京都大学 / 新制・課程博士 / 博士(工学) / 甲第25246号 / 工博第5205号 / 新制||工||1993(附属図書館) / 京都大学大学院工学研究科高分子化学専攻 / (主査)教授 田中 一生, 教授 大北 英生, 教授 大内 誠 / 学位規則第4条第1項該当 / Doctor of Agricultural Science / Kyoto University / DGAM

Described is the development of a multispectral image labeling system with emphasis on Unmanned Ground Vehicles(UGVs). UGVs operating in unstructured environments face significant problems detecting viable paths when LIDAR is the sole source for perception. Promising advances in computer vision and machine learning has shown that multispectral imagery can be effective at detecting materials in unstructured environments [1][2][3][4][5][6]. This thesis seeks to extend previous work[6][7] by performing pixel level classification with multispectral features and texture. First the images are spatially registered to create a multispectral image cube. Visual, near infrared, shortwave infrared, and visible/near infrared polarimetric data are considered. The aligned images are then used to extract features which are fed to machine learning algorithms. The class list includes common materials present in rural and urban scenes such as vehicles, standing water, various forms of vegetation, and concrete. Experiments are conducted to explore the data requirement for a desired performance and the selection of a hyper-parameter for the textural features. A complete system is demonstrated, progressing from the data collection and labeling to the analysis of the classifier performance. / Master of Science

The objective of this study was to evaluate the relationship between tissue oxygen saturation (StO2) and oxygen delivery index (DO2I). Oxygen delivery index is product of two factors arterial oxygen content (CaO2) and cardiac index (CI). In this study the relationship between DO2I and StO2 was evaluated by manipulating both of these factors independently. In phase one of the study, CaO2 was altered by manipulating the fractional inspired oxygen (FiO2) concentration. Anesthetized dogs were evaluated at both high (0.40 and 0.95) and low (0.15 and 0.10) FiO2 sequences. In phase two of the study, CI was altered by manipulating the volemic state. Anesthetized dogs were evaluated at hypovolemic, normovolemic and hypervolemic states. In each phase dogs were instrumented for thermodilution cardiac index (CI) and sartorius muscle StO2. Data collected included hemoglobin concentration, heart rate (HR), MAP, CI, StO2. Arterial oxygen content and DO2I were calculated at each time point. Data analysis included Pearson's correlation and mixed model ANOVA (p < 0.05). In both phases one (r = 0.97; p = 0.0013) and two (r = 0.97; p = 0.005) there was a strong correlation between StO2 and DO2I. Under the conditions of this study, there was a strong correlation between StO2 and DO2I, suggesting that StO2 may be used to estimate the adequacy of oxygen delivery in dogs. / Master of Science

The neurovisceral integration model (NVM) proposes that an organisms ability to flexibly adapt to their environment is related to biological flexibility within the central autonomic network. One important aspect of this flexibility is behavioral inhibition (Thayer and Friedman, 2002). During a behavioral inhibition task, the central autonomic network (CAN), which is comprised of a series of feedback loops, must be able to integrate information and react to these inputs flexibly to facilitate optimal performance. The functioning of the CAN is shown to be associated with respiratory sinus arrhythmia (RSA), as the vagus nerve is part of this feedback system. While the NVM has been examined through neural imaging and RSA, only a few studies have examined these measures simultaneously during the neuroimaging procedure. Furthermore, these studies were done at rest or used tasks that were not targeted at processes associated with the NVM, such as behavior inhibition and cognitive flexibility. For this reason, the present study assessed RSA and neural activation in the prefrontal cortex simultaneously while subjects completed a behavior inhibition task. Using a series of go/no-go tasks, RSA and functional near-infrared spectroscopy (fNIRS) were collected to investigate the relationship between prefrontal activation and vagal activity at rest and during behavioral inhibition. There are three primary aims of this study. First, examine prefrontal activation during various inhibition tasks through fNIRS. Second, evaluate the NVM during a cognitive task using simultaneous fNIRS and RSA analysis. Third, relate task performance, imaging, and RSA measures during behavioral inhibition to deficits in flexible everyday responding, as indicated by self-report measures of behavior. Doing so will elucidate the connection with prefrontal activation and RSA as proposed by the NVM model and determine whether neural and RSA metrics can be related to broader symptoms of inflexibility. / PHD / The neurovisceral integration model (NVM) proposes that the ability to adapt to the environment is related to biological flexibility within the brain. One important aspect of the ability to adapt to the environment is behavior inhibition (e.g., the ability to stop from engaging in a habitual response, Thayer & Friedman, 2002). During a behavior inhibition task, the brain must be able to integrate information and react to these inputs flexibly to facilitate optimal performance. The brain’s ability to do this is related to a measure of heart activity known as respiratory sinus arrhythmia (RSA). The present study assessed RSA and brain activity while subjects completed a behavior inhibition task. Neural activation was measured through functional near-infrared spectroscopy (fNIRS). fNIRS measures the amount of oxygenated blood in different areas of the brain. Greater concentrations of oxygenated blood indicated greater brain activity in an area. Through simultaneous fNIRS and RSA measurement the present study examined their relationship during various inhibition tasks. Doing so clarified the connection between brain activation and RSA as proposed by the NVM model.

