisons. I conclude that more MR brain image data, age appropriate atlases, and nonparametric statistical models are needed to define the true limits of normal brain structure. Accurate definition of these limits will ultimately improve diagnoses, treatment, and outcome of neurodegenerative disease.

A novel framework for quantitative analysis of shape and function in magnetic resonance imaging (MRI) of the brain is proposed. First, an efficient method to compute invariant spherical harmonics (SPHARM) based feature representation for real valued 3D functions was developed. This method addressed previous limitations of obtaining unique feature representations using a radial transform. The scale, rotation and translation invariance of these features enables direct comparisons across subjects. This eliminates need for spatial normalization or manually placed landmarks required in most conventional methods [1-6], thereby simplifying the analysis procedure while avoiding potential errors due to misregistration. The proposed approach was tested on synthetic data to evaluate its improved sensitivity. Application on real clinical data showed that this method was able to detect clinically relevant shape changes in the thalami and brain ventricles of Parkinson's disease patients. This framework was then extended to generate functional features that characterize 3D spatial activation patterns within ROIs in functional magnetic resonance imaging (fMRI). To tackle the issue of intersubject structural variability while performing group studies in functional data, current state-of-the-art methods use spatial normalization techniques to warp the brain to a common atlas, a practice criticized for its accuracy and reliability, especially when pathological or aged brains are involved [7-11]. To circumvent these issues, a novel principal component subspace was developed to reduce the influence of anatomical variations on the functional features. Synthetic data tests demonstrate the improved sensitivity of this approach over the conventional normalization approach in the presence of intersubject variability. Furthermore, application to real fMRI data collected from Parkinson's disease patients revealed significant differences in patterns of activation in regions undetected by conventional means. This heightened sensitivity of the proposed features would be very beneficial in performing group analysis in functional data, since potential false negatives can significantly alter the medical inference. The proposed framework for reducing effects of intersubject anatomical variations is not limited to functional analysis and can be extended to any quantitative observation in ROIs such as diffusion anisotropy in diffusion tensor imaging (DTI), thus providing researchers with a robust alternative to the controversial normalization approach. / Applied Science, Faculty of / Electrical and Computer Engineering, Department of / Graduate

Current models of face perception suggest independent processing of identity and expression, though this distinction is still unclear. Using converging methods of psychophysics and functional magnetic resonance imaging (fMRI) in healthy and patient populations we assessed the relationship between these two perceptual processes. First, using perceptual aftereffects, we explored the neural representations underlying identity and expression. The expression aftereffect only partially transferred across different identities, suggesting adaptation within identity-invariant and identity-dependent expression representations. Contrarily, the identity aftereffect fully transferred across different expressions. This asymmetry cannot be explained through low-level adaptation. The identity-dependent component of the expression aftereffect relies on adaptation to a coherent expression, not low-level features, in the adapting face. Thus adaptation generating the expression aftereffect must occur within high-level representations of facial expression. Second, using fMRI adaptation, we examined identity and expression sensitivity in healthy controls. The fusiform face area and posterior superior temporal sulcus showed sensitivity for both identity and expression changes. Independent sensitivity for identity and expression changes was observed in the precuneus and middle superior temporal sulcus respectively. Finally, we explored identity and expression perception in a neuropsychological population. Selective identity impairments were associated with inferior occipitotemporal damage, not necessarily affecting the occipital or fusiform face areas. Impaired expression perception was associated with superior temporal sulcus damage, and also with deficits in the integration of identity and expression. In summary, psychophysics, neuroimaging and neuropsychological methods all provide converging evidence for the independent processing of identity and expression within the face network. However, these same methods also supply converging evidence for a partial dependence of these two perceptual processes: in the expression aftereffect, the functional sensitivities of the FFA and pSTS, and identity deficits observed in a patient with primarily impaired expression perception and a spared inferotemporal cortex. Thus, future models of face perception must incorporate representations or regions which independently process identity or expression as well as those which are involved in the perception of both identity and expression. / Medicine, Faculty of / Graduate

