Doutorado

URI permanente desta comunidadehttps://repositorioacademico.universidadebrasil.edu.br/handle/123456789/914

Navegar

Agora exibindo 1 - 3 de 3

Análise e otimização dos algoritmos para angiografia por tomografia por coerência óptica e desenvolvimento de Phantom por impressão 3D
(Universidade Brasil, 2021) Rodrigues, Karina de Cássia; Amaral, Marcelo Magri; Almeida, Vilson Rosa de
Skin grafts are surgically applied to repair skin lesions such as burns and extensive necrosis. The success of this surgical procedure is associated with good blood supply in the grafted region. Thus, the assessment of angiogenesis during the tissue repair process is essential for its prognosis. The development of non-invasive evaluation techniques is extremely important for the success of this procedure. One of the promising techniques is Optical Coherence Tomography Angiography (OCT-A), a non invasive technique that can be used to obtain images of the vascularization of biological tissues. The adoption of this technique as a clinical practice in dermatology involves reducing its cost, and the use of equipment with a low acquisition rate (low cost) is a possible path. Thus, this work aimed to implement and optimize algorithms for obtaining angiography images by optical coherence tomography (OCT-A) for applications in images acquired with low acquisition rate and cost equipment. To test those methods, it is requiring the use of phantom that simulate the behavior of the microvascular system. Thus, this work also aimed at the development of a phantom to simulate a microvascular system using 3D printing technology. Phantoms containing microchannels were designed and printed on polylactic acid (PLA) using a 3D printer by fused filament deposition. These PLA phantoms were imaged with the OCT system (OQLabScope - Lumedica, USA). Seven different OCT-A methods were implemented (HFM, STS, CM, SV, OSV, ISC and UHS-OMAG) and compared against their processing time, signal-to-noise ratio, contrast, and contrast-to-noise ratio. The OSV and CM methods showed better overall performance based on these parameters, but CM shown higher processing time. An optimization of the CM method was proposed in this work, reducing the processing time by 99.2%, a significant gain for the algorithm that presented better performances in contrast.
Aplicações de modelos computacionais de análise de dados biomédicos em plataformas de dispositivos móveis
(Universidade Brasil, 2020) Sousa, José Vigno Moura; Almeida, Vilson Rosa de; Costa, Mardoqueu Martins da
This work develops CNNPulmona and CNNCardio, implementations of computational models for optimization of analysis of biomedical data from chest radiography (CXR) and electrocardiogram (ECG), respectively, deployable in mobile device platforms, in addition to a comparison of several methods of signal compression. CNNPulmona is an approach for classifying chest Xray images into three classes: bacterial pneumo nia, viral pneumonia (Covid19 or other type) and healthy lung. Convolutional Neural Networks are used, based on pretrained networks in conjunction with a quantization process, by means of the TensorFlow Lite platform method, thereby reducing the com putational cost. The cascade classification method is used, which makes it possible to divide the classifications into different stages; thus, it was possible to obtain 99.16% ac curacy in the classification of images with suspicion of Covid19. The resulting mobile application program also features a simple and intuitive user interface. In CNNCardio, a new method to classify electrocardiogram signals on mobile devices is proposed, which can classify different arrhythmias according to the EC57 standard of the Asso ciation for the Advancement of Medical Instrumentation. A convolutional neural net work was built, trained and validated with the MITBIH arrhythmia dataset, in which this database has 5 different classes: normal beat, premature supraventricular beat, pre mature ventricular contraction, ventricular beat fusion, normal and unclassifiable beat. After being trained and validated, the model is submitted to a posttraining quantization stage using the TensorFlow Lite conversion method.The results obtained were very sat isfactory, before and after quantization; the convolutional neural network obtained an accuracy of 99%. With the quantization technique, it was possible to obtain a significant reduction in the size of the model, thus enabling the development of the mobile applica tion; this reduction was approximately 90% in relation to the size of the original model. Additionally, the behavior of different signals was compared, when applied to different compression techniques, in order to test and find the best compression techniques for distinct types of biomedical signals, also proving that different types of biomedical sig nals behave distinctly in different types compression of biomedical signals, the results of this comparison of signal compression methods were very satisfactory, demonstrating that different types of compression can be used on signals for better results.
Avaliação do uso de sensor de contração muscular no treino sensório-motor em indivíduos com dor lombar persistente
(Universidade Brasil, 2020) Bonfim, Rafael Victor Ferreira do; Almeida, Vilson Rosa de; Garcia, Lívia Assis
Low back pain is a chronic health problem, with a high socioeconomic impact, due to its high prevalence; it corresponds to any pain felt in the lower back. It is also an expensive health problem, both on personal and social level. The deficiency in the mechanical stability of the lumbar spine is known to decrease spinal muscle activation and could result in the occurrence of pain symptoms in the lower back. The stabilizing musculature is activated to protect the spine during body movements; however, its training could be carried out through specific exercises. Sensor technology could facilitate this task, being implemented, for example, in applications such as monitoring the effectiveness of home rehabilitation interventions in individuals with low back pain. The aim of the present study is to analyze the effect of stabilizer muscle training using a contraction sensor in individuals with nonspecific persistent low back pain, by assessing pain, balance, plantar pressure, muscle isometric resistance test and muscle ultrasound thickness. The sample consisted of 30 individuals with chronic low back pain, aged between 18 and 45 years. In the present horizontal, double-blind, study, in which each individual was evaluated before and after the intervention with 16 visits of 30 minutes each, performed on alternate days, using the muscle contraction sensor for the abdominal muscles. Numerical pain scale, baropodometry, stabilometry, Biering Sorensen muscle isometric resistance test and examination of the ultrasound thickness of the transverse abdomen were used for assessment. As a result of the research, a reduction in pain levels was observed, as well as an improvement in plantar distribution and balance. The resistance of the multifidus and the thickness of the transversus abdominis increased with training employing the contraction sensor. It is concluded that the portable system, developed with low cost, could help in the control of the stabilizing muscle contraction exercises, improving its contraction capacity and motor control and, consequently, improving balance and reducing low back pain, which could cause a major socioeconomic impact.