INTRODUÇÃO: Imagens por tensores de difusão (diffusion tensor imaging - DTI) do cérebro são parte fundamental de exames de ressonância magnética (RM) na rotina clínica. Entretanto, não há informação científica suficiente para inferir a base histológica de parâmetros de imagens de DTI. Um dos achados frequentes nos exames de RM são as áreas de alteração de sinal na substância branca (AASSB). Neste estudo, analisamos imagens de RM post-mortem, em particular as AASSB, e seus correlatos histológicos e imuno-histoquímicos. OBJETIVOS: Descrever os parâmetros quantitativos de imagens de DTI imagens obtidas in cranio post-mortem, comparar com parâmetros de relaxometria e transferência de magnetização e avaliar seus correlatos histológicos nas AASSB e AHAN, utilizando técnicas de correspondência espacial ponto-a-ponto. MÉTODOS: Analisamos áreas distribuídas na substância branca de encéfalos de quatro sujeitos submetidos ao exame de RM post-mortem, aproximadamente 12,87 horas (±2,59). A RM foi realizada em sistema com campo de 3,0 T, sendo utilizadas imagens FLAIR; T2 multi-eco, DTI, Densidade de Prótons para cálculo da taxa de transferência de magnetização (magnetization transfer ratio - MTR). Foram delimitadas regiões de interesse (ROI) em 20 áreas contendo AASSB e 20 áreas homólogas aparentemente normais (AHAN). Para cada área, foram obtidos valores de anisotropia fracional (FA), coeficiente de difusão aparente (ADC), MTR e tempo T2. Os encéfalos foram seccionados em fatias de 0,4mm e fotografados em alta resolução. Estas imagens foram co-registradas tridimensionalmente com as imagens de RM, utilizando técnica semi-automatizada para correlação ponto-a-ponto. Os fragmentos analisados foram corados por hematoxilina-eosina (HE), Kluverx Barrera (KB), galocianina (GALL), anti-neurofilamento (anti-NF), antiproteína básica de mielina (anti-MBP) e anti-proteína ácida fibrilar glial (anti- GFAP). RESULTADOS: Os valores de FA foram 0,40 ±0,12 nas AASSB, e 0,52±0,15 nas AHAN. Os valores de ADC foram nas AASSB de 0,129±0,04 x 10-3mm2/s, e 0,108±0,03 x10-3mm2/s nas AHAN. Os valores de MTR foram 51,59±0,88 % nas AASSB, e 52,88±4,09 % nas AHAN. Os valores de RT2 foram 91,79±1,95 % nas AASSB, e 69,94±4,54 % nas AHAN. Os valores de HE foram 162,9±20,0 nas AASSB, e 163,3±20,4 nas AHAN. Os valores de GALL foram de 181,8±4,36 nas AASSB, e 167,11±20,65 nas AHAN. Os valores de KB foram de 168,9±19,08 nas AASSB, e 182.82±20,65 nas AHAN. Os valores de NF médios mensurados foram de 131,6±12,23 nas AASSB, e 128,03±4,74 nas AHAN. Os valores de MBP médios mensurados foram de 153,9 ±13,67 nas AASSB, e 153,18±12,69 nas AHAN. Os valores de GFAP médios mensurados foram de 94.61±7,85 nas AASSB, e 81,23±6,29 nas AHAN. Os testes não paramétricos de correlação (Spearman) entre variáveis de imagens de RM e histologia apresentaram correlação apenas entre FA e GFAP (r=-0,59; p < 0,005). DISCUSSÃO E CONCLUSÕES: De acordo com os resultados obtidos, podemos inferir que a presença de gliose é um processo patológico que pode ser detectado utilizando os parâmetros de FA nas imagens de DTI. Os resultados apresentados demonstram que imagens de RM adquiridas post-mortem podem servir para compreender melhor as alterações estruturais do tecido cerebral / BACKGROUND: Diffusion tensor imaging (DTI) is part of magnetic resonance imaging (MRI) acquisition methods available in clinical exams. However, there are not enough scientific basis to infer the histological substract of DTI parameters. White matter hyperintensities (WMH) are frequent findings in clinical MRI routine. In the last years, scientific evidences show that these injuries are not just age-related benign changes, but they could be associated to pathological processes. In this study we analyzed post-mortem MRI DTI data, specifically investigating WMH, their histological and immunohistochemical correlates. OBJECTIVES : At present study, we aimed to analyze quantifiable DTI parameters - Fractional Anisotropy (FA) and Apparent Diffusion Coefficient (ADC) - relaxometry (RT2) and