Complaints of reduced cognitive abilities are frequent following chemotherapy. Research in the breast cancer population has revealed some patients may experience treatment-related decline in cognitive domains such as executive function, information processing speed, memory and learning, attention and concentration, and working memory. The extent and mechanism of action of this phenomenon remain poorly understood. Neuroimaging research can characterize the neural underpinnings of chemotherapy-related cognitive impairment; however, with few longitudinal studies, more prospective studies are needed to elucidate this important topic. The aim of this thesis was to use magnetic resonance imaging and contemporary analysis techniques to better understand the influence chemotherapy exerts on both the brain and cognition. This was achieved in two studies that measured cognitive function and brain structure and function at three time points: pre-treatment, one month post-chemotherapy, and at one-year follow-up. In the first study, the association between regions of brain structural changes and cognitive function was examined. The second study took a narrower approach and investigated the functional profile of brain activity during a working memory task. Patients had more pronounced structural and functional disruptions shortly after treatment, relative to both pre-treatment and one-year post-chemotherapy intervals. Regions of structural compromise were largely associated with information processing speed. Functional disruptions occurred in a frontoparietal network. Overall, this thesis provides more evidence of the injurious role chemotherapy plays on cognition, particularly in the short term. This thesis also provides the first longitudinal neuroimaging study to illustrate a complete resolution of working memory related brain disruption one year post-treatment.

Trabalhos anteriores sobre a síntese de imagens de difusão por ressonância magnética se limitaram a estudos sobre estruturas microscópicas, menores que as dimensões típicas de um voxel (e.g., [BF08] [BF13] [LFS + 10] e [BA94]). Isto decorre em parte devido às metodologias utilizadas, que têm como ponto em comum serem simulações de tipo Monte Carlo, nas quais os elementos mínimos da simulação são as partículas de água. Portanto o custo computacional destas simulações é proporcional ao número de partículas a simular, e isto limita os volumes que podem ser simulados a tamanhos microscópicos. Propomos uma metodologia alternativa, que utiliza a imagem T 2 de uma amostra para sintetizar imagens de difusão por ressonância magnética. Os elementos mínimos desta nova metodologia são os pontos da imagem T 2 , e portanto seu custo computacional é proporcional à resolução da imagem T 2 utilizada, o que permite a síntese a partir de amostras de qualquer tamanho físico. Estas sínteses são realizadas por meio da integração numérica da equação do artigo seminal de Stejskal e Tanner [ST65] que relaciona a atenuação do sinal de ressonância magnética devida à difusão com os parâmetros da sequência de pulsos PGSE. Usamos os parâmetros típicos dessa sequência (b, gamma, tau\', g 0, g, delta e Delta), que podem ser configurados explicitamente em máquinas de ressonância magnética, para calcular valores do coeficiente de difusão aparente D em direções arbitrárias. Desenvolvemos software, disponibilizado [Bor] por licença GPL [Fou07], para realizar estas simulações, e para especificar uma máscara de direções, útil para modelar a difusão de uma amostra. Estas ferramentas permitem o estudo sistemático das variações dos parâmetros na síntese de imagens de difusão por ressonância magnética. Apresentamos um estudo de um fantoma de capilares imersos em água, exemplificando como utilizar as ferramentas para investigar a influência destes parâmetros na difusão da água da amostra. / Previous work on the synthesis of diffusion-weighted magnetic resonance imaging are limited to microscopic structures, smaller than the typical dimensions of a single voxel (e.g., [BF08] [BF13] [LFS + 10] and [BA94]). This is consequence, in part, of the methodologies used, that have in common the adoption of Monte Carlo simulation strategies, in which the minimal elements of simulation are the water particles. Therefore the computational cost of these simulations is proportional to the number of particles to simulate, and this limits the volume to be simulated to microscopic sizes. We propose a novel methodology, that uses the T 2 image from a sample to synthesize diffusion-weighted magnetic resonance images. The mininal elements of this novel methodology are the points of the T 2 image, and therefore its computational cost is proportional to the resolution of the T 2 image to be used, which allows the synthesis from samples of any physical size. These syntheses are made through numerical integration of the equation from the seminal article by Stejskal and Tanner [ST65] that relates the attenuation of the magnetic resonance signal due to diffusion to the parameters of the PGSE pulse sequence. We use the typical parameters of this sequence (b, gamma, tau\', g 0, g, delta and Delta), that can be explicitly configured in magnetic resonance machines, to calculate apparent diffusion coefficients D in arbitrary directions. We developed software, available [Bor] through GPL license [Fou07], to run these simulations, and to specify a mask of directions useful to model diffusion. These tools allow the systematic study of parameter variation in the synthesis of diffusion-weighted magnetic resonance images. We present a case study of a phantom made of capillary tubes immersed in water, to exemplify the use of these tools and how to investigate the influence of parameter variation on diffusion in the sample.

