Search

The detection of tumoural cells from whole slide images is an essential task in medical diagnosis and research. In this article, we propose and analyse a novel approach that combines computer vision-based models with graph neural networks to improve the accuracy of automated tumoural cell detection in lung tissue. Our proposal leverages the inherent structure and relationships between cells in the tissue. Experimental results on our own curated dataset show that modelling the problem with graphs gives the model a clear advantage over just working at pixel level. This change in perspective provides extra information that makes it possible to improve the performance. The reduction of dimensionality that comes from working with the graph also allows us to increase the field of view with low computational requirements. Code is available at https://github.com/Jerry-Master/lung-tumour-study, models are uploaded to https://huggingface.co/Jerry-Master/Hovernet-plus-Graphs, and the dataset is published on Zenodo https://zenodo.org/doi/10.5281/zenodo.8368122., Peer Reviewed, Postprint (published version)

Purpose: The diagnosis and prognosis of breast cancer relies on histopathology image analysis. In this context, proliferation markers, especially Ki67, are increasingly important. The diagnosis using these markers is based on the quantification of proliferation, which implies the counting of Ki67 positive and negative tumoral cells in epithelial regions, thus excluding stromal cells. However, stromal cells are often very difficult to distinguish from negative tumoral cells in Ki67 images and often lead to errors when automatic analysis is used. Approach: We study the use of automatic semantic segmentation based on convolutional neural networks (CNNs) to separate stromal and epithelial areas on Ki67 stained images. CNNs need to be accurately trained with extensive databases with associated ground truth. As such databases are not publicly available, we propose a method to produce them with minimal manual labelling effort. Inspired by the procedure used by pathologists, we have produced the database relying on knowledge transfer from cytokeratin-19 images to Ki67 using an image-to-image (I2I) translation network. Results: The automatically produced stroma masks are manually corrected and used to train a CNN that predicts very accurate stroma masks for unseen Ki67 images. An F-score value of 0.87 is achieved. Examples of effect on the KI67 score show the importance of the stroma segmentation. Conclusions: An I2I translation method has proved very useful for building ground-truth labeling in a task where manual labeling is unfeasible. With reduced correction effort, a dataset can be built to train neural networks for the difficult problem of separating epithelial regions from stroma in stained images where separation is very hard without additional information., This work was supported by the Spanish Research Agency (AEI) (Project No. PID2020-116907RB-I00) of the call MCIN/AEI/10.13039/50110001103 and the European Regional Development Funds under program FEDER Catalunya 2014–2020 and SA18-014623 DIGIPATICS., Peer Reviewed, Postprint (published version)

Finding new ways to cost-effectively facilitate population screening and improve cancer diagnoses at an early stage supported by data-driven AI models provides unprecedented opportunities to reduce cancer related mortality. This work presents the INCISIVE project initiative towards enhancing AI solutions for health imaging by unifying, harmonizing, and securely sharing scattered cancer-related data to ensure large datasets which are critically needed to develop and evaluate trustworthy AI models. The adopted solutions of the INCISIVE project have been outlined in terms of data collection, harmonization, data sharing, and federated data storage in compliance with legal, ethical, and FAIR principles. Experiences and examples feature breast cancer data integration and mammography collection, indicating the current progress, challenges, and future directions., This research received funding mainly from the European Union’s Horizon 2020 research and innovation program under grant agreement no 952179. It was also partially funded by the Ministry of Economy, Industry, and Competitiveness of Spain under contracts PID2019-107255GB and 2017-SGR-1414., Peer Reviewed, Article signat per 30 autors/es: Ivan Lazic (1), Ferran Agullo (2), Susanna Ausso (3), Bruno Alves (4), Caroline Barelle (4), Josep Ll. Berral (2), Paschalis Bizopoulos (5), Oana Bunduc (6), Ioanna Chouvarda (7), Didier Dominguez (3), Dimitrios Filos (7), Alberto Gutierrez-Torre (2), Iman Hesso (8), Nikša Jakovljević (1), Reem Kayyali (8), Magdalena Kogut-Czarkowska (9), Alexandra Kosvyra (7), Antonios Lalas (5) , Maria Lavdaniti (10,11), Tatjana Loncar-Turukalo (1),Sara Martinez-Alabart (3), Nassos Michas (4,12), Shereen Nabhani-Gebara (8), Andreas Raptopoulos (6), Yiannis Roussakis (13), Evangelia Stalika (7,11), Chrysostomos Symvoulidis (6,14), Olga Tsave (7), Konstantinos Votis (5) Andreas Charalambous (15) / (1) Faculty of Technical Sciences, University of Novi Sad, 21000 Novi Sad, Serbia; (2) Barcelona Supercomputing Center, 08034 Barcelona, Spain; (3) Fundació TIC Salut Social, Ministry of Health of Catalonia, 08005 Barcelona, Spain; (4) European Dynamics, 1466 Luxembourg, Luxembourg; (5) Centre for Research and Technology Hellas, 57001 Thessaloniki, Greece; (6) Telesto IoT Solutions, London N7 7PX, UK: (7) School of Medicine, Faculty of Health Sciences, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece; (8) Department of Pharmacy, Kingston University London, London KT1 2EE, UK; (9) Timelex BV/SRL, 1000 Brussels, Belgium; (10) Nursing Department, International Hellenic University, 57400 Thessaloniki, Greece; (11) Hellenic Cancer Society, 11521 Athens, Greece; (12) European Dynamics, 15124 Athens, Greece; (13) German Oncology Center, Department of Medical Physics, Limassol 4108, Cyprus; (14) Department of Digital Systems, University of Piraeus, 18534 Piraeus, Greece; (15) Department of Nursing, Cyprus University of Technology, Limassol 3036, Cyprus, Postprint (published version)

