Cardiff University | Prifysgol Caerdydd ORCA
Online Research @ Cardiff 
WelshClear Cookie - decide language by browser settings

Investigating facial shape using multilevel principal component analysis

Galloway, Jennifer ORCID: https://orcid.org/0000-0003-4800-5970 2021. Investigating facial shape using multilevel principal component analysis. PhD Thesis, Cardiff University.
Item availability restricted.

[thumbnail of JGalloway_Final Corrected PhD Thesis.pdf] PDF - Accepted Post-Print Version
Available under License Creative Commons Attribution Non-commercial No Derivatives.

Download (26MB)
[thumbnail of Cardiff University Electronic Publication Form] PDF (Cardiff University Electronic Publication Form) - Supplemental Material
Restricted to Repository staff only

Download (1MB)

Abstract

SUMMARY Aims: 1. To determine the influence of geographical location, sex, height, Body Mass Index (BMI), age (14-16 years old), pubertal stage, metabolic factors, atopy, breathing disorders, maternal smoking and alcohol consumption during pregnancy on facial shape. 2. To explore the usefulness of Multilevel Principal Component Analysis (mPCA) in facial shape research. Method: The influence of geographical location and sex was assessed using 21 landmarks on 3D facial scans of subjects from Croatia (n=73), England (n=79), Wales (n=50) and Finland (n=47). The influence of sex, height, BMI, age (14-16 years old), pubertal stage, metabolic factors, atopy, breathing disorders, maternal smoking and alcohol consumption during pregnancy on adolescent facial shape was assessed using 1000 and 7160 quasi-landmarks on 3D facial scans of the Avon Longitudinal Study of Parents and Children (ALSPAC) cohort (n=1411). The results of mPCA were compared to those using landmarks only, conventional Principal Component Analysis (PCA), Discriminate Function Analysis (DFA) and Partial Least Squares Regression (PLSR). mPCA was also assessed as a variable selection tool prior to PLSR. Results: mPCA provided more meaningful information in the exploratory phase of data analysis than conventional PCA and DFA. However, the results must be interpreted with caution when group sizes are imbalanced. All variables reached significance, except for age, in their respective mPCA models. Geographical location, sex, height, BMI and fasting insulin explained greater than 5% of the total variation. These variables also reached significance in the PLSR models. Therefore 5% may be a useful threshold for PLSR variable selection. Conclusions: Sex, geographical location, height, BMI and fasting insulin had the most influence on facial shape. mPCA appears to be a useful tool for visualising the maximum variation between groups of subjects when group sizes are balanced and as a variable selection tool to inform more sophisticated models such as PLSR.

Item Type: Thesis (PhD)
Date Type: Completion
Status: Unpublished
Schools: Dentistry
Subjects: R Medicine > RK Dentistry
Date of First Compliant Deposit: 23 September 2021
Last Modified: 10 Nov 2022 09:46
URI: https://orca.cardiff.ac.uk/id/eprint/144386

Actions (repository staff only)

Edit Item Edit Item

Downloads

Downloads per month over past year

View more statistics