Dual Energy Computed Tomography (DECT) is an emerging technique that offers new possibilities to determine composition of tissues in clinical applications. Accurate knowledge of tissue composition is important for instance for brachytherapy (BT) treatment planning. However, the accuracy of CT numbers measured with contemporary clinical CT scanners is relatively low since CT numbers are affected by image artifacts. The aim of this work was to estimate the accuracy of CT numbers measured with the Siemens SOMATOM Definition Flash DECT scanner and the accuracy of the resulting volume or mass fractions calculated via the three material decomposition method. CT numbers of water, gelatin and a 3rd component (salt, hydroxyapatite or protein powder) mixtures were measured using Siemens SOMATOM Definition Flash DECT scanner. The accuracy of CT numbers was determined by (i) a comparison with theoretical (true) values and (ii) using different measurement conditions (configurations) and assessing the resulting variations in CT numbers. The accuracy of mass fractions determined via the three material decomposition method was estimated by a comparison with mass fractions measured with calibrated scales. The latter method was assumed to provide highly accurate results. It was found that (i) axial scanning biased CT numbers for some detector rows. (ii) large volume of air surrounding the measured region shifted CT numbers compared to a configuration where the region was surrounded by water. (iii) highly attenuating object shifted CT numbers of surrounding voxels. (iv) some image kernels caused overshooting and undershooting of CT numbers close to edges. The three material decomposition method produced mass fractions differing from true values by 8% and 15% for the salt and hydroxyapatite mixtures respectively. In this case, the analyzed CT numbers were averaged over a volumetric region. For individual voxels, the volume fractions were affected by statistical noise. The method failed when statistical noise was high or CT numbers of the decomposition triplet were similar. Contemporary clinical DECT scanners produced image artifacts that strongly affected the accuracy of the three material decomposition method; the Siemens’ image reconstruction algorithm is not well suited for quantitative CT. The three material decomposition method worked relatively well for averages of CT numbers taken from volumetric regions as these averages lowered statistical noise in the analyzed data.

The current thesis concerns Dual Energy Computed Tomography and specifically the physical principles and methods it is based on. Dual Energy CT offers the potential of not only anatomical, but also functional information from Computed Tomography (CT) exams. This is achieved by utilizing the energy dependence of X-rays’ attenuation within matter. In this way, materials are divided into those that are characterized by energy-dependent attenuation (strong spectral behavior), due to strong photoelectric effect contribution to total attenuation, and those that do not exhibit important photoelectric attenuation at radiological energies and therefore they attenuate X-rays in a much less energy dependent way. This information is useful for the identification of materials that, despite the fact that they are completely different as far as their chemical composition is concerned, they have the same or similar CT number values at a particular kVp level. The energy dependence of attenuation leads to the determination of a polychromatic linear attenuation coefficient. This