Complete digital pathology transformation for primary histopathological diagnosis is a challenging yet rewarding endeavor. Its advantages are clear with more efficient workflows, but there are many technical and functional difficulties to be faced. The Catalan Health Institute (ICS) has started its DigiPatICS project, aiming to deploy digital pathology in an integrative, holistic, and comprehensive way within a network of 8 hospitals, over 168 pathologists, and over 1 million slides each year. We describe the bidding process and the careful planning that was required, followed by swift implementation in stages. The purpose of the DigiPatICS project is to increase patient safety and quality of care, improving diagnosis and the efficiency of processes in the pathological anatomy departments of the ICS through process improvement, digital pathology, and artificial intelligence tools., This project was funded by European Regional Development Funds, Programa operatiu FEDER de Catalunya 2014–2020 and SA18-014623 DIGIPATICS. UPC activity in this project was partially supported by PID2020-116907RB-I00 and funded by MCIN/AEI/10.13039/501100011033, Peer Reviewed, Article signat per 18 autors/es: Jordi Temprana-Salvador (1), Pablo López-García (2), Josep Castellví Vives (1),Lluís de Haro (2), Eudald Ballesta (2), Matias Rojas Abusleme (3), Miquel Arrufat (4), Ferran Marques (5), Josep R. Casas (5),Carlos Gallego (6), Laura Pons (7), José Luis Mate (7), Pedro Luis Fernández (7), Eugeni López-Bonet (8), Ramon Bosch (9), Salomé Martínez (10), Santiago Ramón y Cajal (1), and Xavier Matias-Guiu (11,12) // (1) Department of Pathology, Vall d’Hebron University Hospital, CIBERONC, 08035 Barcelona, Spain; (2) Functional Competence Center, Information Systems, Catalan Health Institute (Institut Català de la Salut), 08006 Barcelona, Spain; (3) Center for Telecommunications and Information Technology (Centre de Telecomunicacions i Tecnologies de la Informació, CTTI), Catalan Health Institute (Institut Català de la Salut), 08006 Barcelona, Spain; (4) Economic and Financial Management, Catalan Health Institute (Institut Català de la Salut), 08006 Barcelona, Spain; (5) Image Processing Group, Technical University of Catalonia (UPC), 08034 Barcelona, Spain; (6) Digital Medical Imaging System of Catalonia (SIMDCAT), TIC Salut, 08005 Barcelona, Spain, (7) Department of Pathology, Germans Trias i Pujol University Hospital, 08916 Badalona, Spain; (8) Department of Pathology, Doctor Josep Trueta Hospital of Girona, 17007 Girona, Spain; (9) Department of Pathology, Verge de la Cinta Hospital of Tortosa, 43500 Tarragona, Spain; (10) Department of Pathology, Joan XXIII University Hospital of Tarragona, 43005 Tarragona, Spain; (11) Department of Pathology, Arnau de Vilanova University Hospital, 25198 Lleida, Spain, (12) Department of Pathology, Bellvitge University Hospital, CIBERONC, 08907 Barcelona, Spain, Postprint (published version)

A system to integrate microwave imaging with optical colonoscopy is presented. The overarching goal is to improve the prevention and early diagnosis of one of the main health and economic burdens of an increasingly aging population, i.e., colorectal cancer. For a colonoscopy, the gold standard for colorectal cancer diagnosis, 22% of polyps are not detected, and the risk of cancer after a negative colonoscopy can be up to 7.9%. To remedy this, a microwave imaging system able to generate an alarm when a polyp is detected is designed, manufactured and validated with a colon phantom composed of tissue-mimicking oil-gelatin materials reproducing the anatomy and dielectric properties of a human colon with a polyp. The acquisition was performed by a miniaturized ring-shaped switched array of 16 antennas attachable at the tip of a conventional colonoscope. This has been conceived to satisfy endoscopy size restrictions, patient safety and intercompatibility with current clinical practice. A Modified Monofocusing imaging method preceded by a previous frame average subtraction as a calibration technique shows a perfect detection of a 10-mm polyp (100% sensitivity and specificity) in the eight analyzed trajectories. The phantom results demonstrate the feasibility of the system in future preclinical trials., The work of Alejandra Garrido was supported by DIN2019-010857. The work of Roberto Sont, Ignasi Belda, and Marta Guardiola was supported in part by the European Union’s Horizon 2020 Research and Innovation Programme under Grant 960251 and in part by the European Institute of Innovation and Technology (EIT). The work of Jordi Romeu was supported by PID2019-107885GB-C31/AEI/10.13039/501100011033., Peer Reviewed, Postprint (published version)

Education in eHealth is becoming of transcendental importance in the European social context. The progressive aging of the population has generated an increase in the prevalence of people with disabilities who demand more effective, adapted and flexible services. Advances in computer science and technology are enabling a qualitative leap in the care of these people, offering solutions to health demands. However, the incorporation of technology in the field of health requires the acquisition of new skills. The eHealth Eurocampus is an EU-funded Erasmus+ Strategic Partnership, whose primary aim is to support the development of innovative teaching materials and activities that will improve the relevance and quality of higher education in the field of eHealth. A total of 10 European institutions participate in the project, including universities, hospitals and technology companies. We present the results corresponding to the third edition of the summer school that took place in Barcelona during the month of July 2019. On this occasion, the main topic was the use of robotics and medical imaging for the treatment of physical and cognitive disability. A teaching methodology based on the presentation of challenges was used and for which the different groups had to work and give a technological solution. The challenges were proposed by real patients and healthcare professionals who, from their experience, presented a problem of their daily life: they went from issues related to the mobility of people with physical disabilities to the usability of medical devices in hospital settings. The students were organized in multi-disciplinary and multicultural teams, and they had to develop their prototypes to respond to the proposed challenge. The resulting projects were presented publicly to the patients and professionals themselves for further discussion. The results presented refer to the perceptions of participating students about the effectiveness of this methodology in the trai, Peer Reviewed, Postprint (author's final draft)

© 2019 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes,creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., Human lungs are essential respiratory organs. Different Obstructive Lung Diseases (OLD) such as bronchitis, asthma, lungs cancer etc. affects the respiration. Diagnosing OLD in the initial stage is better than diagnosing and curing them later. The delay in diagnosing OLD is due to expensive diagnosing tool and experts requirement. Therefore, a non-invasive diagnosing tool for OLD is required that identifies dysfunctional lungs without the support of expert, complex and expensive diagnosing types of equipment. In this work, we design an Iris based Lungs Pre-diagnostic System (ILPS). The ILPS takes iris images as input and identifies dysfunctional Lungs based on iridology map. While testing with 50 lungs patients, the results confirm that the ILPS identifies dysfunctional lungs patients with the accuracy of 88%., The research leading to these results has received funding from the Higher Education Commission under NRPU 2017/18., Peer Reviewed, Postprint (author's final draft)