magnetization transfer ratio (MTR) in WMH compared and normal appearing white matter (NAWM) from images obtained postmortem and in situ, and assess their histological substracts at WMH and NAWM using a point-topoint correlation platform. METHODS : We analyzed 20 regions of interest (ROI) encompassing WMH and NAWM selected from four subjects using in situ post-mortem MRI data acquired in 3.0T MR system. The subjects were scanned with a post-mortem interval of approximately 12h54m (±2h36m). The MRI analysis included 3D T1, FLAIR; multi-echo T2, DTI, and proton density to calculate the rate of magnetization transfer (magnetization transfer ratio - MTR). In each ROI we obtained measures of FA, ADC, RT2 and MTR. Brain specimens were posteriorly fixed in celloidin, sectioned into 0.4mm slices and photographed in high resolution. These images were co-registered with three-dimensional MR images using semi-automated technique for correlation using a point-to-point method established in our group. WMH and NAWM ROIs from MRI were submitted to quantitative histological analysis based on optical density color-deconvolution technique. Histological section were stained in hematoxylin-eosin (HE) , Kluver-Barrera (KB) , galloccyanin (GALL), anti-neurofilament (anti-NF) , anti-basic myelin protein (anti-MBP) and anti-glial fibrillary acidic protein (anti-GFAP). RESULTS: The mean FA values were 0.40±0.12 in WMH, and 0.52±0.15 in NAWM. ADC in WMH was 0.1290±0.04, x10-3mm2/s and 0.1081± 0.03 x10-3mm2/s in NAWM. MTR was 51.59± 0.88% in WMH, and 52.88±4.09% in NAWM. RT2 was 91.79±1.95% in WMH, and 69.94±4.54% in NAWM. FA, ADC and RT2 values were different between WMH and NAWM in a non-parametric analysis. On the quantitative histological analysis, the HE was 162.9±20.0 in WMH , and 163.3±20.4% in NAWM. GALL values at WMH was 181.8±4.36 and 167.11±20.65 at NAWM. KB was 168.9±19.08 in WMH, and 182.82±20.65 at NAWM. NF mean were 131.6 ±12.23 in WMH, and 128.03 ±14.74 at NAWM. MBP was 153.9±13.67 at WMH, and 153.18±12.69 at NAWM. Only GFAP values were different between WMH and NAWM. Mean GFAP was 94.61±7.85 in WMH, and 81.23±6.29 at NAWM. Non-parametric test (Spearman) between MR images and histology, showed correlation only for FA vs GFAP (r=-0.594, p < 0.005). DISCUSSION AND CONCLUSIONS: DTI FA and ADC parameters (and RT2 values) showed different patterns in WMH compared to NAWM, indicating that MR images acquired post-mortem may serve to understand the structural changes of the brain tissue. The presence of gliosis is a pathological process seems to be related to FA value from DTI

INTRODUÇÃO: Imagens por tensores de difusão (diffusion tensor imaging - DTI) do cérebro são parte fundamental de exames de ressonância magnética (RM) na rotina clínica. Entretanto, não há informação científica suficiente para inferir a base histológica de parâmetros de imagens de DTI. Um dos achados frequentes nos exames de RM são as áreas de alteração de sinal na substância branca (AASSB). Neste estudo, analisamos imagens de RM post-mortem, em particular as AASSB, e seus correlatos histológicos e imuno-histoquímicos. OBJETIVOS: Descrever os parâmetros quantitativos de imagens de DTI imagens obtidas in cranio post-mortem, comparar com parâmetros de relaxometria e transferência de magnetização e avaliar seus correlatos histológicos nas AASSB e AHAN, utilizando técnicas de correspondência espacial ponto-a-ponto. MÉTODOS: Analisamos áreas distribuídas na substância branca de encéfalos de quatro sujeitos submetidos ao exame de RM post-mortem, aproximadamente 12,87 horas (±2,59). A RM foi realizada em sistema com campo de 3,0 T, sendo utilizadas imagens FLAIR; T2 multi-eco, DTI, Densidade de Prótons para cálculo da taxa de transferência de magnetização (magnetization transfer ratio - MTR). Foram delimitadas regiões de interesse (ROI) em 20 áreas contendo AASSB e 20 áreas homólogas aparentemente normais (AHAN). Para cada área, foram obtidos valores de anisotropia fracional (FA), coeficiente de difusão aparente (ADC), MTR e