coefficient may be approximated either by considering an equivalent monoenergetic attenuation coefficient that is characterized by the same half value layer as the the polyenergetic beam, or by a local linear attenuation coefficient that is determined by knowledge of the local x-ray spectrum. The energy dependence of attenuation is the cause of beam hardening effects. The basic fields where dual energy CT has become feasible and its current clinical applications are described in the thesis. The utility of DECT ranges from artifact elimination (beam hardening, metal artifacts) to tissue discrimination, material selective images and “conventional CT acquisition” equivalent images. The implementations of dual energy CT are also presented in the thesis and include consecutive scans at two different kVp values, fast kV-switching, dual source CT and dual layer CT. / Η παρούσα διπλωματική εργασία αφορά στην τομογραφία διπλής ενέργειας και ειδικότερα τις φυσικές αρχές και μεθόδους στις οποίες βασίζεται. H τομογραφία διπλής ενέργειας δίνει την προοπτική της απόκτησης όχι μόνο ανατομικής, αλλά και λειτουργικής πληροφορίας από τις εξετάσεις αξονικής τομογραφίας. Αυτό το επιτυγχάνει χρησιμοποιώντας την εξάρτηση της εξασθένησης των ακτίνων X μέσα στην ύλη από την ενέργεια των φωτονίων. Με αυτό τον τρόπο, τα υλικά διαχωρίζονται σε αυτά που εξασθενούν τα φωτόνια με πολύ διαφορετικό τρόπο σε διαφορετικές ενέργειες λόγω έντονου φωτοηλεκτρικού φαινομένου και σε αυτά που η ενεργειακή εξάρτηση του συντελεστή εξασθένησης τους είναι λιγότερο έντονη (λιγότερη συμμετοχή του φωτοηλεκτρικού φαινομένου στη συνολική εξασθένηση). Η πληροφορία αυτή είναι χρήσιμη για την ταυτοποίηση της σύστασης υλικών, που ενώ είναι εντελώς διαφορετικά σε χημική σύσταση, έχουν τον ίδιο αριθμό CT (CT number) σε συγκεκριμένη τάση λειτουργίας της πηγής ακτίνων X. Στην εκτεταμένη περίληψη αυτής της εργασίας, αρχικά γίνεται σύντομη αναφορά στην εξάρτηση του συντελεστή εξασθένησης από την ενέργεια και πώς αυτή επηρεάζει τον ορισμό του συντελεστή γραμμικής εξασθένησης για πολυχρωματικές ακτινοβολίες. Στη συνέχεια ακολουθεί μια συνοπτική περιγραφή υλοποιήσεων της τομογραφίας διπλής ενέργειας καθώς επίσης και των αλγορίθμων και των κλινικών εφαρμογών που είναι διαθέσιμες σήμερα. Τέλος, περνώντας στην περιγραφή του κυρίως μέρους της διπλωματικής, ακολουθεί συνοπτική περιγραφή της θεωρητικής και πειραματικής μελέτης της συμπεριφοράς διπλής ενέργειας του ιωδίου, του ασβεστίου, του οστού και υλικών που προσομοιάζουν το μαλακό και λιπώδη ιστό. Στη θεωρητική περιγραφή, η διπλωματική εστιάζει στις εξαρτήσεις του συντελεστή γραμμικής εξασθένησης, πρώτα από τα χαρακτηριστικά του ίδιου του υλικού (πυκνότητα, χημική σύσταση) και έπειτα από την ενέργεια των φωτονίων. Ο προβληματισμός σχετικά με την εξάρτηση του συντελεστή εξασθένησης με την ενέργεια έχει οδηγήσει συχνά σε τρόπους προσέγγισης και ορισμού ενός συντελεστή γραμμικής εξασθένησης που να ανταποκρίνεται σε πολυενεργειακές δέσμες, όπως αυτές που χρησιμοποιεί ο αξονικός τομογράφος. Ο συντελεστής γραμμικής εξασθένησης μιας πολυενεργειακής δέσμης από ένα συγκεκριμένο υλικό, προσδιορίζεται είτε μέσω ενός ισοδύναμου μονοενεργειακού συντελεστή εξασθένησης που να χαρακτηρίζεται από το ίδιο half value layer (HVL) με την πολυενεργειακή δέσμη είτε μέσω ενός «τοπικού» συντελεστή εξασθένησης. που για να προσδιοριστεί απαιτείται γνώση του φάσματος στην συγκεκριμένη θέση και στο συγκεκριμένο υλικό. Η εξάρτηση του συντελεστή εξασθένησης από την ενέργεια των φωτονίων ευθύνεται για τεχνήματα σκλήρυνσης δέσμης (beam hardening artifacts).

