Statistical Shape Analysis for the Human Back
Abstract
In this research, Procrustes and Euclidean distance matrix analysis (EDMA) have been investigated for analysing the three-dimensional shape and form of the human back. Procrustes analysis is used to distinguish deformed backs from normal backs. EDMA is used to locate the changes occurring on the back surface due to spinal deformity (scoliosis, kyphosis and lordosis) for back deformity patients. A surface topography system, ISIS2 (Integrated Shape Imaging System 2), is available to measure the three-dimensional back surface. The system presents clinical parameters, which are based on distances and angles relative to certain anatomical landmarks on the back surface. Location, rotation and scale definitely influence these parameters. Although the anatomical landmarks are used in the present system to take some account of patient stance, it is still felt that variability in the clinical parameters is increased by the use of length and angle data. Patients also grow and so their back size, shape and form change between appointments with the doctor. Instead of distances and angles, geometric shape that is independent of location, rotation and scale effects could be measured. This research is mainly focusing on the geometric shape and form change in the back surface, thus removing the unwanted effects. Landmarks are used for describing back information and an analysis of the variability in positioning the landmarks has been carried out for repeated measurements. Generalized Procrustes analysis has been applied to all normal backs to calculate a mean Procrustes shape, which is named the standard normal shape (SNS). Each back (normal and deformed) is then translated, rotated and scaled to give a best fit with the SNS using ordinary Procrustes analysis. Riemannian distances are then estimated between the SNS and all individual backs. The highest Riemannian distance value between the normal backs and the SNS is lower than the lowest Riemannian distance value between the deformed backs and the SNS. The results shows that deformed backs can be differentiated from normal backs. EDMA has been used to estimate a mean form, variance-covariance matrix and mean form difference from all the normal backs. This mean form is named the standard normal form (SNF). The influence of individual landmarks for form difference between each deformed back and the SNF is estimated. A high value indicates high deformity on the location of that landmark and a low value close to 1 indicates less deformity. The result is displayed in a graph that provides information regarding the degree and location of the deformity. The novel aspects of this research lie in the development of an effective method for assessing the three-dimensional back shape; extracting automatic landmarks; visualizing back shape and back form differences.Publisher
University of WolverhamptonType
Thesis or dissertationLanguage
enDescription
A thesis submitted to the department of Engineering and Technology in partial fulfilment of the requirements for the degree of Master of Philosophy in Production and Manufacturing Engineering at the University of WolverhamptonCollections
Related items
Showing items related by title, author, creator and subject.
-
Trait Emotional Intelligence: Evaluating the theoretical construct, its relationship to other psychological variables, and potential interventions to enhance it.(2015)Research suggests that there are now two distinct approaches to Emotional Intelligence (EI): ability and trait. To date, however, the literature indicates that the construct remains poorly defined and not always adequately measured. Focusing on trait EI, the current thesis identifies a number of research questions that centre on what it is that defines EI in relation to existing definitions and other constructs, namely, happiness, self-esteem, mood and personality. Moreover, a programme of empirical study investigates whether a training intervention can enhance levels of EI and thus contribute to the emerging applied field of enquiry. This has been achieved through the employment of a series of studies. The initial study used the Repertory Grid Technique (RGT) and Principal Component Analysis (PCA) to generate a definition of EI, which directed this thesis towards alignment with the trait approach. The second study aimed to identify correlations and explore possible predictor variables through the application of Pearson’s r and Hierarchical Regression analysis. Moreover, a Mediation and Moderation analysis investigated whether EI has a mediating or moderating role when combined with other predictors. Two further experimental studies examined whether EI could be experimentally enhanced through a programme of relaxation and positive thinking when compared with a control group engaged in a non-demanding reading task. The results of the first study produced a definition of EI that included descriptions of work-related qualities with the second study yielding results of high correlations between EI, happiness and self-esteem, which were also identified as predictor variables. EI was found to act as a mediator and moderator. Analysis of Variance generated results for the first experimental study that showed overall non-significant interactions. To investigate beyond these findings, the second programme showed that the training programme induced positive changes. It was concluded that, overall, the results contribute to a definition beyond existing definitions of EI, demonstrating EI’s strong associations particularly with happiness, self-esteem and, its mediating and moderating role with other predictors. Primarily, the results from the second experimental study demonstrate the potential of EI in the applied field, including education, work and health.
-
Speaker identification using multimodal neural networks and wavelet analysis(IET, 2015-03-01)The rapid momentum of the technology progress in the recent years has led to a tremendous rise in the use of biometric authentication systems. The objective of this research is to investigate the problem of identifying a speaker from its voice regardless of the content. In this study, the authors designed and implemented a novel text-independent multimodal speaker identification system based on wavelet analysis and neural networks. Wavelet analysis comprises discrete wavelet transform, wavelet packet transform, wavelet sub-band coding and Mel-frequency cepstral coefficients (MFCCs). The learning module comprises general regressive, probabilistic and radial basis function neural networks, forming decisions through a majority voting scheme. The system was found to be competitive and it improved the identification rate by 15% as compared with the classical MFCC. In addition, it reduced the identification time by 40% as compared with the back-propagation neural network, Gaussian mixture model and principal component analysis. Performance tests conducted using the GRID database corpora have shown that this approach has faster identification time and greater accuracy compared with traditional approaches, and it is applicable to real-time, text-independent speaker identification systems.
