Visual feedback for a semi-automatic micropipette aspiration of single biological cells
Date of Issue2017-08-28
School of Electrical and Electronic Engineering
This project aims to provide visual feedback to a system to mechanically characterize single white blood cells by micropipette aspiration. In this project, image processing algorithms were developed to detect individual cells and the micropipette, to provide feedback to a system to control linear actuators and position a micropipette to approach a cell, form a seal and partially aspirate the cell. Circular Hough Transform was used to detect cells, Normalized Cross-Correlation was used to do template-match and detect the micropipette and Kalman Fiter was used to track the movements. Under a bright field microscope, monocytes (a type of white blood cell) are irregularly shaped, furthermore, these cells may cluster together. Therefore, the circular Hough transform failed to accurately detect the cells. An improved algorithm which can find clusters and segment cells was developed by using Mathematical Morphology and K-means clustering. At the end of this project, the visual feedback algorithms perform well, and can potentially be used as a fundament to support the vision-based control of micropipette aspiration as well as the study of cell structures and relevant disease progressions.
DRNTU::Engineering::Electrical and electronic engineering