Accurate, cross-scanner assessment of in-vivo air density used to quantitatively assess amount and distribution of emphysema in COPD subjects has remained elusive. Hounsfield units (HU) within tracheal air can be considerably more positive than -1000 HU. With the advent of new dual-source scanners which employ dedicated scatter correction techniques, it is of interest to evaluate how the quantitative measures of lung density compare between dual-source and single-source scan modes. This study has sought to characterize in-vivo and phantom-based air metrics using dual-energy computed tomography technology where the nature of the technology has required adjustments to scatter correction. Anesthetized ovine (N=6), swine (N=13: more human-like rib cage shape), lung phantom and a thoracic phantom were studied using a dual-source MDCT scanner (Siemens Definition Flash. Multiple dual-source dual-energy (DSDE) and single-source (SS) scans taken at different energy levels and scan settings were acquired for direct quantitative comparison. Density histograms were evaluated for the lung, tracheal, water and blood segments. Image data were obtained at 80, 100, 120, and 140 kVp in the SS mode (B35f kernel) and at 80, 100, 140, and 140-Sn (tin filtered) kVp in the DSDE mode (B35f and D30f kernels), in addition to variations in dose, rotation time, and pitch. To minimize the effect of cross-scatter, the phantom scans in the DSDE mode was obtained by reducing the tube current of one of the tubes to its minimum (near zero) value. When using image data obtained in the DSDE mode, the median HU values in the tracheal regions of all animals and the phantom were consistently closer to -1000 HU regardless of reconstruction kernel (chapters 3 and 4). Similarly, HU values of water and blood were consistently closer to their nominal values of 0 HU and 55 HU respectively. When using image data obtained in the SS mode the air CT numbers demonstrated a consistent positive shift of up to 35 HU with respect to the nominal -1000 HU value. In vivo data demonstrated considerable variability in tracheal, influenced by local anatomy with SS mode scanning while tracheal air was more consistent with DSDE imaging. Scatter effects in the lung parenchyma differed from adjacent tracheal measures. In summary, data suggest that enhanced scatter correction serves to provide more accurate CT lung density measures sought to quantitatively assess the presence and distribution of emphysema in COPD subjects. Data further suggest that CT images, acquired without adequate scatter correction, cannot be corrected by linear algorithms given the variability in tracheal air HU values and the independent scatter effects on lung parenchyma.

Computed tomography (CT) is a widely used medical imaging modality. By rotating an x-ray tube and an x-ray detector around the patient, a CT scanner is able to measure the x-ray transmission from all directions and form an image of the patient’s interior. CT scanners in clinical use today all use energy-integrating detectors, which measure the total incident energy for each measurement interval. A photon-counting detector, on the other hand, counts the number of incoming photons and can in addition measure the energy of each photon by comparing it to a number of energy thresholds. Using photon- counting detectors in computed tomography could lead to improved signal-to-noise ratio, higher spatial resolution and improved spectral imaging which allows better visualization of contrast agents and more reliable quantitative measurements. In this Thesis, the feasibility of using a photon-counting silicon-strip detector for CT is investigated. In the first part of the Thesis, the necessary performance requirements on such a detector is investigated in two different areas: the detector element homogeneity and the capability of handling high photon fluence rates. A metric of inhomogeneity is proposed and used in a simulation study to evaluate different inhomogeneity compensation methods. Also, the photon fluence rate incident on the detector in a scanner in clinical use today is investigated for different patient sizes through dose rate measurements together with simulations of transmission through patient im- ages. In the second part, a prototype detector module is used to demonstrate new applications enabled by the energy resolution of the detector. The ability to generate material-specific images of contrast agents with iodine and gadolinium is demonstrated. Furthermore, it is shown theoretically and ex- perimentally that interfaces in the image can be visualized by imaging the so-called nonlinear partial volume effect. The results suggest that the studied silicon-strip detector is a promising candidate for photon-counting CT.

It is of clinical interest to decompose a three material mixture into its constituted substances using dual-energy CT. In radiation therapy, for example material decomposition can be used to determine tissue properties for the calculation of dose in treatment planning. Due to use of polychromatic spectrum in CT, beam hardening artifacts prevent to achieve fully satisfactory results. Here an iterative reconstruction algorithm proposed by A. Malusek, M. Magnusson, M.Sandborg, and G. Alm Carlsson in 2008 is implemented to achieve this goal. The iterative algorithm can be implemented with both single- and dual-energy CT. The material decomposition process is based on mass conservation and volume conservation assumptions. The implementation and evaluation of iterative reconstruction algorithm is done by using simulation studies of analyzing mixtures of water, protein and adipose tissue. The results demonstrated that beam hardening artifacts are effectively removed and accurate estimation of mass fractions of each base material can be achieved with the proposed method. We also compared our novel iterative reconstruction algorithm to the commonly used water pre-correction method. Experimental results show that our novel iterative algorithm is more accurate.