BACKGROUND: Sleep deprivation is one of the first symptoms to manifest in schizophrenia patients. An important proponent for both sleep and cognition, sleep spindles have been investigated to understand the connection between sleep and schizophrenia. This thesis aims to conduct a meta-analysis on this topic to conglomerate previous research and come to a definitive conclusion on how sleep and schizophrenia interact. Multiple studies have reported associations between sleep, schizophrenia, and the thalamus. Novel methods have allowed researchers to segment the thalamus into 25 different nuclei. Therefore, this thesis will also attempt to validate these findings and use new segmentation software to investigate which specific nuclei affect schizophrenia. This thesis also extends this thalamic investigation to explore white matter tracts related to the thalamus. Using both arms of this study, we aim to further understand the complex relationship between brain structure, sleep, and schizophrenia. METHODS: The meta-analysis portion of this thesis pooled fifteen studies for a total of 321 patients and 323 healthy controls. The patient population was made up of first-episode psychosis (FEP), family high-risk (FHR), and schizophrenia (SZ) populations. R Studio was utilized to run a meta-analysis on sleep spindle density (SSD) values pulled from each study. This dataset was also used for meta-regressions and funnel plots. The imaging aspect of this thesis pulled subjects from two separate Boston studies for a total of 54 early course patients (EC) and 27 healthy controls (HC). A brain editing software, FreeSurfer, was used to quality control and segment the thalamus. This segmentation provided volumes for each nuclei. A free-water imaging pipeline was used to process diffusion weighted images so that free-water (FW) and corrected fractional anisotropy (FAt) could be collected. These values were run through Tract-Based Spatial Statistics (TBSS) to standardize scans and identify white matter regions of interest. RESULTS: This thesis reports an effect size of -1.24 between HC and the collective subject groups with a confidence interval of -1.63 to -.84. Via meta-regression, we report that illness duration, publication year, and spindle frequency gap (SFG) all are associated with sleep spindles. The thalamic volumetric analysis showed that four nuclei differed between EC and HC and two nuclei differed between EC and FH. The imaging aspect of this thesis discovered no significant FAt findings comparing volume to population, though heat maps revealed a trend where FHR and EC had lower FAt than HC. For FW analysis, we found a similar trend where FHR and EC had higher FW than HC. CONCLUSION: Using both analyses, this thesis connected sleep spindles, schizophrenia, and brain structure. We illuminated consistent reports of schizophrenia populations having lower sleep spindle density compared to healthy controls. This thesis reports a difference in thalamic nuclei volumes between both HC and FHR versus EC as well as FAt and FW differences between both FHR and EC and HC. / 2021-06-07T00:00:00Z

