|
" Analysis of Optical Fiber Speckle Patterns for Detection of IVUS Catheter Tip in 3D Space: "
Razmyar, Soroush
Mostafavi, M. Taghi
| Document Type
|
:
|
Latin Dissertation
|
| Language of Document
|
:
|
English
|
| Record Number
|
:
|
1106640
|
| Doc. No
|
:
|
TLpq2407577649
|
| Main Entry
|
:
|
Mostafavi, M. Taghi
|
|
|
:
|
Razmyar, Soroush
|
| Title & Author
|
:
|
Analysis of Optical Fiber Speckle Patterns for Detection of IVUS Catheter Tip in 3D Space:\ Razmyar, SoroushMostafavi, M. Taghi
|
| College
|
:
|
The University of North Carolina at Charlotte
|
| Date
|
:
|
2020
|
| student score
|
:
|
2020
|
| Degree
|
:
|
Ph.D.
|
| Page No
|
:
|
106
|
| Abstract
|
:
|
This research study presents the architectural design and computational framework for an intelligent tracking sensor constructed from a multimode fiber optic. As laser light travels through an inhomogeneous medium, such as multimode fiber, the random interactions between light rays generate a circular output pattern commonly referred to as speckle patterns. Speckle patterns are highly responsive to the variation in the physical status of a multimode fiber. As a multimode fiber deforms, analysis of speckle pattern variations provides information about the external perturbations causing the deformation. This study presents a novel algorithm for calculating 3D transformations from a series of speckle patterns, which is modeled in three tiers. In the first tier, we have performed a series of experiments to demonstrate, in a deforming multimode fiber, the structural variation of speckle patterns contains deterministic information. That also provides a systematic approach for measuring the deformation parameters of a multimode fiber using a convolutional neural network. Second, we have studied the oscillating behavior of multimode fiber as a function of its length to find the relationship between the sensor's heading direction and the deformation of its sensing fiber tip. By utilizing a Long Short-Term Memory model, we have demonstrated that long-term dependencies between the deformation parameters provide a stable and reliable indication of the intelligent sensor's direction. In the end, we have utilized these findings to develop a novel computational framework for the intelligent sensor. This computational framework includes a pipeline of deep learning models to extract features from a sequence of speckle patterns, and a motion model to estimate the trajectory of the sensor from the extracted features.
|
| Subject
|
:
|
Biomedical engineering
|
|
|
:
|
Computer science
|
|
|
:
|
Optics
|
| |