The goal in medical x-ray imaging is to obtain the image quality requiredfor a given detection task, while ensuring that the patient dose is kept as lowas reasonably achievable. The two most common strategies for dose reductionare: optimizing incident x-ray beams and utilizing energy informationof transmitted beams with new detector techniques (spectral imaging). Inthis thesis, dose optimization schemes were investigated in two x-ray imagingsystems: digital mammography and computed tomography (CT). In digital mammography, the usefulness of anti-scatter grids was investigatedas a function of breast thickness with varying geometries and experimentalconditions. The general conclusion is that keeping the grid is optimalfor breasts thicker than 5 cm, whereas the dose can be reduced without a gridfor thinner breasts. A photon-counting silicon-strip detector developed for spectral mammographywas characterized using synchrotron radiation. Energy resolution, ΔE/Ein, was measured to vary between 0.11-0.23 in the energy range 15-40 keV, which is better than the energy resolution of 0.12-0.35 measured inthe state-of-the-art photon-counting mammography system. Pulse pileup hasshown little effect on energy resolution. In CT, the performance of a segmented silicon-strip detector developedfor spectral CT was evaluated and a theoretical comparison was made withthe state-of-the-art CT detector for some clinically relevant imaging tasks.The results indicate that the proposed photon-counting silicon CT detector issuperior to the state-of-the-art CT detector, especially for high-contrast andhigh-resolution imaging tasks. The beam quality was optimized for the proposed photon-counting spectralCT detector in two head imaging cases: non-enhanced imaging and Kedgeimaging. For non-enhanced imaging, a 120-kVp spectrum filtered by 2half value layer (HVL) copper (Z = 29) provides the best performance. Wheniodine is used in K-edge imaging, the optimal filter is 2 HVL iodine (Z = 53)and the optimal kVps are 60-75 kVp. In the case of gadolinium imaging, theradiation dose can be minimized at 120 kVp filtered by 2 HVL thulium (Z =69). / <p>QC 20160401</p>

Nous avons étudié l’utilisation de détecteurs spectrométriques, qui émergent dans le domaine de l’imagerie médicale, pour leur application à la mammographie. Ces détecteurs permettent de discriminer l’énergie des photons reçus, ce qui apporte une information supplémentaire à l’imagerie d’atténuation traditionnelle. Ainsi, il est possible d’utiliser des techniques de décomposition en base de deux matériaux, notamment pour déterminer la densité glandulaire dans le sein, qui correspond au pourcentage de tissus glandulaires, et qui est un facteur de risque pour le développement d’un cancer, à partir d’une seule irradiation. Jusqu’alors, il était possible d’utiliser cette méthode à partir de deux expositions à deux énergies distinctes. Dans certains cas, une nouvelle tendance consiste à pratiquer des mammographies avec injection d’un produit de constratse iodé, mais cela nécessite également au moins deux irradiations. Nous avons donc proposé d’estimer la densité du sein et la concentration d’iode simultanément, à partir d’une seule irradiation, à une dose 0,93 mGy, en appliquant des méthodes de décomposition en base de trois matériaux. Premièrement, des méthodes polynomiales ont été adaptées pour être comptibles avec l’information spectrale provenant de 3 canaux d’énergies. Ensuite, nous avons montré qu’une deuxième approche, capable de prendre en compte une information spectrale plus fine, basée sur la maximisation de la vraisemblance entre un spectre mesuré et des spectres de références, était capable d’atteindre de meilleurs résultats. Enfin, nous avons développé une méthode capable de prendre en compte la compression du sein en mammographie pour améliorer les résultats obtenus par la méthode de maximum de vraisemblance. / Energy sensitive X-ray detectors are emerging in the field of medical imaging. We have investigated the use of this new type of X-ray detectos for their application to mammography exams. These detectors are able to discriminate the energy of received photons, which provides additional information to a standard mammography image only composed of the total attenuation signal. Thus, these detectors allow the use of basis material decomposition techniques, from a single x-ray exposure, and permit to determine the breast density, which corresponds to the percentage of