Magnetic Resonance Spectroscopy (MRS) can be applied to probe noninvasively the concentrations and distribution of metabolites of human tissue in vivo. As the improving of hardware and localization techniques, MRS becomes more and more important in clinical applications. Furthermore, some post-processing software, like LCModel, provide a graphical user interface for efficient and convenient analysis of MR spectroscopic imaging and thus increase the value of MRS applications. Although LCModel provides an efficient analysis and produces stable results, it can not provide metabolite map to observe the distribution of metabolite concentrations. For this reason our study processes the output data of LCModel and Digital Imaging and Communications in Medicine (DICOM) format MR images for 2D/3D metabolite map displaying. Users can use this software to observe the metabolic distribution in AP, SI and RL slice of brain tissue. In the meanwhile, as the absolute quantification of MRS has played more and more important role in clinical applications, this study also provides the LCModel end users an easy way for interpretation.

In recent years, dramatic evolution of precision agriculture technology has been driven by technologies such as sensors and controllers, telematics, and UAV. An intriguing area in the field of precision agriculture and UAV is big volume of data and its analysis that have not been dealt with in depth. The main objective of this project is proposed to develop a digital data processing tool and deliver them in a useful format to the end users. This involves three faceted methodical approach: 1) collection of large amounts of UAV images with regards to combinations of in-field sensors and UAV imagery 2) analyze and validate the combination of data with geospatial tools 3) Develop a web-based computer data processing program to analyze the big data and assess pant condition. The validation and correlation analysis results showed that UAV assessed NDVI are good indicator of crop nutrition along with the ground-based crop sensors.

Bryne AB has invented a device called Loop for measuring the fluidity of molten metal. Loop is made up of refractory fiber material and has a dimension of 196 sq.mm within which there is an engraved spiral shaped runway for the molten metal to flow. The distance traveled by the molten metal in the spiral determines its fluidity and is measured visually by the operator with the help of the spiral scale printed on the Loop. A workstation is used to support the Loop for an easier and better working environment. At present, the entire operation of fluidity measurement is done manually. The aim of the thesis is to develop an automatic workstation for the fluidity measurement and find a technique to measure the fluidity without human visual aid. Further, an attempt has been made to find a way to store and retrieve the measured values for future use.   The automation of the workstation is done with the help of Arduino Uno circuit board for electrical connections and Arduino IDE for programming the circuit. MATLAB image processing tool is used for measuring the fluidity digitally.   The prototype of the automated workstation is built and showcases the ability to reduce the time and operational errors. The automation of the unplugging operation saves up to 7 seconds.  The digital measuring of the fluidity with FLIR thermal camera and MATLAB image processing tool does not show significant improvement in accuracy in measurement, but it reduces the dependency on the expertise of the operator. There is an error of 12.7% from the actual fluidity value.

Magnetic resonance (MR) has been developed and applied to clinical analysis studies due to its non-invasive properties. Because of the increasing interest of applying magnetic resonance spectroscopy imaging (MRSI) to clinical application, some post-processing softwares, like LCModel, provide a graphical user interface for convenient and efficient analysis. However, the features of combining MR imaging (MRI) with MRS information and browsing all analyzed results are not provided by LCModel. Our study proposed a method to implement the architecture for processing General Electric (GE), Siemens MRSI data sets and provides features including interactive display, selection and analysis of full 2D slices. For multi-console analysis, our tool also provides the combination of MRS, MRI, and data sets generated by LCModel, such as the projection of three planes and metabolite/spectra map, and therefore the three formats of data sets could be obtained from scanners of various manufactures. Especially, it is more complicated when processing GE data sets, so some mechanisms for processing are proposed, like the transformation, the three plane loc images detection and MRSI detection, etc. Additionally, our tool also has the advantage of the compatibility of further extended functionalities, which would be more flexible and useful for clinical applications.

Spray drying is a process, which produces powders from the fluid state. This type of process is mostly used in the industrial sector. In this process, a liquid slurry is atomized, forming droplets, which are dried with hot air. During spray drying these droplets will interact and upon impact can show different types of interactions; droplet-droplet collisions as well as interactions with partially or completely dried particles, leading to agglomeration. The result of collision gives properties of the dried powder. The focus of the thesis is to investigate the droplet-droplet collision outcomes of WPC 80 (Whey Protein Concentrate 80) and Lactose. Then the effects of the absolute droplet diameter and the droplet diameter ratios are to be determined. Existing experimental setup and Image Processing Tool of MATLAB is used to study the collision outcome. The outcomes are shown in a regime map. The present results are compared with different products result and literature study. It is observed that there is an effect on collision outcome for different droplet size ratios and no effect for absolute droplet diameter.

