Web Application for Intracranial Hemorrhage Detection

The objective of this project is to create a web tool for neuro-specialists to assist them in brain hemorrhage detection and localization task. An abstract of this project has been submitted to the Society for Imaging Informatics in Medicine 2019.

Table of Contents

• Directory Layout
• How to Run
  • Requirements
  • Steps
• Sample DICOMS
• Demo
• Courtesy
• Contributors
• Supervisors

Directory layout

├── app.py
├── demo
├── Sample DICOM Images
├── gradcam
├── inference_rsna.py
├── model
├── models
│   └── fold0_ep3.pt
├── README.md
├── requirements.txt
├── static
├── templates
├── test.py
├── uploads
└── utils.py

Requirements

• Python packages: Available in requirements.txt
• ngrok

Steps

• Set Up ngrok:

  • Open an account using your mail address here
  • Find your authtoken here.
  • Make ngrok executable on your system and run the following command:

  $ ./ngrok authtoken <YOUR_AUTH_TOKEN>
  

• Running Web server:

  • Go to Intracranial_Hemorrhage_Web_App directory.
  • Run python app.py. The web server will start running on localhost:9999
  • Open another terminal and run ngrok http 9999. After a couple of seconds, it will generate a texts that will look like this

ngrok by @inconshreveable                                       (Ctrl+C to quit)
                                                                                
Session Status                online                                            
Account                       <your_name> (Plan: Free)                               
Update                        update available (version 2.3.35, Ctrl-U to update
Version                       2.3.34                                            
Region                        United States (us)                                
Web Interface                 http://127.0.0.1:4040                             
Forwarding                    http://********.ngrok.io -> http://localhost:9999 
Forwarding                    https://********.ngrok.io -> http://localhost:9999
                                                                                
Connections                   ttl     opn     rt1     rt5     p50     p90       
                              243     0       0.00    0.00    0.01    3.81    

• Copy the link after Forwarding (http://********.ngrok.io). Now anyone can access the web server from this link.

Sample DICOMS

Sample DICOM Images directory contains 12 images. Their labels are available in the Ground_Truth.csv files.

Demo

• Step step tutorial for using the web application
• Demo Video

Courtesy

• 11th place solution of the RSNA Hemorrhage Detection Competition.
• Grad-CAM ++ Pytorch

Contributors

• Tahsin Mostafiz
• Shajib Ghosh

Supervisors

• Dr. Taufiq Hasan
• Dr. Paul Naggy