Atrial fibrillation (AF), the most common type of cardiac arrythmia, has been a huge concern of public health. It affects more than 6 million people in the United States and over 33 million people worldwide. In the current standard of care, an electrogram and geometry map is generated by electroanatomic mapping (EAM) using a mapping catheter, to determine the origins of irregular heart rhythm. Followed by radiofrequency ablation (RFA) using ablation catheter, the targeted sites are ablated as lesions to change the electrical conduction pathway of abnormal electrogram, thus restoring the patients to normal sinus heart rhythm with minimally invasive procedure. However, a significant proportion of patients suffer from AF recurrences and requires repeated procedures, due to the lack of reliable methods to assess the cardiac structural substrates which are the potential maintaining mechanism of AF signals. In recent years, optical imaging modalities are developed to compensate this limitation, among which near-infrared spectroscopy (NIRS) is a catheter-based technique to enable direct, independent characterization of cardiac tissue pathology from spectrum morphology. In this thesis, we validate the capability of NIRS to generate map with repeatability and identify AF substrate to improve the efficacy of treatment. First, a near-infrared imaging spectroscopy was combined with an electromagnetic tracking modality, and the system was operated with high acquisition speed and real-time display to generate high-density map. Further, the robustness of NIRS optical parameters was assessed under blood mapping and various, large catheter-tissue contact angle, to simulate the dynamic circumstance of clinical procedures. A classification algorithm was introduced to predict lesion probability including both PBS and blood data, as well as to evaluate the mapping equivalence of blood and PBS. Next, the spatial resolution and the sampling density requirement of NIRS mapping method was characterized based on small gap, and the spectral properties of gap was assessed comparing to normal tissue and lesion by statistical analysis and machine learning. Lastly, we demonstrate the identification of human left atrial complex substrates using NIRS catheter with different source-detector-separations (SDSs), and reported the spectral features for the AF-related structures such as fibrosis and adipose. To summarize, the catheter-based NIRS technology is robust for in-vivo application and structural target localization, with the potential to enhance the recognition of underlying AF pathology and improve treatment efficacy.

Exercise-induced hypoalgesia (EIH) is the reduction in pain sensitivity following exercise. High-intensity or prolonged exercise is typically required to elicit an EIH response, but there is limited evidence suggesting that low-load resistance exercise with blood flow restriction (LL+BFR) may be able to elicit a robust EIH response. The purpose of these investigations was to assess the magnitude, duration, and proposed mechanisms of EIH following LL+BFR, LL with normobaric systemic hypoxemia, BFR only without exercise, high-load exercise, and a control intervention. The first study evaluated local and systemic pain pressure threshold and tolerance responses one-hour post-exercise. The results indicated that LL+BFR induced similar EIH responses to high-load resistance exercise 0- and 15-minutes post-exercise but only LL+BFR elicited an EIH response present 60-minutes post-exercise. The second study evaluated neuromuscular and perceptual responses, both proposed EIH mechanisms, during exercise. Similar neuromuscular responses were observed in all interventions. Participants reported higher ratings of perceived exertion during LL+BFR and high-load exercise, and higher levels of perceived pain during LL+BFR. These results suggested that despite high levels of motor unit recruitment, there were divergent EIH responses. However, increased pain during exercise may be a mediating factor of EIH after resistance exercise. The third study evaluated peripheral and central cardiovascular responses, which have also been hypothesized to mediate EIH. LL+BFR resulted in greater increases in systolic blood pressure during the first set of exercise, and diastolic blood pressure during all sets of exercise. LL+BFR and BFR only attenuated changes in heart rate variability (HRV). LL+BFR induced the largest increase in deoxyhemoglobin and total hemoglobin and lowered tissue saturation index. BFR only progressively increased oxyhemoglobin and total hemoglobin levels. The local and systemic cardiovascular responses suggested that prolonged EIH following LL+BFR could be related to increased central or peripheral cardiovascular stress.