tempo T2. Os encéfalos foram seccionados em fatias de 0,4mm e fotografados em alta resolução. Estas imagens foram co-registradas tridimensionalmente com as imagens de RM, utilizando técnica semi-automatizada para correlação ponto-a-ponto. Os fragmentos analisados foram corados por hematoxilina-eosina (HE), Kluverx Barrera (KB), galocianina (GALL), anti-neurofilamento (anti-NF), antiproteína básica de mielina (anti-MBP) e anti-proteína ácida fibrilar glial (anti- GFAP). RESULTADOS: Os valores de FA foram 0,40 ±0,12 nas AASSB, e 0,52±0,15 nas AHAN. Os valores de ADC foram nas AASSB de 0,129±0,04 x 10-3mm2/s, e 0,108±0,03 x10-3mm2/s nas AHAN. Os valores de MTR foram 51,59±0,88 % nas AASSB, e 52,88±4,09 % nas AHAN. Os valores de RT2 foram 91,79±1,95 % nas AASSB, e 69,94±4,54 % nas AHAN. Os valores de HE foram 162,9±20,0 nas AASSB, e 163,3±20,4 nas AHAN. Os valores de GALL foram de 181,8±4,36 nas AASSB, e 167,11±20,65 nas AHAN. Os valores de KB foram de 168,9±19,08 nas AASSB, e 182.82±20,65 nas AHAN. Os valores de NF médios mensurados foram de 131,6±12,23 nas AASSB, e 128,03±4,74 nas AHAN. Os valores de MBP médios mensurados foram de 153,9 ±13,67 nas AASSB, e 153,18±12,69 nas AHAN. Os valores de GFAP médios mensurados foram de 94.61±7,85 nas AASSB, e 81,23±6,29 nas AHAN. Os testes não paramétricos de correlação (Spearman) entre variáveis de imagens de RM e histologia apresentaram correlação apenas entre FA e GFAP (r=-0,59; p < 0,005). DISCUSSÃO E CONCLUSÕES: De acordo com os resultados obtidos, podemos inferir que a presença de gliose é um processo patológico que pode ser detectado utilizando os parâmetros de FA nas imagens de DTI. Os resultados apresentados demonstram que imagens de RM adquiridas post-mortem podem servir para compreender melhor as alterações estruturais do tecido cerebral / BACKGROUND: Diffusion tensor imaging (DTI) is part of magnetic resonance imaging (MRI) acquisition methods available in clinical exams. However, there are not enough scientific basis to infer the histological substract of DTI parameters. White matter hyperintensities (WMH) are frequent findings in clinical MRI routine. In the last years, scientific evidences show that these injuries are not just age-related benign changes, but they could be associated to pathological processes. In this study we analyzed post-mortem MRI DTI data, specifically investigating WMH, their histological and immunohistochemical correlates. OBJECTIVES : At present study, we aimed to analyze quantifiable DTI parameters - Fractional Anisotropy (FA) and Apparent Diffusion Coefficient (ADC) - relaxometry (RT2) and magnetization transfer ratio (MTR) in WMH compared and normal appearing white matter (NAWM) from images obtained postmortem and in situ, and assess their histological substracts at WMH and NAWM using a point-topoint correlation platform. METHODS : We analyzed 20 regions of interest (ROI) encompassing WMH and NAWM selected from four subjects using in situ post-mortem MRI data acquired in 3.0T MR system. The subjects were scanned with a post-mortem interval of approximately 12h54m (±2h36m). The MRI analysis included 3D T1, FLAIR; multi-echo T2, DTI, and proton density to calculate the rate of magnetization transfer (magnetization transfer ratio - MTR). In each ROI we obtained measures of FA, ADC, RT2 and MTR. Brain specimens were posteriorly fixed in celloidin, sectioned into 0.4mm slices and photographed in high resolution. These images were co-registered with three-dimensional MR images using semi-automated technique for correlation using a point-to-point method established in our group. WMH and NAWM ROIs from MRI were submitted to quantitative histological analysis based on optical density color-deconvolution technique. Histological section were stained in hematoxylin-eosin (HE) , Kluver-Barrera (KB) , galloccyanin (GALL), anti-neurofilament (anti-NF) , anti-basic myelin protein (anti-MBP) and anti-glial fibrillary acidic protein (anti-GFAP). RESULTS: The mean FA values were 0.40±0.12 in WMH, and 0.52±0.15 in NAWM. ADC