Active Sensors: Calibration & Applications

  • Author / Creator
    Faraji, Mehdi
  • Active sensors, such as active cameras and ultrasound transducers, are becoming more popular. One particular type of active camera, the Pan-Tilt-Zoom (PTZ) camera, has become ubiquitous in surveillance platforms. Given their active nature, active cameras are omnipresent in robotic systems as well. Intravascular Ultrasound (IVUS) is an intra-operative imaging modality that facilitates observing and appraising the vessel wall structures of the human coronary arteries. In this thesis, I propose novel algorithms for two tasks on images acquired by the above-mentioned modalities. First, I propose new mathematical equations for calibrating an active PTZ camera along with a novel algorithm called Simplified Active calibration (SAC). The proposed equations are closed-form and linear and hence can be used in real-time. Also, SAC needs only 4 pairs of images to calibrate the camera. It can estimate the focal length in each direction with only one point correspondence between each pair without any calibration pattern. I have extensively evaluated and analyzed the proposed equations using synthetic and real images. The results illustrate that SAC can be employed in practical real-world applications. In the second part of the thesis, I propose a novel segmentation method for delineating two different parts of the blood vessel walls from IVUS images. Segmentation of arterial wall boundaries from the IVUS images is not only crucial for quantitative analysis of the vessel walls and plaque characteristics, but is also necessary for reconstructing 3D models of the artery. Using a feature detection algorithm, namely, Extremal Region of Extremum Level (EREL), the proposed method delineates the luminal and media-adventitia borders in IVUS frames acquired by 20 MHz probes. Next, I propose a region selection strategy to label two ERELs as lumen and media based on the stability of their texture information. I extensively evaluated our selection strategy on the test set of a standard publicly available dataset containing 326 IVUS B-mode images. The results of my experiments reveal that our selection strategy is able to segment the lumen with < 0.3 mm Hausdorff distance (HD) from the ground truth even though the images contain major artifacts, such as bifurcations, shadows, and side branches. Moreover, even when there are no artifacts, my method is able to delineate media-adventitia boundaries with 0.31 mm HD. Furthermore, my segmentation method runs in linear time. Based on this work, by using a 20-MHz IVUS probe with controlled pullback, not only can we now more accurately analyze the internal structures of human arteries, we can also segment each frame during the pullback procedure in real-time.

  • Subjects / Keywords
  • Graduation date
    Spring 2021
  • Type of Item
  • Degree
    Doctor of Philosophy
  • DOI
  • License
    This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.