Tesis por compendio / [ES] Esta tesis aborda el desarrollo de métodos de procesado y análisis de datos ALSFW para la caracterización de la estructura vertical del bosque y, en particular, del sotobosque. Para responder a este objetivo general, se establecieron seis objetivos específicos: En primer lugar, se analiza la influencia de la densidad de pulso, de los parámetros de voxelización (tamaño de vóxel y valor de asignación) y de los métodos de regresión sobre los valores de las métricas ALSFW y sobre la estimación de atributos de estructura del bosque. Para ello, se redujo aleatoriamente la densidad de pulsos y se modificaron los parámetros de voxelización, obteniendo los valores de las métricas ALSFW para las diferentes combinaciones de parámetros. Estas mismas métricas ALSFW se emplearon para la estimación de atributos de la estructura del bosque mediante diferentes métodos de regresión. En segundo lugar, se integran métodos de procesado y análisis de datos ALSFW en una nueva herramienta llamada WoLFeX (Waveform Lidar for Forestry eXtraction) que incluye los procesos de recorte, corrección radiométrica relativa, voxelización y extracción de métricas a partir de los datos ALSFW, así como nuevas métricas descriptoras del sotobosque. En tercer lugar, se evalúa la influencia del ángulo de escaneo utilizado en la adquisición de datos ALS y la corrección radiométrica en la extracción de métricas ALSFW y en la estimación de atributos de combustibilidad forestal. Para ello, se extrajeron métricas ALSFW con y sin corrección radiométrica relativa y empleando diferentes ángulos de escaneo. En cuarto lugar, se caracteriza la oclusión de la señal a lo largo de la estructura vertical del bosque empleando y comparando tres tipos diferentes de láser escáner (ALSFW, ALSD y láser escáner terrestre: TLS, por sus siglas en inglés), determinando así sus limitaciones en la detección de material vegetativo en dos ecosistemas forestales diferenciados: el boreal y el mediterráneo. Para cuantificar la oclusión de la señal a lo largo de la estructura vertical del bosque se propone un nuevo parámetro, la tasa de reducción del pulso, basada en el porcentaje de haces láser bloqueados antes de alcanzar una posición dada. En quinto lugar, se evalúa la forma en que se detectan y determinan las clases de densidad de sotobosque mediante los diferentes tipos de ALS. Se compararon los perfiles de distribución vertical en los estratos inferiores descritos por el ALSFW y el ALSD con respecto a los descritos por el TLS, utilizando este último como referencia. Asimismo, se determinaron las clases de densidad de sotobosque aplicando la curva Lorenz y el índice Gini a partir de los perfiles de distribución vertical descritos por ALSFW y ALSD. Finalmente, se aplican y evalúan las nuevas métricas ALSFW basadas en la voxelización, utilizando como referencia los atributos extraídos a partir del TLS, para estimar la altura, la cobertura y el volumen del sotobosque en un ecosistema mediterráneo. / [EN] This thesis addresses the development of ALSFW processing and analysis methods to characterize the vertical forest structure, in particular, the understory vegetation. To answer this overarching goal, a total of six specific objectives were established: Firstly, the influence of pulse density, voxel parameters (i.e., voxel size and assignation value) and regression methods on ALSFW metric values and on estimates of forest structure attributes are analyzed. To do this, pulse density was randomly reduced and voxel parameters modified, obtaining ALSFW metric values for the different parameter combinations. These ALSFW metrics were used to estimate forest structure attributes with different regression methods. Secondly, a set of ALSFW data processing and analysis methods are integrated in a new software named WoLFeX (Waveform Lidar for Forestry eXtraction), including clipping, relative radiometric correction, voxelization and ALSFW metric extraction, and proposing new metrics for understory vegetation. Thirdly, the influence of the scan angle of ALS data acquisition and radiometric correction on the extraction of ALSFW metrics and on modeling forest fuel attributes is assessed. To do this, ALSFW metrics were extracted applying and without applying relative radiometric correction and using different