nib

POSTGRADUATE THESES






Posgrado en curso

IN PROGRESS


Updated on December 11, 2023




Lic. NFC Natalia Garay, Thesis Advisors: Franco Simini and Humberto Prinzo, PROINBIO master's degree (desde 2020)

Ing. Isabel Morales, Thesis Advisors: Franco Simini and Joaquim Mendes (FEUP-Oporto), FING doctorate (desde 2018)

Ing. Paola Sciarra, Thesis Advisor: Franco Simini, FING master's degree (desde 2017)

Dr. Alejandro Masner, Thesis Advisors: Franco Simini and Fernanda Blasina, PEDECIBA master's degree (desde 2014)

Ing. Gabriel Slomovitz, Thesis Advisors: Eduardo Mizraji and Franco Simini, FING master's degree (desde 2014)

APPLICANTS FOR MASTER/DOCTORATE PROJECTS




Ing. Lourdes Cairelli, FING master's degree

Ing. Gustavo Sosa, FING master's degree

Ing. Rodolfo Grosso, Thesis Advisors: Franco Simini, FING master's degree

Dr. Marcelo Monzón, Thesis Advisors: Sofía González and Franco Simini, PEDECIBA master's degree

Dra. Andrea Mattiozzi, Thesis Advisors: Luis Dibarboure and Franco Simini, PEDECIBA master's degree


DEFENDED THESES


Updated on December 11, 2023




Instrument for clinical use for the three-dimensional study of the knee in motion with inertial sensors: CHAKAMO - FING master's degree 2023 - Colibri | PDF

In recent years, the analysis of human movement has been used in various applications such as patient rehabilitation, analysis to improve performance in sports, among others. To make this possible, the standard method consists of tracking optical markers located in each body segment. These systems are large and expensive, in addition to requiring qualified personnel for their correct operation, and the analyses are carried out in a laboratory with a defined workload. With the advancement of technology, new methods for tracking human movement began to emerge, one of which uses magneto inertial sensors (MIMU) containing miniature electronics, an accelerometer, a gyroscope and a magnetometer to estimate the orientation of the moving object. MIMUs present great versatility of use in a variety of work environments that are no longer limited to laboratories. Orientation estimation using MIMUs is possible through the use of sensor fusion (SFA) algorithms. These algorithms combine accelerometer, gyroscope and magnetometer information to estimate the orientation of the MIMU. There are several AFSs in the literature that can be classified into two major groups: deterministic and stochastic. The present work addressed four SFA, two deterministic (Madgwick and Valenti) and two stochastic (Valenti and Guo) in order to compare their behavior. For this purpose, two low-cost MBIENTLAB MIMUS and a lower limb simulator with its hinge joint were used. The simulator consists of two hinged wooden rulers without any ferromagnetic element to avoid its interference in the magnetometer measurements. A comparison of the four AFSs was carried out under a static test at 0◦, 30◦, 45◦, 60◦, 90◦ rotation about the swing axis for 60 seconds. In this test, the standard deviation of the SFAs and the mean squared error of the comparisons between them were estimated. A dynamic test was also performed consisting of two motions around the roll axis from 0◦ at 90◦ at two speeds similar to those of human motion. In it, the behaviors between algorithms were analyzed and it was observed that the two algorithms proposed by Valenti had inaccurate behavior in the estimates especially in the heading axis. It was also observed that the estimates varied between MIMUs despite being of the same make and model. After these tests, the SFA of Guo et al. was selected for its stable behavior in both static and dynamic tests. This AFS was used to estimate the kinematics of the knee with a healthy volunteer. The test with the volunteer consisted of going up and down a step with only one leg. This motor task sought to analyze the loaded knee and the instability of the joint when supporting the body with one leg. For this purpose, a wooden staircase with two steps, one of 17 cm and the other of 34 cm, was used. To estimate the behavior of the knee it was necessary to use two MIMUs, one on the thigh and one on the leg, being the segments that articulate at the knee. When using the MIMUs in the analysis of human movement it was necessary to calibrate the instruments to the segments in which they are attached. The calibration selected in this work was of two static positions; one standing and the other lying on the ground. The purpose of the calibration was to determine an initial reference position for the detection of the motor task. Finally, two tests were performed on the volunteer’s right knee. For this, 10 up and down cycles were performed on the 17 cm step and another test on the 34 cm step. At the end of the test, a clinical report was generated with the average number of cycles of the motor task and a 3D dynamic representation of the knee movement. These results were calculated in a script in the Python programming language, this work being the first step towards the development of an instrument for clinical use called ChakaMo.