in WMH was 0.1290±0.04, x10-3mm2/s and 0.1081± 0.03 x10-3mm2/s in NAWM. MTR was 51.59± 0.88% in WMH, and 52.88±4.09% in NAWM. RT2 was 91.79±1.95% in WMH, and 69.94±4.54% in NAWM. FA, ADC and RT2 values were different between WMH and NAWM in a non-parametric analysis. On the quantitative histological analysis, the HE was 162.9±20.0 in WMH , and 163.3±20.4% in NAWM. GALL values at WMH was 181.8±4.36 and 167.11±20.65 at NAWM. KB was 168.9±19.08 in WMH, and 182.82±20.65 at NAWM. NF mean were 131.6 ±12.23 in WMH, and 128.03 ±14.74 at NAWM. MBP was 153.9±13.67 at WMH, and 153.18±12.69 at NAWM. Only GFAP values were different between WMH and NAWM. Mean GFAP was 94.61±7.85 in WMH, and 81.23±6.29 at NAWM. Non-parametric test (Spearman) between MR images and histology, showed correlation only for FA vs GFAP (r=-0.594, p < 0.005). DISCUSSION AND CONCLUSIONS: DTI FA and ADC parameters (and RT2 values) showed different patterns in WMH compared to NAWM, indicating that MR images acquired post-mortem may serve to understand the structural changes of the brain tissue. The presence of gliosis is a pathological process seems to be related to FA value from DTI

Le traitement d'images numériques a suivi l'évolution de l'électronique et de l'informatique. Il est maintenant courant de manipuler des images à valeur non pas dans {0,1}, mais dans des variétés ou des distributions de probabilités. C'est le cas par exemple des images couleurs où de l'imagerie du tenseur de diffusion (DTI). Chaque type d'image possède ses propres structures algébriques, topologiques et géométriques. Ainsi, les techniques existantes de traitement d'image doivent être adaptés lorsqu'elles sont appliquées à de nouvelles modalités d'imagerie. Lorsque l'on manipule de nouveaux types d'espaces de valeurs, les précédents opérateurs peuvent rarement être utilisés tel quel. Même si les notions sous-jacentes ont encore un sens, un travail doit être mené afin de les exprimer dans le nouveau contexte. Cette thèse est composée de deux parties indépendantes. La première, « Morphologie mathématiques pour les images non standards », concerne l'extension de la morphologie mathématique à des cas particuliers où l'espace des valeurs de l'image ne possède pas de structure d'ordre canonique. Le chapitre 2 formalise et démontre le problème de l'irrégularité des ordres totaux dans les espaces métriques. Le résultat principal de ce chapitre montre qu'étant donné un ordre total dans un espace vectoriel multidimensionnel, il existe toujours des images à valeur dans cet espace tel que les dilatations et les érosions morphologiques soient irrégulières et incohérentes. Le chapitre 3 est une tentative d'extension de la morphologie mathématique aux images à valeur dans un ensemble de labels non ordonnés.La deuxième partie de la thèse, « Estimation de densités de probabilités dans les espaces de Riemann » concerne l'adaptation des techniques classiques d'estimation de densités non paramétriques à certaines variétés Riemanniennes. Le chapitre 5 est un travail sur les histogrammes d'images couleurs dans le cadre de métriques perceptuelles. L'idée principale de ce chapitre consiste à calculer les histogrammes suivant une approximation euclidienne local de la métrique perceptuelle, et non une approximation globale comme dans les espaces perceptuels standards. Le chapitre 6 est une étude sur l'estimation de densité lorsque les données sont des lois Gaussiennes. Différentes techniques y sont analysées. Le résultat principal est l'expression de noyaux pour la métrique de Wasserstein. / Digital image processing has followed the evolution of electronic and computer science. It is now current to deal with images valued not in {0,1} or in gray-scale, but in manifolds or probability distributions. This is for instance the case for color images or in diffusion tensor imaging (DTI). Each kind of images has its own algebraic, topological and geometric properties. Thus, existing image processing techniques have to be adapted when applied to new imaging modalities. When