Abstract:
Optical Coherence Tomography (OCT) and Ultrasound (US) are non-ionizing and non-invasive imaging modalities that are clinically used to visualize anatomical structures in the body. OCT has been widely adopted in clinical practice due to its micron-scale resolution to visualize in-vivo structures of the eye. Ultra-High Frequency Ultrasound (UHFUS) can capture images at a depth of $\sim$1cm with 30 micron resolution.

One key application area for OCT is assessing the stem cell distribution residing inside the Palisades of Vogt (POV) in the limbus. The limbus is located at the intersection of the clear cornea and the white sclera in the eye. Another application area is identifying corneal tissue interfaces for surgical procedures, such as Laser In-Situ Keratomileusis (LASIK). As for UHFUS, the key application in this thesis is vascular measurements, including monitoring the rejection of hand transplants.

Achieving higher resolutions with OCT and UHFUS increases the speckle noise during imaging due to smaller resolution cells. In addition to the speckle noise present at every pixel, other localized imaging artifacts, such as shadowing and specular saturation, substantially diminish the visibility of tissue interface boundaries. These boundaries and edges are crucial for diagnosing particular pathological conditions or diseases, and for developing a pre-operative surgical plan. Moreover, abrupt tissue motion obfuscates the analysis of long image sequences. An ideal approach to circumventing these issues would be thoroughly validated, produce measurements that are clinically/surgically relevant, and generate these metrics in a time-sensitive manner.

The main contribution of this thesis is in developing classical and learning-based approaches to address the problems faced in these clinical application areas, thereby advancing the clinical capabilities of OCT and UHFUS. Our methods seek to leverage the most salient regions of images as a basis for additional analysis of application-specific features and quantitative metrics. Vessel boundaries in UHFUS sequences are segmented in real-time using local phase analysis and an efficient level-set based approach, and vessel motion is tracked using an Extended Kalman Filter (EKF). Corneal and limbal tissue interfaces are segmented using classical and learning-based methods. The classic approach involves local phase analysis, combined with a fast Graphics Processor Unit (GPU) based surface detection algorithm. The learning-based approach takes advantage of modern deep learning methods to identify and segment limbal and corneal tissue interfaces.

In the thesis, we plan to extend our approaches to analyze quantitative measurements over time. We plan to extract and visualize cross-sectional images of the Palisades of Vogt (POV) from OCT volumes acquired by different OCT scanners with different settings to quantify the POV structural changes over time. We will explore deep learning approaches to scanner-agnostic corneal tissue interface segmentation. Finally, we plan to extend our algorithms to develop the first computer vision system for measuring intimal vessel wall thickness, and prospectively engage human subjects with our system.

More Information

Thesis Committee Members:
John Galeotti, Chair
George Stetten
Ruslan Salakhutdinov
Ajay Gopinath, Abbott