Author: Ing. Maria Rene Ledezma
Thesis Advisors: Prof. Ing. Franco Simini
Co-Thesis Advisor: Prof. Dr. Darío Santos


Instance-based learning following physician reasoning for assistance during medical consultation - PEDECIBA master's degree 2021 - Colibri | PDF

The search for better medical practices is a perpetual challenge for modern medicine. In this regard, computational intelligence has emerged as a promising subject for developing smart systems in healthcare practice (Castellano and Casalino, 2020). Computational intelligence allows implementing automatic tools, enabling physicians to provide patients with a better quality of attention by performing early and accurate diagnosis and improving treatment. Furthermore, automatic systems and technologies based on computational intelligence have proven to be useful solutions to be applied in clinical practice. Some important advantages of intelligent automatic methods over traditional ones include better efficiency, accuracy, and consistency, among others. Consequently, intelligent automatic systems provide physicians with more time for face-to-face consultation, and more time for critical tasks and critical cases (López-Rubio et al., 2015). A specific subject where the learning capabilities of computational intelligence methods is very helpful to improve medical practice is analyzing and processing Electronic Medical Records (EMRs). EMRs refer to digital records, collected by the individual medical practice, that contain the general health information of patients (Habib, 2010). They usually consist of several types of health data, including, but not limited to, demographics, past surgical history, medical family history, social history, medication, allergies, test results, and medical images. Currently, the majority of medical history recording products are based on predefined templates, which provide very limited flexibility for writing patients medical records. Several drawbacks are identified on medical recording products. In particular, structured data entries are an obstacle to flexible writing in medical record applications, and are disapproved by physicians, who usually prefer writing free text (González et al., 2018). Despite the fact that physicians are getting used to work with electronic medical records, they still have difficulties dealing with long lists of pre-conceived variables, usually included in EMR systems. Although conventional EMR systems are useful to achieve legible, accessible, and complete documentation of medical consultations, they also cause several difficulties for physicians who adopt them. In many cases, physicians spend a lot of time searching for an option to record what they intend to annotate. Another drawback of conventional products concern to the alerts and suggestions they provide, which are generally based on previously defined rules, or according to mechanisms, whose behaviour remains the same throughout their operational life.

Author: Matías Galnares
Thesis Advisors: Sergio Nesmanchnow y Franco Simini


Kinematic Study of the Knee with injury to the anterior cruciate ligament - PROINBIO PhD 2021 - Colibri | PDF

ACL injury is the most frequent and devastating knee ligament injury, mainly as a consequence of sports activity. The usual clinical evaluation is based on subjective tests of joint instability and laxity, such as Lachman's maneuvers, anterior drawer maneuvers, pivot shiftand subjective patient-reported questionnaires. This thesis shows that they do not correlate with the function of the knee joint. In this doctoral thesis, a new methodology for assessing the knee with ACL injury is proposed, which quantifies the dynamic function (arthrokinematics) of the tibiofemoral joint during a specific motor task. The following arthrokinematic parameters of the tibiofemoral joint were studied: (i) Tibio Femoral Contact Point using the Baltzopoulos method and (ii) Tibial Rotation. For this, a stereophotogrammetry system and a novel videofluoroscopy processing of the knee (CINARTRO) were used. Based on the knowledge generated about knee joint function before and after devastating ACL injury, a Minimally Supervised Incremental Home Exercise Program is proposed here. To optimize this treatment and delegate its monitoring to the patient, a portable, reliable and low-cost instrument (DINABANG) was conceived and developed. Thanks to the methodology and the instruments developed in this thesis.

