Med-Vault

Decentralized Electronic Medical Record

Team Members

• Dhruv Verma
• Abhinav Kumar
• Hriday Ahuja

Getting Started

Step 1: Install The Required Softwares

• Install Ganache from the following link [https://trufflesuite.com/ganache/]
• Install Metamask browser extension [https://chrome.google.com/webstore/detail/metamask/nkbihfbeogaeaoehlefnkodbefgpgknn]
• Download and Install Python 3.7+ from here
• Download and Install Node from here. It will also install node package manager

Step 2: Clone the repository and cd into the project directory

• Input the subsequent instructions within the command terminal

  git clone https://github.com/AbhiSinha08/med-vault
  

  cd med-vault
  

Step 2: Setting up Local Blockhain network

• Open Ganache and click on 'Quick Setup'

• Change network on Metamask and import accounts from Ganache

• Compile and deploy the smart contract with truffle

• Input the subsequent instructions within the command terminal

  npm install -g truffle
  

  cd truffle
  

  truffle compile
  

  truffle deploy
  

Step 3: Executing the frontend

• Input the subsequent instructions within the command terminal

  cd ../client
  

  npm install
  

  npm start
  

By default, the client runs on port 3000 on localhost.
Head over to your local host or paste this URL in your browser http://localhost:3000 to test it out.

Brain MRI Segmentation Model

We've developed a model that's designed to perform segmentation on brain MRI images. This means that it's capable of identifying and highlighting specific regions of interest within these images. The architecture we chose for this task is called U-Net, which is the ideal choice for medical image segmentation tasks. When you provide an MRI image as input, the U-Net model processes it through a series of layers that progressively learn to recognize different features within the image.

Input Image to the Model

Output Masked Image

Once the model processes the input image, it generates an output mask that highlights the segmented regions. The output mask assigns a value to each pixel, indicating its class or category. This way, you can clearly see which parts of the brain the model has identified as belonging to the region of interest.

• Binary Accuracy: 0.9108
• Intersection Over Union: 0.0492
• dice Coefficients: 0.0897

Accuracy Graph for the Segmentation Model

The U-Net architecture has proven to be quite effective for tasks like medical image segmentation, where precise identification of specific structures is crucial. It's capable of producing accurate segmentation results, which can be extremely valuable for medical professionals in diagnosing and treating various brain conditions. In a nutshell, my brain MRI Segmentation model based on the U-Net architecture takes MRI images as input, processes them through a specialized neural network, and produces a masked output that highlights the segmented regions of interest within the brain images.