Computer assisted analysis of contrast enhanced ultrasound images for quantification in vascular diseases

Abstract

Contrast enhanced ultrasound (CEUS) with microbubble contrast agents has shown great potential in imaging microvasculature, quantifying perfusion and hence detecting vascular diseases. However, most existing perfusion quantification methods based on image intensity, and are susceptible to confounding factors such as attenuation artefacts. Improving reproducibility is also a key challenge to clinical translation. Therefore, this thesis aims at developing attenuation correction and quantification techniques in CEUS with applications for detection and quantification of microvascular flow / perfusion.

Firstly, a technique for automatic correction of attenuation effects in vascular imaging was developed and validated on a tissue mimicking phantom. The application of this technique to studying contrast enhancement of carotid adventitial vasa vasorum as a biomarker of radiation-induced atherosclerosis was demonstrated. The results showed great potential in reducing attenuation artefact and improve quantification in CEUS of carotid arteries. Furthermore, contrast intensity was shown to significantly increase in irradiated carotid arteries and could be a useful imaging biomarker for radiation-induced atherosclerosis.

Secondly, a robust and automated tool for quantification of microbubble identification in CEUS image sequences using a temporal and spatial analysis was developed and validated on a flow phantom. The application of this technique to evaluate human musculoskeletal microcirculation with contrast enhanced ultrasound was demonstrated. The results showed an excellent accuracy and repeatability in quantifying active vascular density. It has great potential for clinical translation in the assessment of lower limb perfusion.

Finally, a new bubble activity identification and quantification technique based on differential intensity projection in CEUS was developed and demonstrated with an in-vivo study, and applied to the quantification of intraplaque neovascularisation in an irradiated carotid artery of patients who were previously treated for head and neck cancer. The results showed a significantly more specific identification of bubble signals and had good agreement between the differential intensity-based technique and clinical visual assessment. This technique has potential to assist clinicians to diagnose and monitor intraplque neovascularisation.