Author: Darío Santos
Thesis Advisors: Alberto Leardini and Franco Simini


Epidemiology of periodontal disease in Uruguayan adults. Analysis of its associated individual and contextual factors. Population-based study in Uruguay - PROINBIO PhD 2018 - Colibri | PDF

This work has contributed to the understanding of the determinants of periodontal health, noting what contextual and individual characteristics, articulated, must be considered in explaning the phenomenon. For clinicians it will be useful to investigate these characteristics as individual risk factors, without losing sight of the determining contextual factors. At the population level, it is important to urge health Authorities to integrate oral health component to the National Strategy for the Prevention and Treatment of Non-communicable Diseases based on a Common Risk Factor Approach.

Author: Susana Lorenzo Erro
Thesis Advisors: Marco Peres, Silvana Blanco y Franco Simini


Project alternatives and implementation of circuits and reconstruction software for an electrical impedance tomograph for the compact display of the oedematic state of thoracic slices in real time. - FING master's degree 2014 - Colibri | PDF

Electrical impedance tomography is a technique that allows estimation of the conductivity distribution within a body based on surface measurements. To achieve this estimation, electrodes are placed on the body through which current is injected and voltage differences are measured. These voltages are used to arrive at the conductivity distribution by means of a so-called reconstruction. The result is a tomographic slice in which the different conductivity values are represented by different colours. The biomedical engineering nucleus of the University of the Republic has been working on this technique for more than a decade, arriving at a working prototype called IMPETOM. This prototype was very valuable at the time of its completion, but in order to continue with the line of research a redesign was necessary. The aim of this thesis is to design a functional prototype tomograph that allows real-time monitoring of lung function and acute lung injury. The proposed design has 16 electrodes that are placed in a row, using a Velcro strap, on the patient's chest. Current is injected at a frecuency between 20kHz and 100kHz using the adjacent electrode method. Digital demodulation is used to obtain the voltage values and pass the vector of measurements to the PC where the software that performs the reconstruction is located. Image reconstruction in Electrical Impedance tomography is an ill-conditioned inverse problem, so small errors and noise added to the measurements cause arbitrarily large errors in the reconstructed conductivity. To counteract this ill-conditioning our design includes regularisation of the problem and the addition of a priori information that guarantees that the reconstruction is within a set of expected solutions. Several regularisation methods exist, most of which restrict the solutions to regular distributions, while other methods allow for discontinuities and jumps in the conductivity. This thesis analyses and compares several of these methods and determines which is the most suitable for use in the new prototype. Simulated data and real data taken with the first IMPETOM prototype were used for the comparison.

Author: Eduardo Santos
Thesis Advisors: Franco Simini


Study of instantaneous knee rotation centres in patients with Anterior Cruciate Ligament plasty. - PEDECIBA master's degree 2014 - Colibri | PDF

A two-dimensional kinematic analysis of the Knee Joint (KJ) was performed in a population of 6 patients with anterior cruciate ligament (ACL) plasty, in order to identify the Centre of Rotation by Contact Point (CRPC), during the extension movement executed in open kinematic chain. Three situations were studied: prior to ACL plasty, after ACL plasty and in the contralateral uninjured knee. Serial radiographic images were obtained throughout knee extension by video-fluoroscopy. After studying the methods described in the literature, the method proposed by Baltzopoulos was adapted. An interactive computer application was developed to identify the anatomical structures and to determine the CRPC. The results obtained from the patients in the various situations were subjected to the Wilcoxon, Nemenyi, McDonald-Thomson`s test, with a significance level of 5%. No statistically significant differences in CRPC excursion were found between the injured knee and the healthy contra lateral knee. On the other hand, the data show a difference between the injured and reconstructed knee. The results obtained could suggest performing the plasty early, as soon as the clinical situation of the patient allows it, thus avoiding the natural evolution of the movement of the healty RA towards a type of movement of the injured knee. The experience gained and the observation that no instruments for dynamic assessment of the relative movement of the RA surfaces are available in the clinic suggest specifying and then constructing an instrument aimed at an assessment of greater specificity. The work developed in this thesis integrated basic knowledge (biomechanics, computation and physiology) with clinical activity, resulting in a deepening of the knowledge of phenomena related to ACL rupture and plasticity.