A fully automatic technique for segmenting the liver and localizing its unhealthy tissues is a convenient tool in order to diagnose hepatic diseases and assess the response to the according treatments. In this work we propose a method to segment the liver and its lesions from Computed Tomography (CT) scans using Convolutional Neural Networks (CNNs), that have proven good results in a variety of computer vision tasks, including medical imaging. The network that segments the lesions consists of a cascaded architecture, which first focuses on the region of the liver in order to segment the lesions on it. Moreover, we train a detector to localize the lesions, and mask the results of the segmentation network with the positive detections. The segmentation architecture is based on DRIU, a Fully Convolutional Network (FCN) with side outputs that work on feature maps of different resolutions, to finally benefit from the multi-scale information learned by different stages of the network. The main contribution of this work is the use of a detector to localize the lesions, which we show to be beneficial to remove false positives triggered by the segmentation network., Postprint (published version)

We propose a novel Active Learning framework capable to train effectively a convolutional neural network for semantic segmentation of medical imaging, with a limited amount of training labeled data. Our contribution is a practical Cost-Effective Active Learning approach using dropout at test time as Monte Carlo sampling to model the pixel-wise uncertainty and to analyze the image information to improve the training performance., Postprint (published version)

The analysis of the morphology and content of the gut is necessary in order to understand metabolic and functional gut activity and for diagnostic purposes. Magnetic resonance imaging (MRI) has become an important modality technique since it is able to visualize soft tissues using no ionizing radiation, and hence removes the need for any contrast agents. In the last few years, MRI of gastrointestinal function has advanced substantially, although scarcely any publication has been devoted to the analysis of the colon content. This paper presents a semi-automatic segmentation tool for the quantitative assessment of the unprepared colon from MRI images. This application has allowed for the analysis of the colon content in various clinical experiments. The results of the assessment have contributed to a better understanding of the functionality of the colon under different diet conditions. The last experiment carried out by medical doctors showed a marked influence of diet on colonic content, accounting for about 30% of the volume variations., Peer Reviewed, Postprint (published version)

The recent emergence of deep learning methods for medical image analysis has enabled the development of intelligent medical imaging-based diagnosis systems that can assist the human expert in making better decisions about a patient’s health. In this paper we focus on the problem of skin lesion classification, particularly early melanoma detection, and present a deep-learning based approach to solve the problem of classifying a dermoscopic image containing a skin lesion as malignant or benign. The proposed solution is built around the VGGNet convolutional neural network architecture and uses the transfer learning paradigm. Experimental results are encouraging: on the ISIC Archive dataset, the proposed method achieves a sensitivity value of 78.66%, which is significantly higher than the current state of the art on that dataset., Postprint (author's final draft)

Postprint (published version)

En la última década la técnica PET-CT con 18F-FDG, ha demostrado ser de gran utilidad en el diagnóstico, extensión y control de diversas patologías inflamatorias e infecciosas. El territorio miocárdico, no obstante, sigue presentando un reto a la hora de valorar dichas patologías, dada su variabilidad a la hora de presentar o no, captación fisiológica de 18F-FDG. Las metodologías prínceps para intentar frenar el metabolismo cardiaco fisiológico, son el ayuno prolongado, la dieta pobre en hidratos de carbono y rica en ácidos grasos, y la administración endovenosa de heparina minutos antes de la inyección de 18F-FDG, siempre y cuando no existan contraindicaciones. Existen muchos pacientes que por sus antecedentes patológicos y por sus características individuales o sociales no pueden realizar ayunos prolongados o cumplir estrictamente dietas preestablecidas previas a la exploración tomográfica. Como último recurso muchos de estos pacientes únicamente disponen de la metodología heparinizante para intentar optimizar la captación miocárdica y a pesar de este hecho, la heparina endovenosa no acaba siendo efectiva en todos los pacientes. La línea de investigación desarrollada en esta tesis doctoral responde a la necesidad de poder prever la efectividad del fármaco heparinizante, y poder determinar a priori, en que sujetos, dicho tratamiento podría ser óptimo dados sus antecedentes patológicos, uso de fármacos o características personales. Por consiguiente, se podría evitar la administración de un fármaco en aquellos sujetos que previsiblemente no va a ser efectivo, o indicarla claramente al grupo efectivo. Se evaluaron 479 sujetos, (230 mujeres) con una media de edad de 65 años (22-86) con indicación de realizarse un PET-CT secundario a su enfermedad de base (proceso oncológico o sospecha de proceso infeccioso o inflamatorio en territorio cardiaco o vascular). Se estudiaron 5 grupos de sujetos: sujetos con diabetes mellitus, sujetos con dislipemia en tratamiento crónico hi, 18F-FDG PET/CT is a valuable diagnostic tool in the evaluation of inflammatory and infection diseases. Appropriate patient preparation is important because the diagnostic accuracy of this procedure depends on an adequate suppression of physiologic glucose uptake in the myocardium. There are different methods to reduce myocardial 18F-FDG uptake, such as long fasting period, high fat and low-carbohydrate diet and the administration of heparin some minutes before the 18F-FDG dose injection only if there is no counter-indication Due to their pathological background or their individual characteristics, many patients are unable to do fast for many hours or to follow high fat and low-carbohydrate diet before the scan. The administration of heparin some minutes before the 18F-FDG dose injection is the only method to try to do an adequate suppression of physiologic glucose uptake in the myocardium. However, not always the heparin's method works properly doing a suppression of physiologic glucose uptake in the myocardium. The aim of this project is to try to know when the heparin is going to be effective, depending on the pathological background of patients, chronic treatment and individual characteristics. Therefore, if it is known in which subjects is not going to be effective we could avoid treating them. 479 subjects were evaluated (230 women), mean age 65 years old (22-86) with indication of 18F-FDG PET/CT due to an oncological illness, or infection or inflammatory suspect of a cardiac or vascular process. Five groups of subjects were studied: Subjects with diabetes mellitus, subjects with dyslipidemia under chronic lipid lowering therapy, subjects with dyslipidemia but without chronic treatment for dyslipidemia, subjects with metabolic syndrome (diabetes and dyslipidemia coexistence) and the last group of normal subjects, without diabetes nor dyslipidemia. We compare two methods. One is based on at least 6 hours of fasting and the second one is based on at least 6 hours

Alterations in the health of muscles can be evaluated through the use of electrical impedance myography (EIM). To date, however, nearly all work in this field has relied upon the measurement of muscle at rest. To provide an insight into the contractile mechanisms of healthy and disease muscle, we evaluated the alterations in the spectroscopic impedance behavior of muscle during the active process of muscle contraction. The gastrocnemii from a total of 13 mice were studied (five wild type, four muscular dystrophy animals, and four amyotrophic lateral sclerosis animals). Muscle contraction was induced via monophasic current pulse stimulation of the sciatic nerve. Simultaneously, multisine EIM (1 kHz to 1 MHz) and force measurements of the muscle were performed. Stimulation was applied at three different rates to produce mild, moderate, and strong contractions. We identified changes in both single and multifrequency data, as assessed by the Cole impedance model parameters. The processes of contraction and relaxation were clearly identified in the impedance spectra and quantified via derivative plots. Reductions in the center frequency f(c) were observed during the contraction consistent with the increasing muscle fiber diameter. Different EIM stimulation rate-dependencies were also detected across the three groups of animals., Peer Reviewed, Postprint (author's final draft)