glandular tissues in the breast. Breast density is known for being a risk factor for the development of breast cancers. Without energy sensitive X-ray detectors, this method requires two X-ray exposures at different energies. Contrast enhanced mammography is also developing but it requires the use an iodinated contrast media and at least two irradiations. Hence, we proposed to take benefit of energy-sensitive detectors to simultaneously estimate the breast density and the iodine concentration, using a single X-ray exposure at a mean glandular dose of 0.93 mGy. This approach is based on three basis material decomposition methods. First, different polynomial methods have been adapted to comply with spectral information from 3 energy channels. Then, we showed that a second approach, based on the maximisation of the likelihood between a measured spectrum and reference spectra, was able take into consideration finer spectral information and achieved better results. Finally, we have developed a method that can take into consideration the thickness of the compressed breast during a mammography exam to improve the results obtained by the maximum likelihood method.

La tomographie par rayons X multi-énergétiques (MECT) permet d'obtenir plus d'information concernant la structure interne de l'objet par rapport au scanner CT classique. Un de ses intérêts dans l’imagerie médicale est d'obtenir les images de fractions d’eau et d’os. Dans l'état de l'art, les intérêts de MECT n'est pas encore découvert largement. Les approches de reconstruction existantes sont très limitées dans leurs performances. L'objectif principal de ce travail est de proposer des approches de reconstruction de haute qualité qui pourront être utilisés dans la MECT afin d’améliorer la qualité d’imagerie.Ce travail propose deux approches de reconstruction bayésiennes. La première est adaptée au système avec un détecteur discriminant en énergie. Dans cette approche, nous considérons que les polychromaticités de faisceaux sont négligeables. En utilisant le modèle linéaire de la variance et la méthode d'estimation maximum à postériori (MAP), l'approche que nous avons proposé permets de prendre en compte les différents niveaux de bruit présentés sur les mesures multi-énergétiques. Les résultats des simulations montrent que, dans l'imagerie médicale, les mesures biénergies sont suffisantes pour obtenir les fractions de l'eau et de l'os en utilisant l'approche proposée. Des mesures à la troisième énergie est nécessaire uniquement lorsque l'objet contient des matériaux lourdes. Par exemple, l’acier et l'iode. La deuxième approche est proposée pour les systèmes où les mesures multi-énergétiques sont obtenues avec des faisceaux polychromatiques. C'est effectivement la plupart des cas dans l'état actuel du practice. Cette approche est basée sur un modèle direct non-linéaire et un modèle bruit gaussien où la variance est inconnue. En utilisant l’inférence bayésienne, les fractions de matériaux de base et de la variance d'observation pourraient être estimées à l'aide de l'estimateur conjoint de MAP. Sous réserve à un modèle a priori Dirac attribué à la variance, le problème d'estimation conjointe est transformé en un problème d'optimisation avec une fonction du coût non-quadratique. Pour le résoudre, l'utilisation d'un algorithme de gradient conjugué non-linéaire avec le pas de descente quasi-optimale est proposée.La performance de l'approche proposée est analysée avec des données simulées et expérimentales. Les résultats montrent que l'approche proposée est robuste au bruit et aux matériaux. Par rapport aux approches existantes, l'approche proposée présente des avantages sur la qualité de reconstruction. / Multi-Energy Computed Tomography (MECT) makes it possible to get multiple fractions of basis materials without segmentation. In medical application, one is the soft-tissue equivalent water fraction and the other is the hard-matter equivalent bone fraction. Practical MECT measurements are usually obtained with polychromatic X-ray beams. Existing reconstruction approaches based on linear forward models without counting the beam polychromaticity fail to estimate the correct decomposition fractions and result in Beam-Hardening Artifacts (BHA). The existing BHA correction approaches either need to refer to calibration measurements or suffer from the noise amplification caused by the negative-log pre-processing and the water and bone separation problem. To overcome