[home] [Personal Program] [Help]

---

Session: Soft Tissue & Abdominal Organs
Session starts: Thursday 24 January, 13:30
Presentation starts: 14:15
Room: Lecture room 559

---

Niels Petterson (Eindhoven University of Technology)
Marc van Sambeek (Catharina Hospital Eindhoven)
Frans van de Vosse (Eindhoven University of Technology)
Richard Lopata (Eindhoven University of Technology)

Abstract:
Abdominal aortic aneurysms (AAAs), dilations of the abdominal aorta, can rupture and cause a fatal haemorrhage. Treatment is possible, however, not without risk for the patient. Surgery is decided upon when the diameter exceeds a threshold, based on a population average outcome. However, some aneurysms rupture below this threshold, while others remain stable above this criteria. A better rupture risk assessment is needed. Ultrasound is widely used to examine both geometry and dynamic motion of the aorta. These parameters serve as input for novel computer models which estimate the stress and stiffness of the vascular wall. The stress and stiffness can be used as possible rupture risk markers. The accuracy of the models is greatly dependent on their input. Low lateral contrast and resolution in ultrasound cause large variance in both geometry and motion estimation. Therefore, this study proposes multi-perspective imaging to enhance both contrast and resolution. 2-D cross-sectional ultrasound views of the abdominal aorta were acquired with a curved array probe in 6 healthy volunteers. Images from multiple perspectives were taken by physically moving the probe over the abdomen from side to side. 5 acquisitions were made under angles between -30° and +30°. An arch construction was used to fixate the probe, restricting it to a single plane. To manually segment the spine, its visibility was enhanced by calculating the cross-correlation coefficient of image sectors over an entire heart cycle. The stationary spine will yield high correlations. Segmentations of both the spine and the aorta were used to register the acquisitions. Electrocardiograms were used for time-alignment. Compounding was realized by averaging the envelope intensities. Compounding showed a significant increase in image clarity, especially of the aortic wall. Contrast-to-noise ratios increased by a factor of 3.2, from 0.074 to 0.240. The contrast increase is present during the entire heart cycle. Slight changes in heart frequency caused only minor temporal misalignment. The enhanced images increase accuracy of vascular geometry and motion detection, and could improve rupture risk assessment in AAAs. Ongoing work focuses on more advanced compounding schemes as well as the combination of motion vectors from multiple perspectives.