dealing with new kind of value spaces, former operators can rarely be used as they are. Even if the underlying notion has still a meaning, a work must be carried out in order to express it in the new context.The thesis is composed of two independent parts. The first one, "Mathematical morphology on non-standard images", concerns the extension of mathematical morphology to specific cases where the value space of the image does not have a canonical order structure. Chapter 2 formalizes and demonstrates the irregularity issue of total orders in metric spaces. The main results states that for any total order in a multidimensional vector space, there are images for which the morphological dilations and erosions are irregular and inconsistent. Chapter 3 is an attempt to generalize morphology to images valued in a set of unordered labels.The second part "Probability density estimation on Riemannian spaces" concerns the adaptation of standard density estimation techniques to specific Riemannian manifolds. Chapter 5 is a work on color image histograms under perceptual metrics. The main idea of this chapter consists in computing histograms using local Euclidean approximations of the perceptual metric, and not a global Euclidean approximation as in standard perceptual color spaces. Chapter 6 addresses the problem of non parametric density estimation when data lay in spaces of Gaussian laws. Different techniques are studied, an expression of kernels is provided for the Wasserstein metric.

Le but de cette thèse était d'étudier l'organisation anatomo-fonctionnelle du système sensorimoteur humain et la façon dont les mouvements volontaires sont produits et contrôlés. Avec le développement de l’imagerie cérébrale, des méthodes de corrélation anatomo-clinique et de stimulation électrique directe cérébrale, de nombreuses avancées scientifiques ont pu être réalisée. Ces trois approches complémentaires ont été utilisé dans cette thèse afin d’améliorer la compréhension de l’organisation sensorimotrice cérébrale. Dans la première étude (soumise à publication), nous avons montré que la chirurgie cérébrale éveillée utilisant la stimulation électrique directe est une procédure sûre et efficace chez les enfants afin de réduire le déficit neurologique postopératoire. L'approche améliore la précision de la détection des zones éloquentes, avec une bonne tolérance neuropsychologique et psychologique. Une évaluation psychologique et neuropsychologique est essentielle. Dans une deuxième série de deux études, nous avons montré que la partie dorso-postérieure dorsal du cortex pariétal (DPPr) est une structure clé dans l'organisation complexe du mouvement manuel fin chez l'homme à travers la mise en oeuvre d'une boucle sensori-parieto-motrice.La première étude (publiée, Current Biology 2018) montre que la stimulation électrique directe d’une region corticale focale dans la partie dorso-postérieure du cortex pariétal entraine l’inhibition de la production du mouvement manuel, c’est-à-dire bloque l'initiation et la réalisation de ce dernier, sans produire de contraction musculaire ni de sensation consciente de mouvement. Dans la seconde étude (en cours de soumission), nous avions pour objectif d'identifier précisément les bases anatomiques du circuit parietal inhibiteur précédemment décrit. Grâce à la tractographie de diffusion (DTI), nous avons réussi à isoler des projections ipsilatérales spécifiques reliant les sites d’inhibition du DPPr, retrouvés dans la première étude, avec la zones dévolues au contrôle distal fin dans les cortex primaires moteur (M1) et sensoriel (S1). Ces données montrent que la boucle pariétale inhibitrice est directe depuis S1 vers DPPr vers M1 (même s'il n'est pas possible d'exclure l'existence d'échanges bidirectionnels entre ces aires). Dans la dernière étude (en cours de soumission), nous nous sommes intéressé à une structure motrice fondamentale, qui supporte 50 % des invasions tumorales chez l'enfant : le cervelet. Il s'agissait de déterminer si les lésions précoces étaient