A Deep Learning Based approach for Detecting CT Diseases

Today in this age we have much more technologies as compared before, AI is developing rapidly, we have different algorithms different machinery to work on by the use of them we can handle many complexities which we are facing especially the areas which are not so developed. In our daily life we hand

Project Title

A Deep Learning Based approach for Detecting CT Diseases

Project Area of Specialization

Artificial Intelligence

Project Summary

Today in this age we have much more technologies as compared before, AI is developing rapidly, we have different algorithms different machinery to work on by the use of them we can handle many complexities which we are facing especially the areas which are not so developed. In our daily life we handle or go through from many viruses from which some can take us to death and we can’t even identify what happened. Such viruses or disease like influence (fastly spread), new coronavirus, respiratory syncytial viruses. COVID-19 and influenza viruses have a similar disease presentation. However we can’t stop working but to stop problems.

Our project aims to propose a high-speed and accurate fully-automated method to detect viruses from the patient's chest CT scan images. There are several methods for the definitive diagnosis of viruses, including reverse transcriptase-polymerase chain reaction (RT-PCR), Isothermal nucleic amplification test, Antibody test, Serology tests, and medical imaging. RT-PCR is the primary method of diagnosing COVID-19 and many viral diseases. In most patients with viruses, infections are found in the lungs of people. Some patients with early-onset COVID-19 symptoms were found to have new coronavirus infections on their CT scans. At the same time, their RT-PCR test results were negative, then both tests were repeated several days later, and RT-PCR confirmed the CT scan’s diagnostic results. Although medical imaging is not recommended for the definitive diagnosis of such harmful viruses.

The advantage of using medical imaging is the ability to visualize viral infections by machine vision. Machine vision has many different methods, one of the best of which is deep learning. Machine vision and deep learning have many applications in medicine, agriculture, economics, etc., which have eliminated human errors and created automation in various fields.

The use of machine vision and deep learning is one of the best ways to diagnose tumors and infections caused by various diseases. This method has been used for various medical images, such as segmentation of lesions in the brain and skin, Applications to Breast Lesions, and Pulmonary Nodules, sperm detection and tracking, and state-of-the-art bone suppression in x-rays images.

In this project, we introduce a fully-automated method for detecting various virus cases from the output files (images) of the lung HRCT scan device or X-rays. This system does not need any medical expert for system configuration and takes all the CT scans of a patient and clarifies if that patient is infected with such viruses or not.

Project Objectives

1. Devlop the novel approach for automizing the triaging process for CT-scan images.
2. A high-speed and accurate fully-automated method to detect viruses from the patient's chest CT scan images.
3. Formulate the local dataset from Quetta hospitals such as Civil Hospital, and CMH.
4. To Digitlizaed the existing reports and train the model for the local dataset.
5. Various approaches to compare local data vs public available data for classification and segmentation.

Project Implementation Method

This project will mainly focus on three sections. Which are as follows:

Fast Detecting Algorithm:

The deep learning approach has shown more exact outcomes than RTX in the COVID-19 pandemic and is accepted to be the quickest calculation of the current period. Hence using it in the medical sector will be much easier for Hospitals and especially the diseases which cause a human loss in a few days like Influenza, Pneumonia and Covid-19

Implementation of Local dataset collected:

Many countries involved artificial intelligence in their medical sector but still, Balochistan is somehow backward at this, we have better technologies today, and we get to know about these modern technologies in the universities of Balochistan so why not implement them here, where they are needed the most. So our aim is to implement the triage process and get data from the hospitals of Balochistan and do apply all the techniques to them for more accurate results. We explored this and settled on it as the most ideal decision. We extricated information from Quetta Hospitals. Brokedown it with high-definition digital cameras for image scanning, and scanners. Analyzed it. Imported pictures in google collaborator utilizing Keras. Applied calculations on it, and tangled them utilizing the u-net model. Brokedown the outcomes in like manner.

Hardware Requirements

For data collection, we go through Civil Hospital and CMH traveled many times. Get the data required. Scanned or photograph it with high-definition digital cameras and scanners for more accurate results. Imported the images to google collaborator and applied calculations on them using python. Our project will help the medical sector of Balochistan in many ways.

1. Google Collaborator / Personal computer with GPU capability 

2. High definition Digital Camera (for converting the existing record)

3. Xray, CT images Scanners

4. Travelling Expenses to local hospitals

Benefits of the Project

By considering the approach of deep learning we can tickle with such kinds of harmful diseases in the early stages. In ancient days people deal with their problems through their old techniques which were said by their elders to them. Now in the current era, we live in the modern period. We live in a technical and intelligent world. Diseases that are new can be dealt with with modern technologies. Some diseases which have no symptoms are difficult to handle but can be identified by deep learning algorithms of artificial intelligence more accurately on time. We wanted to use modern technologies in the area where we live. We live here, we are getting an education from here, it is our responsibility to take our area in our research and our area get benefited by us, it would be much better when we will help our people, save the lives of our people, for this we are taking our knowledge to fulfill this lack of technical terms which are not present in our area in Balochistan, We will deal with the medical sector with some modern technologies involved with it. We not only help the medical sector but will be less time consumed and more accurate and would be a new invention in our hospital's lives will be saved from such viral diseases, and human loss will be consumed. Our area Balochistan will be benefited from this project in many ways.
 

Technical Details of Final Deliverable

Many countries involved artificial intelligence in their medical sector but still Balochistan is somehow backward at this, we have better technologies today. We extricated information from Quetta Hospitals. Broke down it using high defination digital cameras and scanners. Imported pictures in Google collabrator utilizing keras. Applied calculations on it, tangled them utilizing u-net model.

Segmentation is the primary task for U-net models. The goal of segmentation tasks is to outline and separate different objects in an image to classify different objects rather than classifying the whole image as our main target in our sector is the time consumption, speed, accuracy are the main focuses of the part so as it is relatively going to support our objectives as it classify different objects rather than classifying the whole image by which time would be reduced and as using the residual net  our speed and accuracy of the system would become high.

The basic structure of a u-net architecture consists of two paths. The first path is the contracting path, also provides classification information. The second is an expansion path, also known as decoder or the synthesis path, consisting of up-convolutions and concatenations with features from the contracting path. This expansion allows the network to learn localized classification information which increase the resolution of the system and the resulting network is almost symmetrical, giving it a u-net shape.

However, classification networks fail to provide pixel-level context information which is much needed in medical image analysis.

For implementation part of our project we are using Google collabrator using keras in Python. Using high defination digital cameras and scanners.

The outcome of the project is to design an API for CT-Scan images that can predict any anomaly from CT-Images to reduce the diagnosis time and start the treatment right away to save human life and doctor's time.

Final Deliverable of the Project

HW/SW integrated system

Core Industry

Medical

Other Industries

IT , Health

Core Technology

Artificial Intelligence(AI)

Other Technologies

Artificial Intelligence(AI), Big Data

Sustainable Development Goals

Good Health and Well-Being for People, Industry, Innovation and Infrastructure

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com