Date of Award

1-1-2021

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Biomedical Engineering

First Advisor

Jeremiah Neubert

Abstract

The purpose of the research work was to develop image processing algorithms for automatic and objective assessment of psoriasis to aid in more effective management and treatment of patient symptoms. For this purpose, multiple classical machine learning-based unsupervised approaches are proposed which achieve a maximum average accuracy of ~94% on a dataset of 45 images of various resolutions. This accuracy improves performance by 14% in comparison to existing approaches reported in the literature on the same dataset of images. An alternative deep learning neural network-based approach was proposed that achieves an average accuracy of ~87% for the same set of images with minimal pre and post-processing.

Additionally, a mechanism to classify images into red, red + scaling, and scaling classes is proposed based on the color profile of the lesion that provides an objective way of measuring the severity of the redness of the lesion and the scaling. Apart from the model development, a survey of existing mobile apps was carried out to determine the need for this application along with the technological gaps. Based on this survey and the interviews with various stakeholders associated with the psoriasis disease treatment, a pathway to commercialization is proposed by developing a complete business model.

Download

Share

COinS