Prognosis and Intervention for Burn Wounds Using Deep Learning

Burn injury is usually a severe case with high stakes for the cause of death. These types of critical cases needs an accurate prognosis and intervention. As per practice doctors using techniques such as straight-ruler method, aseptic film trimming method, and digital camera photography method ca

Project Title

Prognosis and Intervention for Burn Wounds Using Deep Learning

Project Area of Specialization

Artificial Intelligence

Project Summary

Burn injury is usually a severe case with high stakes for the cause of death. These types of critical cases needs an accurate prognosis and intervention. As per practice doctors using techniques such as straight-ruler method, aseptic film trimming method, and digital camera photography method can be accurate but are not able to be done over and over nor can be relatable, which means that there can be a difference in the analysis of burn and make the evaluation handicap. Therefore, the aim of this paper is to build such a solution that can aid the doctors for identifying the degrees of burn images. The first step was to have data, Which was collected from “Civil hospital burns centre”. Secondly, taking an ImageNet pre-trained CNN and do transfer learning for calculating crucial results, such as the percentage of the body which is burned(TBSA).Moreover, aiming to accurately segment and classify different burn depth.

Project Objectives

The purpose of this solution is to improve the overall process of reporting and resolving missing person cases. The application will be capable of answering all the major challenges.

Here is an overview of three key features pertaining to aforementioned problems

Register missing person cases and report them to the concerned authority. Cases will be reported according to the nature of the incident (suicide, kidnapping, burn, blast, etc.)

Using image processing technology to analyze the original face after a severe burn or any other injury that has left the face unrecognizable.

Using old photos to predict a missing person’s current appearance. This is extremely crucial in child kidnapping cases that remain unsolved for years, or in scenarios where the family is not able to provide a recent photo of an old person.

Project Implementation Method

For accurate early-stage burn depth diagnosis.First we will use dataset to data segment and label burns.second part minimal training we will try to show accurately discriminate burnt skin from the rest of an image.also calculate percent of the body which is burn(TBSA).for data set we will use online annotation tool that make easy for surgeons to label various burn on an image labeling based on superficial(S),superficial/deep partial thickness,full thickness(FT) and undebrided(U).architecture is based of a fully convolutional network we will use convolutional nets(like Alexnet,VGG,Googlenet) important thing is to take input on dimension Height x width x 3,and to produce output of shape height x width x classes to represent the segementation mask. We will use metrics to evaluate performance of the system.

Benefits of the Project

A lot of people from a lot of departments can benfit from this project. Our scope to help the society with the recent tecnology and with efficiency. Mostly, our focus is on the burned up people who are hard to be identified, so with this project people like police departments, burn centers, and dermatologists can make use out of it.

Technical Details of Final Deliverable

the aim of this is to build automated computer aided for identifying the degree of burn image patient need to be classified into 4 degrees of burn based on the skin’s thickness, the depth of burns and scalds, and some diagnosis related.

Final Deliverable of the Project

HW/SW integrated system

Type of Industry

IT , Medical

Technologies

Artificial Intelligence(AI)

Sustainable Development Goals

Good Health and Well-Being for People

Required Resources

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