Detecting elementary particles and observing accompanying events in particle colliders is one of the most important field of current research in experimental physics. TimePix and its successor TimePix3 are types of the currently used detectors which are placed beside other in ATLAS experiment conducted by Eu- ropean Organization for Nuclear Research. Such detectors can produce huge amount of data about passing particles at high rate. The goal of the thesis is to develop methods for detecting and classification of elementary particles observed by detector network ATLAS-TPX3. Suitable methods for clustering and/or classification based on semi-labelled data should be identified or new one should be developed. The proposed methods will be implemented and their performance on real data will be evaluated. The results will also include an implementation of framework for preprocessing low level data from detector network ATLAS-TPX3 in real-time and creating outputs that are suitable for subsequent physics investigation (e.g. ROOT framework files) includ- ing the proposed or future methods for particle classification. 5

L'imagerie gamma est une technique qui permet la localisation spatiale de sources radioactives. Les différentes applications de cette technique couvrent les phases de démantèlement des installations nucléaires ou de gestion des déchets nucléaires, mais aussi la radioprotection ou la sécurité intérieure. L'utilisation de caméras gamma permet de réduire la dose reçue par les opérateurs, et, par conséquent, de respecter le principe ALARA. Il existe deux techniques d’imagerie permettant la localisation de radioéléments émetteurs gamma : l’imagerie à masque codé et l’imagerie Compton. L’imagerie à masque codé utilise la modulation spatiale du flux de photons gamma incidents par collimateur multi-trous placé entre la source et le détecteur. Elle présente l’avantage d’être extrêmement performante pour des émetteurs gamma « basses énergies », aussi bien en matière de sensibilité, qu’en matière de résolution angulaire. L'imagerie Compton, quant à elle, repose sur l’utilisation de la mécanique de diffusion Compton. L'énergie déposée pendant le processus de diffusion déterminera l'angle de diffusion, et les positions des interactions détermineront la direction des rayons gamma entrants. La position de la source radioactive peut ainsi être limitée à un cône. Si plusieurs cônes sont utilisés, alors la position où le plus grand nombre de cônes se chevauchent correspond à la position de la source radioactive. Une des limitations de cette technique concerne la localisation des émetteurs gamma « basses énergies », pour lesquels la résolution angulaire est fortement dégradée allant jusqu’à l’impossibilité complète de trouver la position. L’objectif de ces travaux est de développer un prototype d’imageur hybride associant les techniques d’imagerie à masque codé et d’imagerie Compton, afin de tirer profit des avantages de chacun des types d’imagerie. Les différents travaux menés, autour du détecteur pixellisé Timepix3, mais aussi en matière de développement d’algorithmes mathématiques, ont permis de proposer deux prototypes d’imageurs hybrides. Les résultats obtenus à l’issue de ces travaux de recherche ont permis de valider expérimentalement les performances d’un des prototypes d’imageurs et d’illustrer l’intérêt d’un système hybride. / Gamma imaging is a technique that allows the spatial localization of radioactive sources. The various applications of this technique cover decommissioning phases of nuclear facilities, nuclear waste management applications, but also radiation protection or Homeland Security. Using gamma camera reduces the dose received by operators and consequently contributes to the respect of the ALARA principle. There are two imaging techniques for the localization of gamma ray emitters: coded aperture imaging and Compton imaging. Coded aperture imaging relies on the spatial modulation of the incident gamma-ray flux by a multi-hole collimator placed between the detector and the radioactive source. It has the advantage of being extremely efficient for « low energy » gamma-ray emitters in terms of sensitivity and angular resolution. On the other hand, Compton imaging is based of the Compton scattering kinematic. The energy deposited during the scattering process will determine the scattering angle, and the positions of the interactions will determine the direction of the incoming gamma-ray. The position of the radioactive source can thus be limited to a cone. If several cones are used, then, the position where the greatest number of cones overlap corresponds to the position of the radioactive source. One limitations of this technique concerns the location of « low energy » gamma-ray emitters, for which the angular resolution is strongly degraded until it is completely not localizable. The objective of this work is to develop a prototype of hybrid imager that combines coded aperture and Compton imaging techniques in order to take advantage of each type of imaging. The different studies carried out, around the Timepix3 pixel detector, but also in the development of mathematical algorithms, have led to propose two prototypes of hybrid imager. The results obtained from this research work made it possible to validate experimentally the performance of one of the imager prototypes, and to illustrate the interest of a hybrid system.

Medical imaging is a technique that allows us to visualize non surgically the internal structure of the human body in order to diagnose or treat medical conditions. It permits also monitoring of physical processes or functions of different organs inside the body. The medical imaging encompasses wide range of techniques based on different physical prin- ciples, including techniques using ionizing radiation. The quality of the images depends significantly on the quality of the used imaging detectors. There are many types of the detectors, from old analog devices (e.g. films) to fully digital detectors such as flat panels, that are the most widely used today. The newer technology is being developed and the techniques such as photon counting explored. However, the state of the art technology is the single photon counting, where the experimental detectors such as Medipix are able to count and process each individual photon. This works studies the properties, features and applications of the newest detector from the Medipix family Timepix3 in different imaging modalities. Firstly, a design of a new hardware readout interface for Timepix3 is presented together with data acquisition software and new analysis and calibration algorithms. Then, different applications of Timepix3 detector were explored: very...

