A robust technique for the detection and quantification of abdominal aortic calcification using dual energy X-Ray absorptiometry

Elmasri, Karima 2018. A robust technique for the detection and quantification of abdominal aortic calcification using dual energy X-Ray absorptiometry. PhD Thesis, Cardiff University. Item availability restricted.

Preview	PDF - Accepted Post-Print Version Download (13MB) | Preview
	PDF - Supplemental Material Restricted to Repository staff only Download (181kB)

Abstract

Arterial calcification is a manifestation of atherosclerosis, which over the last two decades has become a recognised predictor of cardiovascular disease. Abdominal Aortic Calcification (AAC) and osteoporosis have been shown to coincide in older individuals. The accepted method of diagnosing osteoporosis is through the measurement of bone mineral density by dual energy x-ray absorptiometry (DXA). Vertebral fracture assessment (VFA) images obtained alongside BMD using DXA technology provide an inexpensive resource for AAC diagnosis. Although several simple methods have been proposed for manual semi-quantitative scoring of AAC in x-ray images in the past, these methods have limitations in terms of capturing small changes in atherosclerosis progression and are time-consuming. Several automatic approaches have been proposed to measure AAC on radiographs. However, these methods have not been related to any accepted medical AAC scoring systems and thus are not likely to be adopted easily by the medical community. In addition, there has been no attempt to apply the proposed methods to VFA images. The main focus of the research presented in this thesis is the automatic quantification of AAC in VFA images acquired in single energy mode. The thesis is divided into two main parts. In the first part, an automatic method for AAC detection and quantification in VFA images is proposed and evaluated on a large number of images. In the second part, the performance of both single and dual energy VFA imaging for the detection of uniformly distributed calcification is investigated. The automatic method for AAC detection consists of two stages. In the first stage an active appearance model was employed for the purpose of segmentaion. In the second stage, adaptive thresholding techniques were used to detect AAC, whilst automatic iii classification techniques were used to quantify the detected calcification. The performance of several classifiers were investigated, and the proposed method was evaluated against the manual AC-24 scoring method using several hundred images and two human readers. A thorough statistical analysis of the results showed that, overall, the SVM classifier gave the best results. Weighted accuracy, sensitivity, specificity assessed for 4 AAC categories were 89.2%, 78.5% and 92.3% respectively while the corresponding values for 3 AAC categories were 88.6%, 86%, 90.4%. In the second part, a study using a tissue-mimicking physical phantom is described. The phantom consists of an aluminium strip within Perspex to simulate calcification and abdominal soft tissue respectively. VFA images of different phantom configurations were acquired in single energy (SE) and dual energy (DE) modes. The minimum detectable aluminium thickness was assessed visually and related to contrast and contrast-to-noise ratio. Percentage coefficient of variation was used to quantify uniformity, repeatability and reproducibility with a Perspex width of 25 cm, the smallest thickness of aluminium that could be detected was 0.20- 0.25 mm. The initial results are promising, and the system proposed in this research can be used as an alternative method to the manual scoring system (AC-24) for a wide range of AAC. The principal conclusion from the phantom work is that under idealised imaging conditions, VFA images have the potential to be used for detecting small thicknesses of calcification with good linearity, repeatability and reproducibility in SE and DE modes for patients with a body width < 30 cm.

Item Type:	Thesis (PhD)
Date Type:	Completion
Status:	Unpublished
Schools:	Engineering
Uncontrolled Keywords:	Abdominal aortic calcification (AAC); Dual Energy x-ray absorptiometry (DXA); Vertebral Fracture Assessment (VFA); Active Shape and appearance Models (ASM & AAM); Contrast and Contrast to noise ratio (CNR); AC-24 points scoring system.
Date of First Compliant Deposit:	3 October 2018
Last Modified:	30 Mar 2021 10:57
URI:	https://orca.cardiff.ac.uk/id/eprint/115458

Actions (repository staff only)

Edit Item