scan angles. Fourthly, signal occlusion is characterized along the vertical forest structure using and comparing three different laser scanning configurations (ALSFW, ALSD and terrestrial laser scanning: TLS), determining their limitations in the detection of vegetative material in two contrasted forest ecosystems: boreal and Mediterranean. To quantify signal occlusion along the vertical forest structure, a new parameter based on the percentage of laser beams blocked prior to reach a given location, the rate of pulse reduction, is proposed. Fifthly, the assessment of how understory vegetation density classes are detected and determined by different ALS configurations is done. Vertical distribution profiles at the lower strata described by ALSFW and ALSD are compared with those described by TLS as reference. Moreover, understory vegetation density classes are determined by applying the Lorenz curve and Gini index from the vertical distribution profiles described by ALSFW and ALSD. Finally, the new proposed voxel-based ALSFW metrics are applied and evaluated, using TLS-based attributes as a reference, to estimate understory height, cover and volume in a Mediterranean ecosystem. / [CA] Aquesta tesi aborda el desenvolupament de mètodes de processament i anàlisi de dades ALSFW per a la caracterització de l'estructura vertical del bosc i, en particular, del sotabosc. Per a respondre a aquest objectiu general, s'establiren sis objectius específics: En primer lloc, s'analitza la influència de la densitat de pols, dels paràmetres de voxelització (grandària de vóxel i valor d'assignació) i dels mètodes de regressió sobre els valors de les mètriques ALSFW i sobre l'estimació dels atributs d'estructura del bosc. Per a això, es reduí aleatòriament la densitat de polsos i es modificaren els paràmetres de voxelització, obtenint els valors de les mètriques ALSFW per a les diferents combinacions de paràmetres. Aquestes mètriques ALSFW s'empraren per a l'estimació d'atributs de l'estructura del bosc mitjançant diferents mètodes de regressió. En segon lloc, s'integraren mètodes de processament i d'anàlisi de dades ALSFW en una nova eina anomenada WoLFeX (Waveform Lidar for Forestry eXtraction) que inclou el processos de retallada, correcció radiomètrica relativa, voxelització i extracció de mètriques a partir de les dades ALSFW, així com noves mètriques descriptores del sotabosc. En tercer lloc, s'avalua la influència de l'angle de escaneig emprat en l'adquisició de les dades ALS i la correcció radiomètrica en l'extracció de mètriques ALSFW i en l'estimació d'atributs de combustibilitat forestal. Per a això, s'extragueren mètriques ALSFW amb i sense correcció radiomètrica relativa i emprant diferents angles d'escaneig. En quart lloc, es caracteritza l'oclusió del senyal al llarg de l'estructura vertical del bosc emprant i comparant tres tipus diferents de làser escàner (ALSFW, ALSD i làser escàner terrestre: TLS, per les seues sigles en anglès), determinant així les seues limitacions en la detecció de material vegetatiu en dos ecosistemes diferenciats: un boreal i un mediterrani. Per a quantificar l'oclusió del senyal al llarg de l'estructura vertical del bosc es proposa un nou paràmetre, la taxa de reducció del pols, basada en el percentatge de rajos làser bloquejats abans d'arribar a una posició donada. En cinquè lloc, s'avalua la manera en la qual es detecten i determinen les classes de densitat de sotabosc mitjançant els diferents tipus d'ALS. Es compararen els perfils de distribució vertical en estrats inferiors descrits per l'ALSFW i l'ALSD respecte als descrits pel TLS, emprant aquest últim com a referència. A més a més, es determinaren les classes de densitat de sotabosc aplicant la corba Lorenz i l'índex Gini a partir dels perfils de distribució vertical descrits per l'ALSFW i l'ALSD. Finalment, s'apliquen i avaluen les noves mètriques ALSFW basades en la voxelització, emprant com a referència els atributs extrets a partir del TLS, per a estimar l'alçada, la cobertura i el volum del sotabosc en un ecosistema mediterrani. / Crespo Peremarch, P. (2020). Processing and analysis of airborne fullwaveform laser scanning data for the characterization of forest structure and fuel properties [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/153715 / Compendio