Compton cameras localize γ-ray sources in 3D space by observing evidence of Compton scattering with detectors sensitive to ionizing radiation. This thesis proposes a software system for operating a novel Compton camera device comprised of Timepix3 detectors and Katherine readouts. To communicate with readouts using UDP-based protocol, a dedicated hardware library was developed. The presented software can successfully control the acquisition of multiple Timepix3 detectors and simultaneously process their measurements in a real-time setting. To recognize instances of Compton scattering among observed interactions, a chain of algorithms is applied with explicit consideration for a possibly high volume of measured information. Unlike alternate approaches, the presented work uses a recently published charge drift time model to improve its spatial resolution. In order to achieve localization of γ-ray sources, the software performs conical back projection into a discretized cuboid volume. Results of randomized evaluation with simulated data indicate that the presented implementation is correct and constitutes a viable method of γ-ray source localization in 3D space. Experimental verification with a prototype model is in progress.

In this thesis the verification of a neural network’s proficiency at labeling ionizing radiation particles from the unsorted output of a timepix3 camera is attempted. Focus is put on labeling single particles in separate data sequences with slightly preprocessed input data. Preprocessing of input data is done to simplify the patterns that should be recognized. Two major choices were available for this project, Elman-network and Jordan-network. A more complicated type was not an option because of the longer time needed to implement them. The network type chosen was Elman because of freedom in context size. The neural network is created and trained with the TensorFlow API in python with labeled data that was not created by hand. The network recognized the length difference between gamma particles and alpha particles. Beta particles were not considered by the network. It is concluded that the Elman-style network is not proficient in labeling the sequences, which were considered short enough and to have simple enough input data. A more modern network type is therefore likely required to solve this problem.

Le Timepix3, successeur du Timepix, est un détecteur au silicium composé de deux couches sensibles installées en parallèle. Chaque couche est munie d’une matrice de 65 536 pixels (256x256) et d’une épaisseur de 500 μm. Une des améliorations du Timepix3 par rapport aux générations précédentes est qu’il est possible de récolter simultanément la quantité de charge déposée ainsi que le temps d’arrivée de cette charge. Pour la prise de données Run 3 du LHC qui a débuté en 2022, 16 détecteurs Timepix3 ont été installés dans la caverne du détecteur ATLAS. Les Timepix3 seront utilisés pour mesurer la luminosité du faisceau du LHC ainsi que pour caractériser et mesurer la radiation dans la caverne ATLAS, où beaucoup de composantes électroniques sont installés. L’objet de cette maitrise est de développer un algorithme d’identification des particules qui frappe le détecteur Timepix3. Dans un premier temps, l’information de la quantité d’énergie déposée et du temps d’arrivée sera utilisée pour caractériser un champ de particules incidentes au détecteur Timepix3 (électrons, photons, particules lourdes chargées). La nouvelle méthode consiste à utiliser les paramètres physiques des particules lors de leurs interactions avec le milieu, tels que la trajectoire, l’angle d’incidence, le dépôt d’énergie, la densité spatiale de l’amas, densité énergétique le long de la trajectoire de la particule incidente. Dans un second temps, comme les rayons delta sont des effets récurrents et indésirables qui perturbent l’analyse des données en physique des particules, ce mémoire traitera de la façon de les supprimer pour ne récolter que l’énergie déposée directement par la particule incidente. Il sera aussi question d’utiliser la statistique liée à la production des rayons delta lors du passage d’un flux de particules dans le détecteur pour en déterminer l’énergie cinétique. L’algorithme développé pour caractériser un champ de particules avec le Timepix3 a été confronté aux données obtenues avec un cyclotron de protons à Aahrus au Danemark. Nous avons obtenu des résultats satisfaisants, étant donné que la majorité des particules sont identifiées comme des protons et que nous avons réussi à déterminer l’énergie cinétique de ces protons qui se rapproche de l’énergie cinétique du faisceau de protons utilisé. / The Timepix3, successor to the Timepix, is a silicon detector composed of two sensitive layers mounted in parallel. Each layer has a matrix of 65 536 pixels (256x256) and a thickness of 500 μm. One of the improvements of the Timepix3 compared to previous generations is that it is possible to simultaneously collect the quantity of charge deposited as well as the time of arrival of this charge. For the LHC Run 3 data taking which started in 2022, 16 Timepix3 detectors were installed in the ATLAS detector cavern. The Timepix3 will be used to measure the luminosity of the LHC beam as well as to characterize and measure the radiation in the ATLAS cavern, where many electronic components are installed. The purpose of this master thesis is to develop an algorithm for identifying particles that strike the Timepix3 detector. Initially, information on the amount of energy deposited and the time of arrival will be used to characterize a field of particles incident at the Timepix3 detector (electrons, photons, heavy charged particles). The new method consists in using the physical parameters of the particles during their interactions with the medium, such as the trajectory, the angle of incidence, the energy deposition, the spatial density of the cluster, energy density along the trajectory of the incident particle. Secondly, as delta rays are recurring and undesirable effects which disturb the analysis of data in particle physics, this thesis will deal with how to suppress them in order to harvest only the energy deposited directly by the incident particle. It will also be a question of using the statistics linked to the production of delta rays when a flow of particles passes through the detector to determine their kinetic energy. The algorithm developed to characterize a particle field with the Timepix3 was confronted with data obtained with a proton cyclotron at Aahrus in Denmark. We have obtained satisfactory results, given that the majority of the particles are identified as protons and that we have succeeded in determining the kinetic energy of these protons which is close to the kinetic energy of the proton beam used.