oui ou non prédictives d'une récupération déficitaire à long terme après prise en compte des covariables les plus critiques. Nous avons mesuré la récupération fonctionnelle à long terme chez 3 groupes survivants de lésion de la fosse postérieure. Les 3 groupes étaient comparables en ce qui concerne leurs caractéristiques tumorales mais opérés à différents âges : jeune (≤ 7 ans), moyen (> 7 ans et ≤ 13 ans) et tardif (> 13 ans). La qualité de vie (échelles cliniques : Health-related Quality of Life -hrQol- et Performance Status -PS-), les performances motrices (ataxie -ICARS- et motricité fine -Pegboard-) et cognitif (quotient intellectuel -FSIQ-) furent mesurés. L'âge précoce lors de la chirurgie, une lésion des noyaux profonds cérébelleux et la nécessité d'une radiothérapie postopératoire révélèrent une influence significativement négative et indépendante sur la récupération à long terme des participants. Ces résultats confirment l'existence d'une période critique de développement au cours de laquelle la "machine à apprendre" cérébelleuse revêt une importance cruciale / The aim of the thesis was to investigate the mapping of the anatomofunctional organization of the human sensorimotor system and how volutional movements of human are produced and controlled. Neuroimaging and especially DTI, fine anatomo-functional observation in patient and direct electrical stimulation were considered. This multi-modal approach permitted to improve our understanding of sensorimotor organization in humans. In the first study, we showed that awake brain surgery with the use of direct electrical stimulation is a safe and efficient procedure in children in order to decrease post-operative neurological deficit. It improves the accuracy of detecting eloquent area, with a good tolerance from a neuropsychological and psychological aspect. Age-adapted neuropsychologic preparation may enable offering ABS even to younger children on an individual basis. In a second series of two studies, we showed that the dorso-posterior part of the parietal cortex is a key structure in the complex organization of movement in human with a S1-DPPr-M1 loop. In the first study, direct electrical stimulation of focal cortical site in the dorso-posterior part of the parietal cortex triggered inhibition of movement production and blocked ongoing movement without producing muscle contraction or conscious movement sensation. In the second study, we aimed to find a direct projection from the PRR, defined in the first study (Desmurget et al., 2018), to the primary motor cortex and the primary somatosensory cortex. Thanks to the DTI state-of-the-art tractography, we succeeded in finding such major ipsilateral streamlines projecting in the well-known hand knob region giving new insights of the white matter structures involved in the inhibition of volitional hand movements. These observations confirm clinical per-operative data showing that stimulating the counterpart of PRR in humans can disrupt hand movements ipsilaterally, irrespective of the hemisphere. Moreover, our results shed light on the implication of the PRR for the volitional hand sensorimotor operating behavior. In the last study, we investigate the impact of early cerebellar damage on long-term functional recovery in 3 groups of posterior fossa survivors, comparable with respect to their tumoural characteristics but operated at different ages: young (≤ 7 years), middle (> 7 years and ≤ 13 years) and old (> 13 years). Daily (Health-related Quality of Life -hrQol-, Performance Status -PS-), motor (International Cooperative Ataxia Rating Scale -ICARS-, Pegboard Purdue Test -PegBoard-) and cognitive (Full Scale Intelligence Quotient -FSIQ-) functioning were measured. Early age at surgery, lesion of deep cerebellar nuclei and post-operative radiotherapy had a significant, independent negative influence on long term recovery. These results support the existence of an early critical period of development during which the cerebellar "learning machine" is of critical importance