There is an increasing interest on tele-medicine and tele-diagnostic solutions based on the remote inspection of volume data coming from multimodal imaging. Client-server architectures meet these functionalities. The use of mobile devices is sometimes required due to the portability and easy maintenance. However, transmission time for the volumetric information and low performance hardware properties, make quite complex the design of efficient visualization systems on these devices. In this paper, we present a hybrid approach which is based on regions of interest (ROIs) and on a transfer-function aware compression scheme. It has a good performance in terms of bandwidth requirements and storage needs in the client device, being flexible enough to represent several materials and volume structures in the ROI. Clients store a low-resolution version of the volume data and ROI-dependent high resolution segmented information. Data must be only sent whenever a new ROI is requested, but interaction in the client is autonomous - without any data transmission - while a certain ROI is inspected. A benchmark is presented to compare the proposed scheme with three existing approaches, on two different volume data models. The results show that our hybrid approach is compact, efficient and scalable, with compression rates that decrease when the size of the volume model increases., Peer Reviewed, Postprint (published version)

In computer-assisted beating heart surgeries, accurate tracking of the heart’s motion is of huge importance and there is a continuous need to eliminate any source of error that might disturb the tracking process. One source of error is the specular reflection that appears on the glossy surface of the heart. In this paper, we propose a robust solution for the detection and removal of specular highlights. A hybrid color attributes and wavelet based edge projection approach is applied to accurately identify the affected regions. These regions are then recovered using a dynamic search-based inpainting with adaptive windowing. Experimental results demonstrate the precision and efficiency of the proposed method. Moreover, it has a real-time performance and can be generalized to various other applications., Peer Reviewed, Postprint (published version)

Retinal images often suffer from blurring which hinders disease diagnosis and progression assessment. The restoration of the images is carried out by means of blind deconvolution, but the success of the restoration depends on the correct estimation of the point-spread-function (PSF) that blurred the image. The restoration can be space-invariant or space-variant. Because a retinal image has regions without texture or sharp edges, the blind PSF estimation may fail. In this paper we propose a strategy for the correct assessment of PSF estimation in retinal images for restoration by means of space-invariant or space-invariant blind deconvolution. Our method is based on a decomposition in Zernike coefficients of the estimated PSFs to identify valid PSFs. This significantly improves the quality of the image restoration revealed by the increased visibility of small details like small blood vessels and by the lack of restoration artifacts., Postprint (published version)

L’avaluació de la qualitat òptica ocular és una eina útil que proporciona informació objectiva de la visió. Gràcies als nous avenços tecnològics, la tècnica de doble pas pot enregistrar en un entorn clínic la imatge retinal i fer-ne el seu anàlisi, proporcionant informació de les aberracions i la difusió intraocular. En aquest treball s’utilitza un sistema experimental basat en aquesta tècnica amb una càmera d’alta sensibilitat, per avaluar el grau de difusió d’ulls sans i amb diferent grau de cataracta. Per disminuir la influència de les aberracions sobre la difusió, s’estudia un nou paràmetre anomenat PSI (Peripheral Scatter Index) que utilitza regions més allunyades del centre de la imatge en comparació a les que s’usen actualment en el sistema comercial OQASTM (Optical Quality Analysis System) per al càlcul de l’OSI (Objective Scatter Index) (de 12’ a 20’). Els resultats assenyalen que l’OSI proporciona una bona informació de la difusió, tot i que el PSI permetria una millor quantificació de la mateixa, sobretot en ulls que en mostren amb elevada quantitat. Les dades mostren que les excentricitats on millor es valora la difusió es localitzen en zones diferents en cada individu, tot i que de forma genèrica la millor zona estaria entre 60’ i 124’., La evaluación de la calidad óptica ocular es una herramienta útil que proporciona información objetiva de la visión. Gracias a los nuevos avances tecnológicos, la técnica de doble paso permite registrar en un entorno clínico la imagen retiniana y hacer su análisis, proporcionando información de las aberraciones y la difusión intraocular. En éste trabajo se utiliza un sistema experimental basado en ésta técnica con una cámara de alta sensibilidad con el fin de evaluar el grado de difusión en ojos sanos y con diferente grado de catarata. Para disminuir la influencia de las aberraciones sobre la difusión, se estudia un nuevo parámetro denominado PSI (Pheripheral Scatter Index) que utiliza regiones más lejanas del centro de la imagen en comparación a las que se usan actualmente en el sistema comercial OQASTM (Optical Quality Analysis System) para el cálculo del OSI (Objective Scatter Index) (de 12’ a 20’). Los resultados señalan que el OSI proporciona una buena información de la difusión pero que el PSI permite una mejor cuantificación de la misma, sobretodo en ojos con alto grado de difusión. Los datos muestran que las excentricidades donde se valora mejor la difusión se localizan en zonas diferentes según el paciente, aunque de forma genérica la mejor zona se encuentra entre 60’ i 124’., The assessment of the optical quality is a useful tool that provides objective information of vision. By means of the double pass technique, it is now possible to record the double pass image and its posterior analysis in a clinical environment, which brings information about aberrations and intraocular scattering. In this work, we used an experimental double pass system which includes a high sensitivity refrigerated camera in order to evaluate the intraocular scattering from an assembly of patients with healthy eyes and with different grades of nuclear cataracts. A new index called PSI (Peripheral Scatter Index) was studied, which takes into account information about outer areas of the retinal image in comparison with those used by the OSI (Objective Scatter Index) given by the commercial instrument OQASTM (Optical Quality Analysis System) (from 12’ to 20’), in order to decrease the impact of aberrations. The results reveal that OSI gives good information about scattering, although outer regions of the retinal image could provide additional information for improving the assessment of intraocular scattered light, mainly in eyes with high level of diffusion. On the other hand, even though each patient exhibits a specific region for the optimal evaluation of the scattering, the zone comprised between 60 and 124 minutes of arc seems to be suitable for its assessment.