To address the paucity of available molecularly targeted photoacoustic imaging probes (PIPs) and to generate meaningful data to support the ongoing effort to refine diagnostic photoacoustic imaging (PAI) applications, the work presented here focuses on the design, synthesis, and evaluation of novel PIPs. To this end, various light-absorbing small molecule dyes, targeting strategies, and disease-targeting molecules were evaluated. First, a near-infrared photoacoustic probe was used to image bone in vivo through active and bioorthogonal pre-targeting strategies by utilizing a coupling between a tetrazine-derived cyanine dye and a trans-cyclooctene-modified bisphosphonate. In vitro hydroxyapatite binding and in vivo bone imaging studies showed significant localization of the agent to the target using both active and pre-targeting strategies. The tetrazine-dye building block was then used to create a first-generation bacteria-targeting PIP, using a trans-cyclooctene-modified Zinc (II)-dipicolylamine (ZnDPA). The PIP demonstrated poor aqueous solubility and overlapping photoacoustic (PA) signal with deoxyhemoglobin. Therefore, a commercially available ZnDPA-derived fluorophore, PSVue794, was then repurposed for use as a PIP. PSVue794 demonstrated the ability to differentiate between bacterial infection, sterile inflammation, and healthy tissue in a mouse model, via PA imaging, which prompted its use in a series of proof-of-concept studies towards the generation of a model of implant infection. The feasibility of detecting the PIP in the presence of a PA signal-emitting metallic implant, which was deemed a significant hurdle due to the intensity of the PA signal of the metal, was verified. Although the work requires some follow-up evaluations to demonstrate the practical use of the model, ZnDPA-based PIPs have remained promising candidates for PAI of bacterial infection. Finally, a novel general-purpose dye was designed to possess properties ideal for in vivo PAI. To evaluate the modifications made, the general-purpose dye was conjugated with ZnDPA, and was tested alongside the non-targeted counterpart and PSVue794. Through the studies conducted, it was evident that the rationale that contributed to the design of the general-purpose dye did lead to highly soluble PIP with promising PA properties, however, the PIP did not demonstrate target-specificity, in vivo. Therefore, investigations using the non-targeted PIP with higher affinity targeting vectors for PA-compatible diseases, such as surgical-site/implant infections and prostate cancer, is warranted. / Thesis / Doctor of Philosophy (PhD) / The work conducted within this thesis aims to outline the process of developing photoacoustic diagnostic agents for the detection of various diseases, including bone disease and bacterial infection. To this end, various small molecule, near-infrared absorbing dyes, disease-targeting molecules, and assembly methods were selected to generate several diagnostic agents. To demonstrate their utility, the diagnostic agents were each evaluated in a series of studies designed to assess their ability to generate detectable photoacoustic signal, interact specifically with disease-markers, and localize the sites of disease in living systems. Significant attention was placed on comprehensively evaluating the diagnostic agents through the development of methodology and generating a standard procedure for photoacoustic data production and reporting, which was practiced throughout the work.

No / Knowledge and control of the polymorphic phases of chemical compounds are important aspects of drug development in the pharmaceutical industry. Salmeterol xinafoate, a long acting β-adrenergic receptor agonist, exists in two polymorphic Forms, I and II. Raman and near infrared spectra were obtained of these polymorphs at selected wavelengths in the range of 488–1064 nm; significant differences in the Raman and near-infrared spectra were apparent and key spectral marker bands have been identified for the vibrational spectro-scopic characterisation of the individual polymorphs which were also characterised with X ray diffractometry. The solid-state transition of salmeterol xinafoate polymorphs was studied using simultaneous in situ portable Raman spectroscopy and differential scanning calorimetry isothermally between transitions. This method assisted in the unambiguous characterisation of the two polymorphic forms by providing a simultaneous probe of both the thermal and vibrational data. The study demonstrates the value of a rapid in situ analysis of a drug polymorph which can be of potential value for at-line in-process control.