these problems, statistical DECT reconstruction approaches based on non-linear forward models counting the beam polychromaticity show great potential for giving accurate fraction images.This work proposes a full-spectral Bayesian reconstruction approach which allows the reconstruction of high quality fraction images from ordinary polychromatic measurements. This approach is based on a Gaussian noise model with unknown variance assigned directly to the projections without taking negative-log. Referring to Bayesian inferences, the decomposition fractions and observation variance are estimated by using the joint Maximum A Posteriori (MAP) estimation method. Subject to an adaptive prior model assigned to the variance, the joint estimation problem is then simplified into a single estimation problem. It transforms the joint MAP estimation problem into a minimization problem with a non-quadratic cost function. To solve it, the use of a monotone Conjugate Gradient (CG) algorithm with suboptimal descent steps is proposed.The performances of the proposed approach are analyzed with both simulated and experimental data. The results show that the proposed Bayesian approach is robust to noise and materials. It is also necessary to have the accurate spectrum information about the source-detector system. When dealing with experimental data, the spectrum can be predicted by a Monte Carlo simulator. For a variety of materials, less than 5% estimation errors are observed on their average decomposition fractions.The proposed approach is a statistical reconstruction approach based on a non-linear forward model counting the full beam polychromaticity and applied directly to the projections without taking negative-log. Compared to the approaches based on linear forward models and the BHA correction approaches, it has advantages in noise robustness and reconstruction accuracy.

Les systèmes d’imagerie par rayons X conventionnels utilisent des détecteurs à base de scintillateur en mode intégration d’énergie. La nouvelle génération de détecteurs à base de semi-conducteur CdTe/CdZnTe permet de compter le nombre de photons et de mesurer l’énergie avec laquelle les photons arrivent sur le détecteur. Le laboratoire LDET (CEA LETI) a développé des détecteurs spectrométriques pixellisés à base de CdTe pour l’imagerie par rayons X associés à un circuit de lecture rapide permettant de travailler à fort taux de comptage avec une bonne résolution en énergie. Ces travaux de thèse proposent d’apporter une contribution au traitement des données acquises sur ces détecteurs résolus en énergie pour la quantification des constituants des matériaux en radiographie et en tomographie. Le cadre médical applicatif choisi est l’ostéodensitométrie. Des simulations de radiographie, qui prennent en compte les imperfections du système de détection, comme le partage de charges et les empilements, ont été réalisées. Nous avons choisi d’étudier des méthodes de traitements des données spectrales basées sur la décomposition en base de matériaux. Cette technique de réduction des données consiste à modéliser le coefficient d’atténuation linéique d’un matériau par une combinaison linéaire des fonctions d’atténuation de deux matériaux de base. Deux approches, utilisant toutes les deux un apprentissage par calibrage, ont été adaptées pour notre application. La première est une adaptation de l’approche polynômiale standard, appliquée pour deux et trois canaux d’énergie. Un processus d’optimisation des seuils des canaux a été réalisé afin de trouver la configuration optimale des bandes d’énergie. Une étude sur le nombre de canaux a permis d’évaluer les limites de la formulation polynômiale. Pour aller plus loin dans l’exploitation du potentiel des nouveaux détecteurs, une approche statistique développée dans notre laboratoire a été adaptée pour la décomposition en base de matériaux. Elle peut se généraliser à un grand nombre de canaux (100 par exemple). Une comparaison des deux approches a été réalisée selon des critères de performance comme le bruit et la précision sur l’estimation des longueurs des matériaux traversés. La validation des deux approches étudiées sur des données expérimentales acquises en radiographie, dans notre laboratoire, avec des détecteurs spectrométriques, a montré une bonne quantification des constituants des matériaux, en accord avec les résultats obtenus en simulation. / Scintillator based integrating detectors are used in conventional X-ray