Abdominal aortic aneurysm (AAA) is a vascular disease resulting in a permanent local dilatation of the abdominal aorta. Different studies estimate the prevalence of AAA to 1.3-8.9% of men and 1.0-2.2% of women over 60 years of age. Risk factors include smoking, hypertension, high serum cholesterol, diabetes, and family history. The weakening of the wall and altered wall stress associated with aneurysm formation and progression may eventually lead to aneurysm rupture, which causes haemorrhage and severe blood loss and is associated with very high mortality. AAA is responsible for 1.3% of deaths among men aged 65-85 in developed countries. Elective repair of asymptomatic AAA is recommended when the risk of rupture is estimated to exceed the risk associated with repair. Currently, best clinical practice is to recommend repair when the maximum diameter of the aneurysm exceeds 50-55 mm or increases rapidly. This is a population-based criterion, meaning that in average, an aneurysm with diameter exceeding this criterion is more likely to rupture than to experience complications with repair. Individually, however, some aneurysms rupture before 50 mm, while several aneurysms larger than 55 mm are still intact. More patient-specific information about the state of the individual aneurysm is therefore warranted.
In this PhD thesis I have developed and investigated concepts and methods for ultrasound based strain estimation in AAA. The physiological motivation is that progression of aneurysm is associated with altered wall tissue composition, which leads to altered elastic properties, and altered wall stress (geometry and flow conditions). The underlying hypothesis is that it may be possible to detect and quantify this alteration from dynamic ultrasound images, and through that predict further progression.
We have developed a method for estimation of cyclic circumferential strain from 2D ultrasound. The method relies on the user to define the wall in an ultrasound image, and then automatically tracks a number of points in the wall over the cardiac cycle based on correlation between frames. The relative change in distance between neighboring points are used as a measure for strain estimation. Inhomogeneous strain values were found along the circumference of the aneurysms, suggesting that additional information could be obtained compared to using diameter alone. The method was further used for investigating strain in aneurysms before and after endovascular aortic repair (EVAR) in ten patients. Since insertion of a stentgraft reduces the load imposed on the wall, a successful EVAR should result in reduced strain. The results showed a clear reduction, which means that the expected reduction was indeed detectable using our method. The study included a limited patient material, and it remains to investigate if the strain values can be used for predicting clinical outcome after EVAR.
Because only a limited part of the aneurysm can be imaged in each cross-sectional view, we demonstrated a method for visualizing the circumferential strain from several image planes together in a 3D model using navigation technology. The 3D model may enhance interpretation of results by relating circumferential strain from several parts of the aneurysm to a 3D geometry. This is also an important step towards integration with wall stress simulations for adding more patient specific information.
Abdominal images may have relatively low signal to noise ratios, which will negatively influence the performance of the correlation based tracking method. Before larger clinical trials are initiated, it is therefore important to investigate the quality of the strain estimates obtained by the method. We developed a simulation model, for simulation of wall motion due to a time-varying blood pressure, and for simulation of ultrasound images including speckle, direction dependent reflection and absorption. The simulation model is an important part of future evaluation and tuning of the strain method.
Further refinement includes implementation of the processing method on an ultrasound scanner for real-time data analysis, which would benefit workflow and make it easier to find the most relevant image planes during investigation. Also, strain estimation from real-time 3D ultrasound is interesting for evaluating several strain components. Finally, clinical trials must be implemented for further investigating potential correlation between strain and clinically relevant parameters, including formation, growth and rupture of AAA.