Detection of Microconidia in Microscopy Images of Fusarium oxysporum f. sp. cubense Using Neural networks

Date of Award

2021

Document Type

Thesis

Degree Name

Master of Science in Computer Science

Department

Information Systems & Computer Science

First Advisor

Patricia Angela R. Abu, PhD

Abstract

Fusarium oxysporum f. sp. cubense (Foc) is a soil-borne fungus and the causative agent of the deadly Fusarium wilt disease in banana plants. Left alone, the fungus is able to survive for years and infect multiple plants through the soil. External symptoms only manifest in late stages of infection, with the burning of plants being the only way to eradicate the fungus. It is imperative then that Foc be detected as soon as possible. To achieve this, the study endeavors to detect microconidia, a reproductive structure of the Foc species, in microscopy images of stained soil specimen under three microscopy configurations using convolutional neural networks (CNNs). First, features of microconidia useful in classification through CNNs will be identified. Then, four CNNs per CNN architecture will be developed: one classifying bright field (BF) images only, one classifying dark field (DF) images only, another classifying fluorescent images (FL) only, with the last classifying all images regardless of microscopy technique. Modelling will be followed by a performance comparison of CNN architectures (ResNet, VGGNet, and AlexNet) in terms of accuracy and prediction time on the test set, as well as ROI detection using Gradient-weighted class activation mapping (Grad-CAM). Lastly, two more CNNs classifying images with and without hyphae will be developed to examine the effect that the presence of hyphae has on the models. This study contributes towards the early detection of Foc, and is a step toward mitigating the threat it presents.

This document is currently not available here.

Share

COinS