imaging systems. The new generation of energy-resolved semiconductor radiation detectors, based on CdTe/CdZnTe, allows counting the number of photons incident on the detector and measure their energy. The LDET laboratory developed pixelated spectrometric detectors for X-ray imaging, associated with a fast readout circuit, which allows working with high fluxes and while maintaining a good energy resolution. With this thesis, we bring our contribution to data processing acquired in radiographic and tomographic modes for material components quantification. Osteodensitometry was chosen as a medical application. Radiographic data was acquired by simulation with a detector which presents imperfections as charge sharing and pile-up. The methods chosen for data processing are based on a material decomposition approach. Basis material decomposition models the linear attenuation coefficient of a material as a linear combination of the attenuations of two basis materials based on the energy related information acquired in each energy bin. Two approaches based on a calibration step were adapted for our application. The first is the polynomial approach used for standard dual energy acquisitions, which was applied for two and three energies acquired with the energy-resolved detector. We searched the optimal configuration of bins. We evaluated the limits of the polynomial approach with a study on the number of channels. To go further and take benefit of the elevated number of bins acquired with the detectors developed in our laboratory, a statistical approach implemented in our laboratory was adapted for the material decomposition method for quantifying mineral content in bone. The two approaches were compared using figures of merit as bias and noise over the lengths of the materials traversed by X-rays. An experimental radiographic validation of the two approaches was done in our laboratory with a spectrometric detector. Results in material quantification reflect an agreement with the simulations.

Ring artifacts are a common problem with the use of photon-counting detectors and commercial deployment rests on being able to compensate for them. Deep learning has been proposed as a candidate for tackling the inefficiency or high cost of traditional techniques. In that spirit, we propose a new approach to ring artifact reduction, namely one that employs Residual Networks in sinogram domain. We train them on data simulated via a realistic photon-counting CT model based on numerical phantoms of real scans acquired by the KiTS19 Challenge dataset. By exploring various architectures we find that shallow ResNets achieve a significant artifact reduction by staying more true to the ground truth in terms of not introducing new artifacts. All networks introduce a smoothing effect which is attributed to the use of MSE as a loss function. An alternative training scheme using patches instead of whole sinograms is tested and it shows a slightly improved model stability. Lastly, we demonstrate via a performance metric study that common metrics are not suitable for quantifying the performance in this problem, save for a potential new approach in the virtual mono-energetic domain. / Ringartefakter är ett vanligt problem vid användning av fotonräknande detektorer och kommersiell introduktion kräver att man kan kompensera för dem. Djupinlärning har föreslagits som en kandidat för att hantera ineffektiviteten eller de höga kostnaderna för traditionella tekniker. I den andan föreslår vi ett nytt tillvägagångssätt för att reducera ringartefakter, nämligen en som använder sig av residualnätverk i sinogramdomänen. Vi tränar dem på data simulerad via en realistisk fotonräkning CT modell baserad på numeriska fantomer av verkliga skanningar från datamängen KiTS19 Challenge. Genom att utforska olika arkitekturer finner vi att grunda ResNet uppnår en betydande minskning av artefakter genom bevara en större likhet med den sanna bilden när det gäller att inte introducera nya artefakter. Alla nätverk introducerar en utsmetningseffekt som tillskrivs användningen av MSE som en förlustfunktion. Ett alternativt träningsschema med utsnitt istället för hela sinogram testas och det visar en något förbättrad modellstabilitet. Slutligen visar vi genom en prestandamåttstudie att vanliga prestandamått inte är lämpliga för att kvantifiera prestandan i detta problem med undantag för ett potentiellt nytt tillvägagångssätt i den virtuella monoenergetiska domänen.