El càncer de còlon és la quarta causa més comuna de mort per càncer en el món, havent al voltant de 143.460 nous casos el 2012, segons dades de l'American Cancer Society. La taxa de mortalitat d'aquest càncer depèn de l'etapa en què sigui detectat, decreixent des de taxes majors del 95% en les primeres etapes a taxes inferiors al 35% en les etapes quarta i cinquena, la qual cosa demostra la necessitat d'una exploració precoç del còlon. Durant l'exploració, el metge busca creixements adenomatosos, que es coneixen sota el nom de pòlips, per tal d'esbrinar el seu nivell de desenvolupament. Hi ha diverses tècniques per a l'exloració del còlon, però la colonoscòpia és considerada encara avui com estàndard de facto, encara que presenta alguns inconvenients com una elevada taxa de errors. La nostra contribució està enquadrada dins del camp de sistemes intel·ligents per a la colonoscòpia [13] i té com a objectiu el desenvolupament de mètodes de localització i segmentació de pòlips basant-se en models d'aparen¸ca. Definim la localització de pòlips com el mètode pel qual donada una imatge d'entrada es proporciona com a sortida una altra imatge on es assenyalen les àrees de la imatge on és més probable trobar un pòlip. Per altra banda, la segmentació té com a objectiu definir la regió concreta de la imatge on es troba el pòlip. Per tal de desenvolupar ambdós mètodes s'ha comen¸cat desenvolupant un model d'aparen¸ca per a pòlips, el qual defineix un pòlip com una estructura limitada per valls en la imatge de escala de gris. La novetat de la nostra contribució rau en el fet que incloem en el nostre model altres elements de l'escena endoluminal, com els reflexos especulars o els gots sanguinis, que demostren tenir un impacte en el rendiment global dels nostres mètodes i en elements derivats del procés de formació de la imatge com l'interlacing. Tenint això en compte, definim una imatge de profunditat de valls que integra la sortida del detector de valls amb un gradient morfològic, El cáncer de colon es la cuarta causa más común de muerte por cáncer en el mundo, presentando alrededor de 143.460 nuevos casos en 2012, según datos estimados por la American Cancer Society. La tasa de mortalidad del cancer de colon depende de la etapa en que éste es detectado, decreciendo desde tasas mayores del 95% en las primeras etapas a tasas inferiores al 35% en las etapas cuarta y quinta, lo cual es muestra de la necesidad de una exploración temprana del colon. Durante la exploración el médico busca crecimientos adenomatosos que se conocen bajo el nombre de pólipos, con el fin de averiguar su nivel de desarrollo. Existen varias técnicas para la exloración del colon pero la colonoscopia está considerada aún hoy en d'ıa como estandar de facto, aunque presenta algunos inconvenientes como la tasa de fallos. Nuestra contribución, encuadrada dentro del campo de sistemas inteligentes para la colonoscopia, tiene como objetivo el desarrollo de métodos de localización y segmentación de pólipos basándose en un model de apariencia para los pólipos. Definimos localización de pólipos como el método por el cual dada una imagen de entrada se proporciona como salida donde se señalan las áreas de la imagen donde es más probable encontrar un pólipo. La segmentación de pólipos tiene como objetivo definir la región concreta de la imagen donde está el pólipo. Con el fin de desarrollar ambos métodos se ha comenzado desarrollando un modelo de apariencia para pólipos, el cual define un pólipo como una estructura limitada por valles en la imagen de escala de gris. Lo novedoso de nuestra contribución radica en el hecho de que incluimos en nuestro modelo otros elementos de la escena endoluminal tales como los reflejos especulares o los vasos sangu'ıneos que tienen un impacto en el rendimiento global de nuestros métodos as'ı como elementos derivados del proceso de formación de la imagen, como el interlacing. Teniendo esto en cuenta definimos nuestra imagen de profundidad de valles que in, Colorectal cancer is the fourth most common cause of cancer death worldwide, with about 143.460 new cases expected in 2012 by recent estimates of the American Cancer Society. Colon cancerís survival rate depends on the stage in which it is detected, decreasing from rates higher than 95% in the first stages to rates lower than 35% in stages IV and V, hence the necessity for a early colon screening. In this process physicians search for adenomatous growths known as polyps, in order to assess their degree of development. There are several screening techniques but colonoscopy is still nowadays the gold standard, although it has some drawbacks such as the miss rate. Our contribution, in the field of intelligent system for colonoscopy, aims at providing a polyp localization and a polyp segmentation system based on a model of appearance for polyps. In this sense we define polyp localization as a method which given an input image identifies which areas of the image are more likely to contain a polyp. Polyp segmentation aims at selecting the region of the image that contains a polyp. In order to develop both methods we have started by defining a model of appearance for polyps, which defines a polyp as enclosed by intensity valleys. The novelty of our contribution resides on the fact that we include in our model other elements from the endoluminal scene such as specular highlights and blood vessels, which have an impact on the performance of our methods and also other elements that appear as a result of image formation, such as interlacing. Considering this we define our novel Depth of Valleys image which integrates valley information with the output of the morphological gradient and also takes into account the presence of the before mentioned elements of the endoluminal scene. In order to develop our polyp localization method we accumulate the information that the Depth of Valleys image provides in order to generate accumulation energy maps. In order to obtain polyp segmenta

Las técnicas de diagnóstico por imagen han revolucionado muchas especialidades, como la oncología, donde su uso ha permitido reducir la morbi-mortalidad de lesiones detectadas precozmente. Sin embargo, ni todas las técnicas ni todos los tumores son iguales.

Postprint (author's final draft)

El presente proyecto busca estudiar la viabilidad de la implantación de un servicio de prototipado rápido para el ámbito médico en la Fundació CIM. La Fundació CIM posee en sus instalaciones diversas tecnologías de prototipado rápido así como experiencia en su uso por lo que en este proyecto se deberá desarrollar los métodos y adquirir los conocimientos necesarios para poder aplicar estas tecnologías al mundo de la medicina. El punto de partida es la generación de modelos tridimensionales a partir de imágenes médicas tales como resonancias magnéticas (RM) o tomografías computerizadas (TC). Estos métodos captan imágenes de una parte del cuerpo humano capa a capa, permitiendo segmentar las partes interesantes para el caso y reproducir un modelo tridimensional de estas partes. Así pues, se hará una exploración de las tecnologías de fabricación aditiva susceptibles de ser utilizadas en este proyecto y de los métodos de captación de imágenes que se usaran como fuente de datos. También se analizaran diversos software de modelado 3D a partir de archivos DICOM, valorando diversos aspectos tales como dificultad de uso, calidad de resultados, coste, etc. Entre estos software hay que destacar uno, desarrollado en la UPC por el VirVIG, que se ha usado para las diferentes pruebas a lo largo del proyecto. También se hará una exploración del mercado, detectando posibles aplicaciones y analizando servicios similares existentes en España y en el extranjero. De esta exploración surgirá una relación de trabajo con una empresa intermediaria entre hospitales y fabricantes con lo que se podrá poner a prueba el modelo productivo que se está investigando, así como analizar la viabilidad económica y medioambiental. Otra parte importante del proyecto es la dedicada a aplicaciones a medida, donde se analizaran dos casos reales no convencionales realizados durante el proceso de investigación en colaboración con el hospital Sant Joan de Déu.

