Back to search

National Technical University of Athens (NTUA)

Υπολογιστική ανάλυση μηχανικών χαρακτηριστικών του αρτηριακού τοιχώματος από απεικονιστικές καταγραφές με στόχο την υποβοήθηση της διάγνωσης της καρωτιδικής αθηρωμάτωσης

Abstract

dc:description

This dissertartion focused on motion analysis from B-mode ultrasound image sequences of the arterial wall of the carotid artery, in an attempt to investigate the unexplored role of mechanical features. Firstly, an in silico framework, consisting of 13 computer-generated image sequences that simulate realistic scenarios of ultrasound image recordings, was developed to assess the suitability of an algorithm in the detection of motion of the arterial wall of the carotid artery from B-mode ultrasound image sequences. Then, the most suitable algorithm, in terms of sufficient accuracy and low computational cost, was investigated. Towards this direction, a number of motion estimators were designed: (a) adaptive-block-matching algorithms, (b) algorithms that combine the conventional block matching with Kalman filter, and (c) multiscale versions of popular motion estimators. In a next step, a fully automated algorithm for segmenting the atherosclerotic plaque from B-mode ultrasound images of the carotid artery wall was developed, in an attempt to achieve user-independent identification of regions of interest for subsequent motion analysis. Then, the optimal computer-aided-diagnosis (CAD) scheme, which incorporates mechanical features of the arterial wall to characterize the atherosclerotic plaque, was investigated. To this end, real ultrasound image recordings were collected for subjects who referred to two different hospitals for ultrasound examination of the carotid artery, and the hereafter mentioned steps were followed: (1) a framework for measuring a wide variety of mechanical features was designed, (2) 15 candidate classification schemes were implemented, and (3) the most suitable classification scheme was identified. This classification scheme achieved classification accuracy equal to (a) 88% in a set consisting of two equal-sized subsets of ultrasound image data of the two hospitals and (b) 96% when it was applied separately to the data subsets of the hospitals. The high classification performance was maintained (92%) in a large dataset, which was balanced using the SMOTE algorithm, while, when the mechanical features were combined with textural and biochemical indices, the classification accuracy was even higher (95%). Going one step further in motion-based CAD, a voice-recognition analogue, which relies on hidden Markov models and is guided by spatiotemporal patterns, representing motion and strain activity in the arterial wall, was designed. Upon optimisation, the designed CAD was fed with 22 motion and 2 strain patterns, thereby achieving classification accuracy between 76% and 79%. Finally, the outcomes of the dissertation were incorporated in the web-based platform CAROTID. The methodologies which were developed in the framework of this dissertation were based only on patient data which are collected in the current clinical practice. Therefore, through their incorporation in an integrated tool for optimal management of patients, they are able to upgrade the potential of a low-cost routine examination into a powerful tool for objective and personalized clinical assessment of patients with carotid atherosclerosis.

Degree

thesis:*
Grantor dc:publisher
National Technical University of Athens (NTUA)
Year dc:date
2014

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Gastounioti, Ainilia
  • Γαστουνιώτη, Αιμιλία

Subjects

dc:subject × 16

Rights

dc:rights
Statement dc:rights
  • BY
Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/42828

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Gastounioti, Ainilia; Γαστουνιώτη, Αιμιλία. Υπολογιστική ανάλυση μηχανικών χαρακτηριστικών του αρτηριακού τοιχώματος από απεικονιστικές καταγραφές με στόχο την υποβοήθηση της διάγνωσης της καρωτιδικής αθηρωμάτωσης. National Technical University of Athens (NTUA), 2014. http://hdl.handle.net/10442/hedi/42828