Assignment Task

CTEC3451 - Computer Networks and Security Module Assignment

Introduction

Detecting lung blood clots is crucial for clinicians, as early identification and intervention can prevent potentially fatal outcomes. However, this task poses challenges, particularly in diagnosing pulmonary embolism (a blood clot in the lungs) during its early stages. The current gold standard for detecting lung blood clots is through CT scans, which are costly and expose patients to radiation.

Recent advancements in deep learning offer promise for automating the detection of various medical conditions. In this proposal, the aim is to leverage deep learning techniques for the detection of lung blood clots.

Background

Blood comprises plasma, red blood cells (RBC), white blood cells (WBC), and platelets. Plasma, constituting about 55% of blood volume, is a clear, colorless fluid containing water and other substances essential for various bodily functions such as oxygen transport, blood clotting, and hormonal regulation.

Blood clots are a leading cause of strokes and coronary heart attacks. When blood becomes trapped in sticky vessels, it can obstruct blood flow, leading to life-threatening conditions like strokes or heart attacks. Symptoms may include headaches, chest pain, numbness, rapid heartbeat, and shortness of breath.

Physicians employ various diagnostic techniques when suspecting a blood clot, including imaging tools like duplex ultrasound, MRI, venography, computed tomography scans, and D-Dimer tests. However, these methods vary in accuracy, cost, and availability.

Prior research has explored using neural network models, genetic algorithms, and artificial intelligence to detect blood clots in their early stages.

Objective

This project aims to apply deep learning for the detection of lung blood clots, potentially saving lives by identifying pulmonary embolism.

Specific Objectives

1. Conduct a comprehensive literature review to identify methods, algorithms, and approaches used in this study.
2. Label the data by experts.
3. Prepare training and test datasets.
4. Identify suitable deep learning algorithms.
5. Develop an optimal model for detecting lung blood clots.
6. Test and evaluate the performance of the proposed model.