Projecte fi de carrera elaborat en col·laboració amb Medical Imaging Research Center de Illinois Institute of Technology, [ANGLÈS] This project is a first approach to an assisted diagnosis applied to chest x-ray images. Given a x-ray image, we pretend to develop an artificial observer that determines which areas of the image are relevant for a doctor during the diagnosis. We are using a dataset with several examples of chest x-ray images and a list of pixels gazed by expert human observers, doctors. This data is obtained by eye tracking techniques applied while the doctors were asked to diagnose the images. This research has developed a complete statistical learning algorithm that uses the given dataset and uses it for predicting regions of interest in the image., [CASTELLÀ] Este proyecto es una primera aproximación a la diagnosis asistida de radiografías de pecho. Dada una radiografía, se pretende desarrollar un observador artificial que determine en que zonas de la imagen se fija un médico al diagnosticarlas. Para ello se dispone de una base de datos con diversos ejemplos de radiografías y una lista de píxeles que han sido observados por observadores humanos expertos, médicos. Estos datos han sido captados mediante técnicas de trazado de mirada cuando a los médicos se les pedía que diagnosticasen las imágenes. Este trabajo de investigación pretende desarrollar un completo algoritmo de aprendizaje estadístico que utilice estos datos y los procese para poder predecir zonas de interés en radiografías de pecho., [CATALÀ] Aquest projecte és una primera aproximació a la diagnosi assistida de radiografies de pit. Donada una radiografia, es pretén desenvolupar un observador artificial que determini a quines zones de la imatge es fixa un metge al diagnosticar-les. Per a fer-ho, es disposa de una base de dades amb diversos exemples de radiografies i una llista dels píxels que han estat observats pels observadors humans experts, metges. Aquestes dades han estat captades mitjançant tècniques de traçat de mirada quan als metges se’ls demanava que diagnostiquessin les imatges. Aquest treball d’investigació pretén desenvolupar un complet algoritme d’aprenentatge estadístic que utilitzi aquestes dades i les processi per a poder predir zones d’interès a les radiografies.

Postprint (published version)

L’arribada de noves tecnologies obre moltes possibilitats pràctiques per estudiar més a fons l’ús d’aquestes imatges en l’òptica de l’ull. D’acord amb aquesta evolució, s’han emprat recentment durant la cirurgia de cataractes o en el cribratge de l’estrabisme. Tanmateix, una altra aplicació coneguda és el desenvolupament dels sistemes de seguiment ocular, que utilitzen la PI i la PIV. El propòsit principal d’aquest treball ha estat analitzar la simulació del traçat de raigs de les imatges de Purkinje en el sistema òptic de l’ull teòric. Com que aquestes sempre se solen captar amb instruments oftàlmics, s’ha volgut representar aquesta marxa gràfica per a la configuració objecte-ull en un instrument específic, el queratòmetre de Baush & Lomb, utilitzant el programa informàtic BEAM FOUR. Com es veurà, hi ha algunes discrepàncies entre el què diu l’òptica paraxial i els valors experimentals obtinguts.

Determining the changes in passive electrical properties of human tissues and the geometry changes of the body parts due to breathing and cardiac activity could be a method for monitoring these vital signs. We have designed a model with COMSOL Multiphysics to simulate a magnetic induction method for determining changes of these characteristics. Given the large number of parameters and variables, difficulties of anatomical modeling and various tissue properties, we are facing some numerical challenges. In addition, using magnetic induction method, we have to consider safety issues, standards and limitations. We used COMSOL to simulate the induced currents in the human body and the resulting perturbation in the magnetic field for different tissue properties and geometries. As a result, we obtained the expected changes during breathing and we know the minimum distances and maximum excitation currents that could be used to meet the international safety standards., Postprint (published version)

Web-based applications in computational medicine have become increasingly important during the last years. The rapid growth of the World Wide Web supposes a new paradigm in the telemedicine and eHealth areas in order to assist and enhance the prevention, diagnosis and treatment of patients. Furthermore, training of radiologists and management of medical databases are also becoming increasingly important issues in the field. In this paper, we present MammoApplet , an interactive Java applet interface designed as a web-based tool. It aims to facilitate the diagnosis of new mammographic cases by providing a set of image processing tools that allow a better visualization of the images, and a set of drawing tools, used to annotate the suspicious regions. Each annotation allows including the attributes considered by the experts when issuing the final diagnosis. The overall set of overlays is stored in a database as XML files associated with the original images. The final goal is to obtain a database of already diagnosed cases for training and enhancing the performance of novice radiologists, Peer Reviewed, Postprint (author's final draft)

English: Nowadays, the most popular technique for breast cancer detection is the mammography. Due to its drawbacks, other alternative/complementary methods are being researched. In this master thesis, the feasibility of the UWB Magnitude-Combined (MC-UWB) tomographic algorithm is assessed for the breast cancer detection application. This process is done by creating a numerical breast model, calculating its direct problem, applying the algorithm to solve the inverse problem to finally present the reconstructed images., Spanish: Actualmente, la técnica más usada para detectar cancer de mamas es la mamografía. Debido a que esta presenta algunos inconvenientes, otros métodos alternativos/complementarios están siendo investigados. En esta tesis de master, la viabilidad del algoritmo basado en tomografía llamado UWB Magnitude-Combined (MC-UWB) es evaluada. Este proceso es realizado en varias etapas las cuales constan en: crear un modelo numérico del seno, calcular su problema directo, aplicar el algoritmo para solucionar el problema inverso, y finalmente, presentar las imágenes reconstruídas., Català: Actualment, la tècnica mes utilitzada per detectar càncer de mama es la mamografia. Degut a aixo, es presenten alguns inconvenients, altres mètodes alternatius/complementaris estan siguen desarrollats. En aquesta tesis de màster, la viabilitat de l'algorisme basat en tomografia anomenat UWB Magnitude-Combined (MC-UWB) és avaluada. Aquest procés es realitza en varies etapes les quals consten en: crear un model numèric de la mama, calcular el problema directe, aplicar l'algoritme per sol.lucionar el problema invers, i finalment, presentar les imatges reconstruïdes.