Indiana University-Purdue University Indianapolis (IUPUI) / Graph Neural Networks (GNNs) have become excessively popular and prominent deep learning techniques to analyze structural graph data for their ability to solve complex real-world problems. Because graphs provide an efficient approach to contriving abstract hypothetical concepts, modern research overcomes the limitations of classical graph theory, requiring prior knowledge of the graph structure before employing traditional algorithms. GNNs, an impressive framework for representation learning of graphs, have already produced many state-of-the-art techniques to solve node classification, link prediction, and graph classification tasks. GNNs can learn meaningful representations of graphs incorporating topological structure, node attributes, and neighborhood aggregation to solve supervised, semi-supervised, and unsupervised graph-based problems. In this study, the usefulness of GNNs has been analyzed primarily from two aspects - clustering and classification. We focus on these two techniques, as they are the most popular strategies in data mining to discern collected data and employ predictive analysis.

Samtliga nya detaljplaner behöver genomgå en undersökning tidigt i planprocessen för att bedöma om planen tros kunna orsaka en betydande miljöpåverkan och i så fall behöva genomgå en strategisk miljöbedömning. Structor Miljöteknik arbetar med miljöbedömningar av detaljplaner med målsättningen är att väva in miljöanpassningsåtgärder i ett så tidigt skede av planeringen som möjligt för att bidra till en hållbar stadsplanering. En strategi med tillhörande processverktyg togs fram i samband med ett tidigare examensarbete för att underlätta detta arbetssätt. Verktyget är anpassad för användare som redan är väl insatta i dem framtagna strategin, och kan därmed upplevas som tung att sätta sig in i. För att fler ska kunna tänka sig att använda det och för att den framtagna strategin ska få större genomslag hos Sveriges kommuner behöver därför verktyget vidareutvecklas och bli mer användarvänligt.  En utvärderande studie gjordes av den arbetsprocess som Structor Miljöteknik har för undersökningar i syfte att identifiera dess styrkor och svagheter. Utifrån denna studie, intervjuer med Naturvårdsverket, Boverket, Länsstyrelsen i Örebro län, Motala kommun och konsulter från Structor Miljöteknik samt en egen utvärdering av det befintliga verktyget togs flera kriterier fram för det nya verktyget. Kriterierna gavs olika prioriteringar som baserades på i vilken grad uppfyllandet av kriteriet ansågs bidra till ökad funktionalitet, automatisering och användarvänlighet. Microsoft Excel identifierades som en lämplig programvara för verktyget då majoriteten av kriterierna bedömdes kunna uppfyllas i det. En plan för vidareutvecklingen av verktyget togs fram som delades in i 4 etapper: framtagande av grund, uppfylla satta kriterier, författa en vägledning och tester och utvärdering. På grund av tidsbrist färdigställdes dock endast den första etappen fullt ut. Grundversionen av det nya verktyget gav goda förutsättningar att uppfylla kriterierna med hög prioritet samt merparten av de övriga.  Möjligheten att uppfylla de olika framtagna kriterierna och tillvägagångsättet för att i ett vidare arbete slutföra utvecklingsprocessen analyserades och presenterades. / When a new local plan is crafted an early screening during the planning process is required to examine if the plan has significant environmental impact. If the plan is believed to have significant environmental impact a strategic environmental assessment needs to be conducted. The company Structor Miljöteknik has a way of working with the environmental assessment of plans where the goal is to incorporate measures striving for greater environmental sustainability early in the process. Today the company does this using a strategy and a processing tool that was created by Sandström. The use of the processing tools is limited by the fact that it requires a user with deeper understanding of the tools construction to be able to use it. Therefore, there is a need for further development of the processing tool to achieve a more user-friendly interface, so that its use become more widespread and the environmental profits of it can be reaped. An evaluative study has been conducted to pinpoint the strength and weaknesses in the process used by Structor Miljöteknik currently. Within this study extensive interviews have been conducted with different government agencies, country administrative board, municipality and people working in the private company Structor Miljöteknik. Based on the interviews several criteria for a developed processing tool were identified. All the criteria were then rated based on three parameters: functionality, automation and user friendliness. Microsoft Excel was identified as suitable software product to use for the processing tool since it enabled most of the criteria to be fulfilled. The further development of the processing tool was divided into four steps: creating a foundation, fulfilling the set criteria, creating a manual and tests and evaluation. In this report the first step was completed, laying a foundation that has great potential for achieving the criteria with high priority as well as most of the other criteria when the further steps are developed.

<p><strong>Graph Neural Networks</strong> (GNNs) have become excessively popular and prominent deep learning techniques to analyze structural graph data for their ability to solve complex real-world problems. Because graphs provide an efficient approach to contriving abstract hypothetical concepts, modern research overcomes the limitations of classical graph theory, requiring prior knowledge of the graph structure before employing traditional algorithms. GNNs, an impressive framework for representation learning of graphs, have already produced many state-of-the-art techniques to solve node classification, link prediction, and graph classification tasks. GNNs can learn meaningful representations of graphs incorporating topological structure, node attributes, and neighborhood aggregation to solve supervised, semi-supervised, and unsupervised graph-based problems. In this study, the usefulness of GNNs has been analyzed primarily from two aspects - <strong>clustering and classification</strong>. We focus on these two techniques, as they are the most popular strategies in data mining to discern collected data and employ predictive analysis.</p>