Author: Darío Santos
Thesis Advisors: Gustavo Brum and Franco Simini


Dead space reduction system for assisted ventilation of newborns - FING master's degree 2014 - Colibri | PDF

LAVESPI, from the Spanish conjunction of "Lavado" (washout) and "Espiratorio" (expiratory), was projected and built to improve the performance of new-born ventilators by adding the skill of washing out the instrumental dead space by injecting a slight flow during expiration, technique known as "Tracheal Gas Insufflation" (TGI). The design of LAVESPI consists on a microprocessed equipment with servo-control valves, sensors and a gas pump that includes the newfangled idea of using the same ventilation gas during the expiratory phase through a special endotracheal tube (multichannels incorporated in its wall). LAVESPI controls the airway pressure and stops the insufflation when necessary in order to avoid exceeding pressure limits that could harm the weak alveolar structures, LAVESPI can be added to all ventilators independently from its brands and models without disturbing its performance. LAVESPI includes special edotracheal tubes with 8 channels placed on its wall (e.g. Vygon 6501.30). The trials of LAVESPI show a mean reduction on the arterial carbon dioxide (PaCO2) of 19% (from 61.7 mmHg ( = 7;7) to 49.7 mmHg (σ = 5;6) in a 5 newborn piglets group with healthy lungs and 21% (from 61.2 mmHg (σ = 11;1) to 48.5 mmHg (σ = 5;7)) in a 2 newborn piglets group with injured lungs (mean weight of the 7 pigs 1641 g). LAVESPI is a prototype of USD 900 components cost, which has potential of application in the neonatal intensive care units, without discarding its possible adaptation for adult patients.

Author: Daniel Thevenet
Thesis Advisors: Miguel Martell y Franco Simini


Um computador por aluno fora do contexto escolar: cenas digitais do Plan Ceibal na fronteira do Brasil e Uruguai - UFRGS master's degree 2012 - LUME | PDF

O presente trabalho discute o processo de distribuição de um computador por aluno configurado a partir da implantação do projeto de inclusão digital adotado pelo governo uruguaio, este projeto denomina-se Plan Ceibal e teve seu início em 2007, sendo finalizada a primeira etapa de distribuição em 2010, momento em que foram distribuídos computadores para todos os alunos de escolas públicas que estivessem entre o 1º e 6º ano do ensino primário. Em uma cidade do interior daquele país realizamos um estudo etnográfico do tipo blended, combinando espaços da vida real com espaços virtuais de utilização da tecnologia oferecida pelo Ceibal. O presente estudo ocorreu de janeiro a agosto de 2011, nossa investigação procurou identificar novas formas culturais emergentes, analisando o uso destes computadores portáteis fora do espaço escolar através do estudo etnográfico. Assim conseguimos mapear os espaços de utilização dos computadores pelos alunos na comunidade, obtendo dados que apontam para mudanças em práticas culturais. No estudo etnográfico procuramos entender os tipos de interações que estes alunos estabeleciam no âmbito da internet e que tipo de uso estes faziam da rede para fins de interações. Através deste estudo foi possível ampliar nosso campo de estudo, procurando entender também a cultura gerada a partir do uso da internet e a participação de possíveis redes sociais. Assim identificamos o que muitos alunos faziam em redes sociais, sendo atuantes neste espaço, embora não tenha sido possível identificar relações entre os próprios alunos na rede já constituída. As interações entre estes e a pesquisadora apontam para trocas de capital social do tipo relacional e constituição de laços sociais fortes em alguns casos, mas em sua maioria laços fracos, que apenas estruturam a rede e não foram adiante nas interações. Para esta investigação consideramos que existe um complexo relacionamento entre o uso das tecnologias e as estruturas sociais. Esta pesquisa com viés sócio-histórico, direciona o olhar para o que fazem as pessoas com as tecnologias na construção de zonas de sentidos, e não apenas para as possibilidades pedagógicas da tecnologia. Por isso identificamos que estes alunos fazem uso da tecnologia fora da escola, trazendo novos sentidos para o uso, vinculando as práticas culturais estabelecidas e não a capacidade tecnológica da tecnologia em si. Identificamos ainda várias formas de socialização vinculadas ao uso dos computadores pelos alunos em espaços públicos. Portanto, verifica-se que a tecnologia está sendo utilizada como mais um componente do mio e que as produções a partir deste uso moldam novas práticas na internet e fora dela comunidade estudada, caracterizando a tecnologia como cultura e como artefato desta cultura.

Author: Marcia Maciel
Thesis Advisors: Liliana Passerino