Date of Award

2021-12-01

Degree Name

Master of Science

Department

Engineering

Advisor(s)

Tzu-Liang B. Tseng

Abstract

Pneumonia is a viral or fungal illness that spreads to the lungs of the human body, causing fluid to accumulate in the lungs' air sacs. Millions of people are affected by this disease each year. One of the most common radiological diagnostics for diagnosing and screening this kind of sickness is a chest X-ray. The most commonly available radiological test for diagnosing and screening this kind of illness is a chest X-ray. An inaccurate diagnosis, especially over-diagnosis and under-diagnosis, is a common issue in the medical sector. As another issue, human-assisted diagnosis has limitations like the availability of an expert, cost, etc. To address these issues, researchers focused on deep learning approaches for improved diagnostic outcomes. An application with an automatic system to detect pneumonia is developed in these aids in overcoming diagnosing errors and treating the patient. As discussed above, the authors developed a two-step methodology in this research. In the first step, various models are utilized as the neural network model to be trained with different software for pneumonia detection for chest X-ray images. These trained models are converted to a frozen graph and injected into the Unity software tool by utilizing a C# script that creates a bounding box to overlay the detected region. These processes are mitigated by building and deploying the application from unity to the head-mounted device known as Microsoft HoloLens gen 1. This study also considers the research gap and proposes integrating Augmented Reality (AR) technology and Deep Learning (DL) to handle this issue.

Language

en

Provenance

Received from ProQuest

File Size

47 p.

File Format

application/pdf

Rights Holder

Jeevarathinam Senthilkumar

Share

COinS