Postprint (published version)

Image registration is an important task in medicine, especially when images have been acquired by different scanner/sensor types, since they provide information on different body structures (bones, muscles, vessels...). Several techniques have been proposed in the past, and among those, Normalized Mutual Information has been proven as successful in many cases. Normalized Compression Distance has been proposed as a simple yet effective technique for image registration. It is especially suitable for the case of CT-MRI registration. However, other image modalities such as PET pose some problems and do not achieve accurate registration. In this paper we analyse and propose a valid approach for image registration using compression that works properly for different combinations of CT, MRI and PET images., Peer Reviewed, Postprint (published version)

Projecte final de màster realitzat en col.laboració amb Universitat de Tecnologia de Eindhoven, Castellano: Di usion Tensor Imaging (DTI) es una técnica de imagen basada en la resonancia magnética (MR), que ofrece una visión única de la organización estructural de la sustancia blanca del cerebro. Esto se consigue mediante la medición de la difusión de moléculas de agua en el tejido. En el agua, la difusión es libre y tiene la misma magnitud en todas las direcciones. En este escenario obtenemos un per l de difusión isotrópico. En los tejidos brosos sin embargo, como en la materia blanca del cerebro, la difusión se limitaría en sentido perpendicular a las bras. Esto provoca que, en este escenario, obtengamos un per l de difusión anisotrópico. En las imágenes DTI, el per l de difusión se modela como una distribución de probabilidad de Gauss y por lo tanto puede ser descrito por un tensor de segundo orden [BPD94]. En este modelo, el principal eigenvector del tensor corresponde a la dirección de mayor difusión, que es la misma que sigue la estructura de bras (Figura 1.1). Gracias a estos tensores podemos realizar un recorrido sobre ellos e ir reconstruyendo un modelo 3D de estas bras. Esta reconstrucción recibe el nombre de ber tracking [VAD05] (Figura 1.2). Pese a su potencial, la salida de la reconstrucción, ber tracking, contiene una cantidad considerable de incertidumbre. Este error o incertidumbre se acumula a lo largo del proceso de obtención del modelo. En la fase de obtención de los datos puede introducir errores por ruido en las imágenes, distorsión de la imagen, parámetros del escáaner, etc. En la fase de reconstrucción se introducen errores de aproximación dependiendo del modelo de difusión utilizado. En la mayor parte de los algoritmos, este error no se muestra, lo cual da una sensación de certidumbre en los datos que no se corresponde con la realidad. Una aplicación para neurocirugía, no puede pasar por alto estos errores, ya que la aplicación se usara para tomar decisiones y evaluar el riesgo quirúrgico. Si no mostramos esta incertidumbre a la hora de

Electrical Impedance Spectroscopy (EIS) is a powerful tool to collect data from many biological materials in a wide variety of applications. Body composition fluid or tissue and organ state monitoring are just some examples of these applications. While the classical EIS is based on frequency sweep, the EIS technique using broadband excitations allows to acquire simultaneous impedance spectrum data. The strength and weakness of broadband EIS relies on the fact that it enables multiple Electrical Bio-Impedance (EBI) data collection in a short measuring time but at the cost of losing impedance spectrum accuracy. In general, there is a relationship between the broadband excitation time/frequency properties and the final EBI's accuracy obtained. This paper studies the influence of the multisine broadband excitation amplitude's design over the EBI accuracy by means of the resultant Noise-to-Signal Ratio (NSR) obtained when measuring with a custom impedance analyzer. Theory has been supported by a set of validation experiments., Peer Reviewed, Postprint (published version)

The present paper describes the design and usage of an Internet medical image search engine based on the queryby-example paradigm and the multimedia standard ISO-15938-12:2008. The system, which allows combining Content Based Information Retrieval (CBIR) techniques such as “query by image” with traditional XML metadata filtering, is applied to annotate and query Optical Biopsies-OB (in the present case Confocal endomicroscopy-CEM). The OB-CEM are taken by endoscopists, not trained in microscopic morphology which is the domain of the surgical pathology. To gain diagnostic confidence the endoscopists could consult the images to a pathologist or could use the system proposed in the paper. That is, to search for similar images to compare the diagnosis., Peer Reviewed, Postprint (published version)

Projecte final de carrera fet en col.laboració amb l’Hospital Clínic de Barcelona, Català: El diagnòstic de lesió cerebral en nounats es basa essencialment en la visualització de l’aspecte i de la “densitat ecogràfica” del teixit (el nivell de blanc sobre negre del teixit), interpretat de manera subjectiva pel metge. Tanmateix, els ecògrafs actuals tenen una sensibilitat extremadament elevada per detectar canvis i reflectir-los en la pantalla. El problema principal és que l’ull humà no és capaç de distingir de manera reproduïble, i molt menys quantifica, aquests canvis. Aquest problema, que és comú a totes les tècniques d’imatge, ha portat al naixement d’un camp nou en la biotecnologia que intenta desenvolupar eines d’anàlisi automàtica que permetin convertir els mil.lions de dades que ofereixen les tecnologies d’imatge en parámetres numèrics quantitatius. Per aquest tipus de paràmetres s’ha creat el terme de biomarcadors quantitatius d’image (Quantitative Imaging Biomarkers). Aquest projecte es basa en el desenvolupament d’un mètode automàtic per inferir les característiques del teixit mitjançant l’anàlisi automatitzada de la imatge ecogràfica. La tecnologia es basa en el concepte de que si un teixit conté diferències en la distribució i característiques de les cèl·lules i del líquid que les envolta, l’ona acústica que es reflecteix és diferent i en conseqüència la reconstrucció de la imatge també ho és. Aquests canvis es tradueixen en què l’ecògraf construeix una imatge diferent depenent de les característiques físiques que s’observen, però aquesta diferència és subtil i només es tradueix en un canvi en la “textura” de la imatge. Aquestes textures no poden ser detectades de manera reproduïble per l’ull humà, per molt entrenat que estigui, però sí per un programa d’anàlisi avançada de textures adaptat a l’interpretació d’ecografia i entrenat per reconèixer els patrons que s’associen a dany cerebral. La capacitat d’avaluar les propietats acústiques mitjançant algorismes sofisticats que permetin detectar els canvis induïts en l’ona acústica en func

Retinal images are widely used for diagnostic purposes by ophthalmolo- gists. Therefore, these images are suitable for digital image analysis for their visual enhancement and pathological risk or damage detection. Here, we implement a lu- minosity and contrast enhancement technique based on domain knowledge. We also review and analyze a previous approach in optic nerve head segmentation to extend its applicability to non circular shaped contours. We introduce a di erent strategy based on the use of active contours.