Yes / A transflectance near infra red (NIR) spectroscopy approach has been used to simultaneously measure drug and plasticiser content of polymer melts with varying opacity during hot melt extrusion. A high temperature reflectance NIR probe was mounted in the extruder die directly opposed to a highly reflective surface. Carbamazepine (CBZ) was used as a model drug, with polyvinyl pyrollidone-vinyl acetate co-polymer (PVP-VA) as a matrix and polyethylene glycol (PEG) as a plasticiser. The opacity of the molten extrudate varied from transparent at low CBZ loading to opaque at high CBZ loading. Particulate amorphous API and voids formed around these particles were found to cause the opacity. The extrusion process was monitored in real time using transflectance NIR; calibration and validation runs were performed using a wide range of drug and plasticiser loadings. Once calibrated, the technique was used to simultaneously track drug and plasticiser content during applied step changes in feedstock material. Rheological and thermal characterisations were used to help understand the morphology of extruded material. The study has shown that it is possible to use a single NIR spectroscopy technique to monitor opaque and transparent melts during HME, and to simultaneously monitor two distinct components within a formulation.

Résumé: Lauthentification est un concept large dans lequel il sagit de pouvoir contrôler ladéquation entre le produit et les informations indiquées sur létiquette. Les techniques utilisant la spectrométrie dans le proche infrarouge dune part, et celles basées sur la PCR dautre part, permettent daborder lauthentification des produits alimentaires sous des angles totalement différents. Loriginalité de ce travail était de les associer pour résoudre deux problèmes précis : 1. lauthentification des poulets de chair à croissance lente utilisés dans des productions de qualité différenciée soumises à des cahiers des charges contraignants ; 2. la détection des farines animales dans lalimentation pour le bétail. Nos travaux sur les poulets de chair nous ont permis de développer des modèles de discrimination reposant sur les spectres dans le proche infrarouge de la viande de poulet. Ils distinguent, pour plus de 80 % des individus, les poulets issus de souches à croissance lente de ceux issus de souches à croissance rapide. Les résultats dune expérimentation animale ont également démontré que la spectrométrie dans le proche infrarouge était capable de mettre évidence des fraudes au niveau de lalimentation des animaux. Deux marqueurs moléculaires spécifiques du type de souche de poulets ont été mis en évidence et caractérisés. Pour le marqueur moléculaire caractéristique des poulets à croissance rapide, un test rapide utilisable en routine a été développé. Dans le cas de la détection des farines animales dans lalimentation du bétail, une méthode de PCR en temps réel sensible et spécifique a été mise au point participant avec succès à des études inter-laboratoires internationales. Associée dans une stratégie originale à la MPIR (microscopie dans le proche infrarouge), elle permet la détection spécifique de particules de farines de viande et dos. Summary: Authentication is a large concept focussing on the control of the correspondance between the product and the information provided on the label. Techniques based on near infrared spectroscopy on the one hand, and those based on PCR on the other hand, allow to tackle the authentication of food and feed products by different sides. The originality of this work was to associate both techniques to solve two specific problems : 1. authentication of slow growing chickens bred in high quality productions according to restricting specifications ; 2. the detection of meat and bone meals in feedingstuffs. The work on the chicken allowed us to develop discriminant models using the near infrared spectra of chicken meat. These models discriminate the chicken from slow- vs. fast-growing chicken for more than 80 % of the animals. The results of an animal experimentation also showed that near infrared spectroscopy was able to detect feeding frauds. Two molecular markers specific of the type of chicken strains were found and characterised. For the one related to the fast-growing chicken strains, a rapid assay applicable in routine testing was conceived. In the case of the meat and bone meal detection in feedingstuffs, a specific and sensitive real time PCR method was developed. It participated succesfully to international inter-laboratory studies. Its combination with NIRM (near infrared microscopy) through an original strategy allows the specific detection of meat and bone meal particles.