Balanced steady-state free precession (bSSFP) is a fast, SNR-efficient magnetic resonance (MR) imaging sequence suffering from dark banding artifacts due to its off-resonance dependence. These banding artifacts are difficult to mitigate at high field strengths and in the presence of metallic implants. Recent developments in parametric modelling of bSSFP have led to advances in banding removal and parameter estimation using multiple phase-cycled bSSFP. With increasing number of coils in receivers, more storage and processing is required. Coil combination is used to reduce dimensionality of these datasets which otherwise might be prohibitively large or computationally intractable for clinical applications. However, our recent work demonstrates that some combination methods are problematic in conjunction with elliptical phase-cycled bSSFP.This thesis will present a method for phase estimation of coil-combined multiple phase-cycled bSSFP to reduce storage and computational requirements for elliptical models. This method is general and works across many coil combination techniques popular in MR reconstruction including the geometric coil combine and adaptive coil combine algorithms. A viable phase estimate for the sum-of-squares is also demonstrated for computationally efficient dimension reduction. Simulations, phantom experiments, and in vivo MR imaging is performed to validate the proposed phase estimates.Compressed sensing (CS) is an increasingly important acquisition and reconstruction framework. CS MR allows for reconstruction of datasets sampled well-under the Nyquist rate and its application is natural in MR where images are often sparse under common linear transforms. An extension of this framework is the ordering problem for CS, first introduced in 2008. Although the assumption is made in CS that images are sparse in some specified transform domain, it might not be maximally sparse. For example, a signal ordered such that it is monotonic is maximally sparse in the finite differences domain. Knowledge of the correct ordering of an image's pixels can lead to much more sparse and powerful regularizers for the CS inverse problem. However, this problem has met with little interest due to the strong dependence on initial image estimates.This thesis will also present an algorithm for estimating the optimal order of a signal such that it is maximally sparse under an arbitrary linear transformation without relying on any prior image estimate. The algorithm is combinatoric in nature and feasible for small signals of interest such as T1 mapping time curves. Proof of concept simulations are performed that validate performance of the algorithm. Computationally feasible modifications for in vivo cardiac T1 mapping are also demonstrated.

En la actualidad, la medicina y en especial el área de imagen médica, ha sido uno de los campos de la ciencia que más se ha beneficiado de las tecnologías de la información. En esta tesis nos centramos en las imágenes de resonancia magnética cerebral y los métodos de segmentación automática del volumen cerebral. En el primer capítulo describimos los antecedentes de este trabajo mediante una breve introducción de los principios físicos de la resonancia magnética y una revisión del estado del arte en relación con las técnicas de segmentación y su evolución a lo largo de las últimas décadas. En los capítulos tres, cuatro y cinco presentamos tres métodos de segmentación automática de diferentes partes del cerebro que mejoran el estado del arte en términos de calidad del resultado y velocidad. Finalmente, en el capítulo seis se comentan las conclusiones generales y se proponen líneas de desarrollo futuras. / Nowadays, medicine and specially the medical image area, is one of the science fields that has benefited more from the information technologies. In this thesis we focus on cerebral magnetic resonance imaging and automatic segmentation methods of the brain volume. In the first chapter we describe a background for this work by a brief introduction of the physic fundaments of magnetic resonance and a revision of the state-of-the-art regarding to segmentation techniques and its evolution along the last decades. In the chapters three, four and five we present three methods for automatically segment different parts of the brain that leverage the state-of-the-art in terms of result quality and computation time. Finally, in chapter six we comment the general conclusions and propose lines for future works. / En l'actualitat, la medicina i en especial l'àrea d'imatge mèdica, ha estat un dels camps de la ciència que més s'ha beneficiat de les tecnologies de la informació. En aquesta tesi ens centrem en les imatges de ressonància magnètica cerebral i el desenvolupament de mètodes de segmentaciò automàtica del volum cerebral. En el primer capítol descrvim els antecedents d'aquest treball mitjançant una breu introducció dels principis físics de la ressonància magnètica i una revisió de l'estat de l'art en relació amb les tècniques de segmentació i la seva evolució al llarg de les últimes dècades. En els capítols tres, quatre, i cinc presentem tres mètodes de segmentació automàtica per a diferents parts del cervell que milloren l'estat de l'art en quant a qualitat dels seus resultats i velocitat. Finalment, en el capitol sis es comenten les conclusions generals i es proposen línies de desenvolupament ions generals i es proposen línies de desenvolupament futures. / Romero Gómez, JE. (2018). Nuevos métodos para el análisis automático del volumen de estructuras cerebrales a partir de imágenes de resonancia magnética nuclear [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/106346 / TESIS

京都大学 / 0048 / 新制・課程博士 / 博士(医学) / 甲第22638号 / 医博第4621号 / 新制||医||1044(附属図書館) / 京都大学大学院医学研究科医学専攻 / (主査)教授 武藤 学, 教授 小川 誠司, 教授 小川 修 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM