Propostas de Estágio 2014/2015 - Plurianual

DEI - FCTUC
Gerado a 2024-03-29 00:26:50 (Europe/Lisbon).
Voltar

Titulo Estágio

ORTHODIA II – Automatic characterization and diagnostic of gait disorders based on computer vision

Área Tecnológica

Reconhecimento de Padrões

Local do Estágio

ISR - Instituto de Sistema e Robótica

Enquadramento

Comprehensive measures of gait pathology are useful in clinical practice. They allow stratification of severity, give an overall impression of gait quality, and aid in objective evaluation of treatments outcome. There are many ways to gauge overall gait pathology. While parent and caregiver assessments are useful and practical, they lack the precision and objectivity provided by three-dimensional quantitative gait data. Gait data can be used to assess pathology in a variety of ways. For example, stride parameters such as walking speed, step length, and cadence provide an overall picture of gait quality. These parameters are especially useful after normalization to account for differences in stature. It is possible, however, to walk with adequate stride parameters and still have significantly atypical joint motions and orientations. This suggests a need for three-dimensional gait data in assessing overall gait pathology. Interpreting three dimensional gait data in a global sense is not a simple task. Difficulties arise from the complexity of gait, and from the interdependent nature of gait data. For example, to assess the motions of the lower extremities during a single stride requires the analysis of multiple joints and body segments in multiple planes at multiple instants of time. Furthermore, these motions are coupled across joints, planes, and time. It is clear, therefore, that some method for dealing with this complexity and interdependence is necessary to gain an overall sense of gait pathology.
This internship will be included in the project "Novas Tecnologias para apoio à Saúde e Qualidade de Vida, Projecto B- Diagnosis and Assisted Mobility for People with Special Needs”, financed by QREN-MaisCentro.

Objetivo

The main objective of this research project is to develop a computerized system to characterize and automatically diagnose gait disorders. The system will also provide an indication of the progress of the rehabilitation, either through a comparative analysis of the movement of patients.
The hardware system already developed consists of electronic equipment, a 3D vision system for joint trajectories acquisition system, connected to a personal computer. The patient gait and posture will be analyzed from the data acquired. These data will be processed using classification algorithms based on computational intelligence techniques. The Doctor Páscoa Pinheiro, who is the head of the rehabilitation department of the Hospital of the University of Coimbra, will provide patients for the experiments.

Plano de Trabalhos - Semestre 1

1) State of the art review - Review of the most recent approaches to this problem in the state of the art. Identification of the biggest challenges in the field.
2) Human joint trajectories acquisition from the 3D vision system. Comparison of the obtained data with data from the literature.
3) The data from 3D trajectories gathered in previous task should be preprocessed before used in the next step of gait pathology classification. Typically, denoising, scaling and normalization are the primary steps to be performed in order to eliminate bias and cleaning the data. For feature extraction we intend to make a step further and to implement computationally more efficient methods as ICA (Independent Component Analysis), SVD (Singular Value Decomposition), or even NMF (Non-Negative matrix factorization) than PCA (Principal Component Analysis). In addition, a fundamental and challenge problem in the analysis of data with small number of samples (10 patients in each pathology case) and very large number of features is the selection of relevant features.
4) 1st semester report - Writing and reviewing of the first semester report.

Plano de Trabalhos - Semestre 2

5) The classification module combines several SVM (Support Vector Machine); NN (Neural Network) and ELM (Extreme Learning Machine) classifiers trained with features computed from selected blocks of 3D trajectories until the individual threshold is reached by all classifiers. The final decision can be taken either based on one classifier or on competitive/complementary output of more classifiers. Classifier evaluation is based on their performance and decision speed (which is crucial for on-line implementations).
6) First algorithm implementation - Implementation and application of the first implementation. Results analysis and definition of possible improvements.
7) Algorithm improvements - Final implementation and application. Results analysis and comments.
8) Dissertation - Writing and reviewing of the dissertation document.

Condições

The workplan will be performed in ISR-Coimbra, where the student will be given a workstation and full support of a heterogeneous team composed by Informatics Engineers, Electrical Engineers and Medical specialist, with experience on the human gait capture and gait disorders. Our ISR lab is full equipped with the hardware which will be necessary for the student.

Payment:
A research grant (BI) can be considered depending on the curriculum vitae and motivation of the applicant after the conclusion of the master degree.

Observações

Experience on pattern recognition problems will be preferred.

Orientador

Bernardete Ribeiro e João Paulo Ferreira
bribeiro@dei.uc.pt 📩