Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.716074
Title: Comparison of peripheral quantitative computed tomography and magnetic resonance imaging for tissue characterisation in the gastrocnemius muscle
Author: Al Gohani, Fahad
Awarding Body: Cardiff University
Current Institution: Cardiff University
Date of Award: 2017
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
Rupture of the medial head of the gastrocnemius muscle (GM) is a common injury of the calf muscles. Magnetic resonance imaging (MRI) and ultrasound (US) are the medical imaging modalities that are usually used to assess such injuries. Texture analysis is a digital image processing technique that quantifies the relationship between pixel intensities (grey levels) and pixel positions. Texture can reveal valuable information that cannot be perceived by the naked eye. Dedicated image processing software is required to extract texture parameters. Texture analysis has been implemented for medical imaging modalities such as MRI, US and computed tomography (CT) for the evaluation sports muscle injury. Peripheral quantitative tomography (pQCT) is an adaptation of conventional CT. In this project, texture analysis was implemented on MRI and pQCT images of the gastrocnemius muscle (GM). MRI is an expensive technique that requires specialised facilities. Conversely, pQCT utilises a small-bore, low-dose X-ray scanner, which is portable and less costly than MRI. It has traditionally been used mainly for bone analysis. The aim of this study was to assess the suitability of pQCT for GM tissue characterisation using texture analysis compared with MRI. The study is novel in that it is the first to apply texture analysis to GM images using pQCT Texture analysis was done on image data acquired from MRI (GE, 1.5T) and pQCT (Stratec XCT 2000) in a group of healthy human subjects and an injured subject. A water phantom was also scanned with pQCT. An existing standard imaging protocol was observed for MRI acquisition, while pQCT image acquisition parameters were explored and optimised to yield a standard protocol. The pQCT scanner was shown to be capable of acquiring calf muscle images and distinguishing calf muscle boundaries. Texture parameters (grey level, variance, skewness, kurtosis, co-occurrence matrix, run length matrix, gradient, autoregressive (AR) model and wavelet transform) were extracted from the acquired images. The repeatability of these quantities for pQCT in a healthy human subject and a water phantom was assessed by calculating the coefficient of variation (%CV). The effect of pQCT parameters (scan speed and pixel size) was tested using multiple variate II analysis of variance (MANOVA). The effect of region of interest (ROI) area and anatomical position were evaluated using simple linear regression. The t-test was used to compare the mean values of the texture features in the right and left leg for both MRI and pQCT in a group of healthy human subjects. Neither MRI nor pQCT showed significant differences between the two legs for any of the texture features. In addition, there was no significant difference between the two modalities for the AR model and wavelet transform texture parameters. Reference ranges for the medial head of the GM were defined for both modalities. A study of a single injured subject revealed that the values of the AR model texture parameter fell outside the reference ranges for both MRI and pQCT, and so the AR model was identified as the most sensitive texture parameter for distinguishing injured from uninjured GM. The principal conclusion from this work is that pQCT has the potential to be used for imaging the gastrocnemius muscle and that GM images from both MRI and pQCT scanners can be objectively characterised by texture analysis. In addition, the autoregressive model texture parameter may be the most appropriate for muscle characterisation.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.716074  DOI: Not available
Share: