Browsing by Author "Assis, Gilda Aparecida de"

Now showing 1 - 4 of 4
  • No Thumbnail Available
    Item
    Animação gráfica da marcha humana a partir de dados do Kinect.
    (2021) Leite, Edmo de Oliveira; Assis, Gilda Aparecida de; Yared, Glauco Ferreira Gazel
    A análise da marcha humana a partir de dados biométricos tem aplicações em áreas como segurança, robótica bioinspirada e saúde. Sensores de movimento de baixo custo, como o Kinect, têm permitido a aquisição de dados biométricos da marcha em ambientes terrestres. Entretanto, esses equipamentos têm limitações que podem impactar na qualidade dos dados. Nesse cenário, diferentes técnicas de processamento de sinais podem ser aplicadas para reduzir o ruído. A visualização desses dados, originais ou processados, muitas vezes é realizada na forma de gráficos, tendo utilidade limitada para profissionais não experientes na análise de sinais. Nesse sentido, a visualização dos dados da marcha em um modelo tridimensional pode contribuir para melhorar a decisão dos profissionais, principalmente da saúde. Este trabalho tem como objetivo realizar a animação da marcha humana em um modelo tridimensional, a partir dos dados coletados pelo sensor Kinect 2.0. Para reduzir o ruído dos dados, foi realizado um pré-processamento com filtros de média móvel e Butterworth. Foram elaborados vídeos das animações conforme as vistas isométrica e lateral, que foram incorporados em um questionário on-line e avaliados em uma pesquisa de campo sobre artificialidade/naturalidade da animação, utilizando-se a técnica de pontuação média de opinião (mean opinion score [MOS]). Um total de 22 participantes, estudantes de computação, respondeu ao questionário on-line. A análise de variância simples (analysis of variance [Anova]) one way mostrou que os vídeos a partir das vistas isométrica e lateral processados com filtro de média móvel (janela = 15 e repetições = 3) que obtiveram maiores valores da métrica MOS foram avaliados como significativamente mais naturais do que outros vídeos, processados ou não.
  • No Thumbnail Available
    Item
    Evaluation of a protocol for fMRI assessment associated with augmented reality rehabilitation of stroke subjects.
    (2019) Assis, Gilda Aparecida de; Brandão, Alexandre Fonseca; Corrêa, Ana Grasielle Dionísio; Castellano, Gabriela
    New technologies for rehabilitation involving Augmented Reality (AR) as a complement to conventional therapy have appeared in recent years. An earlier study for shoulder rehabilitation using the AR NeuroR computer system showed improved clinical outcomes for stroke patients. This study aims to analyze a proposed protocol to measure possible changes in functional brain connectivity associated with the use of the NeuroR system in the context of shoulder motor rehabilitation of post-stroke subjects. A pilot study was conducted with a poststroke patient, using resting-state functional magnetic resonance imaging (RS-fMRI). RS-fMRI signals were acquired pre and post use of the NeuroR system (pre-test and post-test), integrated into the patient’s rehabilitation program. Functional connectivity analysis of RS-fMRI was performed using the motor area as seed. he maximum connectivity value in the pre-test occurred in the ipsilesional parietal region while the maximum in the post-test was located in the ipsilesional frontal region. It was observed that the regions strongly associated with motor activity had higher connectivity values at post-test compared to pre-test. The proposed protocol is suitable and safe for verifying if functional brain connectivity was changed after the rehabilitation program with NeuroR training, indicating a possible neuroplasticity effect. Tests with a larger number of patients are still necessary.
  • No Thumbnail Available
    Item
    Investigation of fMRI protocol for evaluation of gestural interaction applied to upper-limb motor improvement.
    (2019) Brandão, Alexandre Fonseca; Casseb, Raphael Fernandes; Almeida, Sara Regina Meira; Assis, Gilda Aparecida de; Camargo, Alline Fernanda de Barros; Min, Li Li; Castellano, Gabriela
    The use of Virtual Reality (VR) systems for rehabilitation treatment as a complement to conventional therapy has grown in recent years. Upper limbs therapy using VR has already been shown useful for stroke patients. In this work, we present a pilot study aiming to investigate the use of a functional magnetic resonance imaging (fMRI) protocol to analyze brain connectivity changes in subjects undergoing upper limb training through a VR environment. Thirteen healthy subjects underwent resting-state fMRI exams before and after a VR session. Although no significant changes are expected in healthy subjects performing only one training session, this study could pave the way for future studies performed with both stroke patients or athletes performing more sessions. Indeed, no significant changes in motor cortex connectivity were found. Nonetheless, an evaluation protocol for this type of VR rehabilitation procedure was successfully established, to be used in further studies with patients or athletes.
  • No Thumbnail Available
    Item
    Wearables and detection of falls : a comparison of machine learning methods and sensors positioning.
    (2022) Pinto, Arthur Bernardo Assumpção; Assis, Gilda Aparecida de; Torres, Luiz Carlos Bambirra; Beltrame, Thomas; Domingues, Diana Maria Gallicchio
    Wearable sensors have many applications to provide assistance for older adults. We aimed to identify the best combination of machine learning algorithms and body regions to attach one wearable for real-time falls detection from a public dataset where volunteers performed daily activities and simulated falls. Accuracy and comfort of the combination of wearables and algorithms were assessed. Raw data from the accelerometer and gyroscope were used for both training and testing stages. We evaluated the confusion matrix between all wear- ables at each of the different body regions (Ankle, Right Pocket, Belt, Neck, and Wrist) for the following machine learning algorithms: Multilayer Perceptron (MLP), Random Forest, XGBoost, and Long Short Term Memory (LSTM) deep neural network. The accuracy was compared by ANOVA two-way repeated measures statistical test. This work has two main technical contributions. First, our results demonstrated the highest accuracy in identifying falls when the sensors were positioned on the neck or ankle. Second, when the machine learning algorithms to detect fall was compared, LSTM deep neural network and Random Forest showed statistically higher accuracy than MLP and XGBoost. Besides, a comfort analysis based on the literature concluded that neck and wrist are the most comfortable regions to wear wearables.