Magnetic Resonance Imaging (MRI) is an important paraclinical tool for diagnosing Multiple Sclerosis (MS) and providing several markers of disease activity and evolution. Traditionally, hypointense lesions on T1-weighted images have been reported to represent areas where demyelination and axonal loss have occurred, and are the images usually selected for segmenting the encephalic parenchyma. Based on the fact that in T1-weighted images MS lesions cannot be located within cerebrospinal fluid regions (CSF), a correct detection of such regions is very useful to filter MS’s false detections. However, the gray levels similarity among some MS lesions and CDF regions makes of the encephalic parenchyma detection process a difficult task. In this work we propose an approach for detecting CSF regions in which, for taking into consideration aforementioned gray-level vagueness, as well as the intrinsic uncertainty of CSF boundaries, we make use of fuzzy techniques. The proposed algorithm performs a fuzzy local analysis based on gray-level and texture characteristics, but considering the location and size of the CSF regions. As a result, the algorithm allows discriminating cerebrospinal fluid regions inside the intracranial region, providing confidence degrees that match with the possibility of including pixels associated to MS lesion, Peer Reviewed, Postprint (published version)

Consultable des del TDX, Títol obtingut de la portada digitalitzada, The aims of this work are: 1. To provide the normal cervical and coelomic radiographic appearance of the Loggerhead sea turtle, in the dorso-ventral view, as well as other useful landmarks, to allow for correlation of shell scutes with internal anatomic structures. 2. To provide the normal radiographic anatomy of the limbs of the Loggerhead sea turtle in combination with data obtained from computed tomography - osteological, gross anatomical and histological data. 3. To describe the normal ultrasonographic appearance of cervical structures and coelomic organs of the Loggerhead sea turtle, and to provide the respective images of frozen cross-sections for anatomical reference. 4. To provide normal computed tomographic images of the vertebral column and coelomic structures of the Loggerhead sea turtle, by establishing reference standards for organ size and position in this species; to provide images of virtual tracheo-bronchoscopy and 3D reconstructions of the respiratory tract and bone structures. 5. To provide the normal magnetic resonance imaging (MRI) appearance of coelomic structures of Loggerhead sea turtles in T1 and T2-weighted scans via comparison with cross-sectional anatomic sections of this species.

Consultable des del TDX, Títol obtingut de la portada digitalitzada, The aims of this work are: 1. To provide the normal cervical and coelomic radiographic appearance of the Loggerhead sea turtle, in the dorso-ventral view, as well as other useful landmarks, to allow for correlation of shell scutes with internal anatomic structures. 2. To provide the normal radiographic anatomy of the limbs of the Loggerhead sea turtle in combination with data obtained from computed tomography - osteological, gross anatomical and histological data. 3. To describe the normal ultrasonographic appearance of cervical structures and coelomic organs of the Loggerhead sea turtle, and to provide the respective images of frozen cross-sections for anatomical reference. 4. To provide normal computed tomographic images of the vertebral column and coelomic structures of the Loggerhead sea turtle, by establishing reference standards for organ size and position in this species; to provide images of virtual tracheo-bronchoscopy and 3D reconstructions of the respiratory tract and bone structures. 5. To provide the normal magnetic resonance imaging (MRI) appearance of coelomic structures of Loggerhead sea turtles in T1 and T2-weighted scans via comparison with cross-sectional anatomic sections of this species.

Volume data models are becoming larger and larger as the capture technology improves. Thus, their visualization requires high computational power. The automatic presentation of volume models through representative images and/or exploration paths becomes more and more useful. Representative views are also useful for document illustration, fast data quality evaluation, or model libraries documentation. Exploration paths are also useful for video demonstrations and previsualization of captured data. In this paper we present a fast, adaptive method for the selection of representative views and the automatic generation of exploration paths for volume models. Our algorithm is based on multi-scale entropy and algorithmic complexity. These views and paths reveal informative parts of a model given a certain transfer function. We show that our method is simple and easy to incorporate in medical visualization tools., Postprint (author’s final draft)

Breast cancer is the second cause of death among women cancers. Computer Aided Detection has been demon- strated an useful tool for early diagnosis, a crucial as- pect for a high survival rate. In this context, several re- search works have incorporated texture features in mam- mographic image segmentation and description such as Gray-Level co-occurrence matrices, Local Binary Pat- terns, and many others. This paper presents an approach for breast density classi¯cation based on segmentation and texture feature extraction techniques in order to clas- sify digital mammograms according to their internal tis- sue. The aim of this work is to compare di®erent texture descriptors on the same framework (same algorithms for segmentation and classi¯cation, as well as same images). Extensive results prove the feasibility of the proposed ap- proach., Postprint (published version)

The registration of cardiac images is relevant for the diagnosis of cardiac pathologies. In the case of SPECT (Single Photon Emission Computer Tomography) datasets, this is difficult because of the low resolution of the images. In this paper we present a new registration method for 3D images with poor resolution. Our method is based on the maximization of the mutual information (MI) using a new interpolation method which significantly improves the result of an MI-based strategy on low resolution images. We also use a different optimization strategy than other MI-based algorithms, typically achieving errors of the order of one pixel. We also discuss experimental results of our method., Postprint (published version)

Peer Reviewed, Postprint (published version)

La utilización de radiación de microondas para la obtención de imágenes tomográficas de cuerpos biológicos es un nuevo campo de investigación que puede englobarse dentro de los métodos de diagnóstico no invasivo. El elemento diferencial con respecto a los demás métodos en el propio sensor, lo que asegura la complementaridad con el resto de procedimientos. Las microondas, zona del espectro electromagnético comprendida entre 300 MHz y 300 GHz, no poseen la energía cuántica necesaria como para romper los enlaces intermoleculares y por tanto constituyen una radiación no ionizante. Este hecho, considerado aisladamente, hace atractivos los procedimientos de detección de imágenes biológicas basados en este tipo de radiación., Peer Reviewed, Postprint (published version)