Quantum SegNet: Brain Analysis With AI and quantum computing

Developer Spotlight: Analysing Brain Images with Intel AI Software and Intel Tiber AI Cloud

Quantum SegNet

In the fast developing disciplines of artificial intelligence and quantum computing, neural networks are increasingly needed to improve medical imaging. Brain scans must be carefully segmented and assessed to diagnose tumours, strokes, and neurodegenerative diseases. However, the accuracy and effectiveness needed for early detection and treatment planning are frequently lacking in older models. To tackle this problem, a group of Intel Student Ambassadors created Quantum SegNet, a quantum-enhanced neural network for brain image interpretation, during the 2024 Intel Student Ambassador Fall Hackathon.

The Intel Tiber AI Cloud‘s AI tools and framework optimisations are utilised by the Quantun SegNet solution. This blog will provide you with a project summary.

Concerning the Quantum SegNet Initiative

Quantum SegNet is an artificial intelligence (AI) solution that uses quantum neural networks (QNNs) to improve brain image analysis. By combining AI-driven picture segmentation models with quantum computing techniques, the initiative greatly increases the precision and effectiveness of medical diagnostics.

A Quanvolution Filter, a revolutionary quantum feature extraction method that improves on conventional convolutional layers, is used in the model’s Quantum-Enhanced Segmentation.

An Overview of Model Architecture

There are several processing levels in the deep learning framework:

• Feature extraction improves image clarity by applying a filtering method influenced by quantum mechanics.
• Brain structure categorisation is improved with segmentation processing.
• Inference Acceleration: To maximise real-time picture segmentation, use the Intel Distribution of OpenVINO Toolkit.

Using the OpenVINO Toolkit for Model Optimisation and Deployment

The PyTorch model was converted by the team to ONNX and then to OpenVINO for deployment. In order to ensure that the model could operate effectively on Intel’s architecture, minimising latency and optimising inference speed, the conversion process was essential. The conversion looked like this:

 import torch
 import onnx
 from openvino.runtime import Core
 
# Convert PyTorch model to ONNX
 model = torch.load('model.pth')
 torch.onnx.export(model, dummy_input, "model.onnx", opset_version=11)
 
# Load ONNX model and optimize with OpenVINO
 core = Core()
 model_onnx = core.read_model("model.onnx")
 compiled_model = core.compile_model(model_onnx, "CPU")

Real-time brain picture analysis is now possible because to the OpenVINO Toolkit’s ability to optimise and compress the model, which greatly reduced processing time. This project phase necessitated extensive testing and adjustment.

Utilised Intel Tools and Technologies

To improve their model, the Quantum SegNet team made use of Intel’s AI and quantum computing ecosystem:

• The Intel Tiber AI Cloud offered the processing power needed to effectively run quantum-enhanced models.
• Real-time inference optimisation for quicker segmentation was made possible by the Intel Distribution of the OpenVINO Toolkit.
• Model deployment and training were aided by the Intel Extensions for PyTorch and Tensor Flow.

What Comes Next?

AI-powered medical imaging could undergo a revolution with Quantum SegNet. The group intends to investigate practical uses in medical facilities and research institutes while also refining the concept.

Intel use its oneAPI-powered developer resources to get AI projects off to a quick start. To take use of the optimised software portfolio on the newest accelerated hardware for high-performance AI and accelerated computing, sign up for Intel Tiber AI Cloud.

Quantum Neural Networks for the Analysis of Brain Images

In order to increase performance and efficiency, this study investigates the application of Quantum Neural Networks (QNNs) for brain image analysis, utilising Intel hardware optimisations. The OpenVINO Toolkit, Intel Gaudi (HPU), and the Intel Extension for PyTorch are the main tools used in this project.

Overview of the Project

This project’s objective is to use Quantum Neural Networks (QNNs) to analyse brain pictures.

Installation

Step 1: create a virtual environment.

It advise managing dependencies in a virtual environment:

python -m venv qnn-env
source qnn-env/bin/activate  # On Windows use `qnn-env\Scripts\activate`

Step 2: Install the necessary dependencies

Install the required packages with these commands:

pip install numpy opencv-python matplotlib torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu
pip install intel-extension-for-pytorch openvino
pip install torchquantum
pip install qiskit==0.43.1 qiskit-terra==0.24.1 qiskit-aer==0.12.0 qiskit-ibm-provider==0.6.1 qiskit-ibm-runtime==0.10.0

Step 3: Confirm the Installation

Verify that every package is installed properly:

pip list

Usage

To manage the project, take these actions:

Make a clone of the repository:

git clone https://github.com/CPD9/Quantum-Neural-Networks-QNNs-for-Brain-Image-Analysis.git
cd Quantum-Neural-Networks-QNNs-for-Brain-Image-Analysis
. QuantumSegnets.ipynb

Turn on your online environment:

source qnn-env/bin/activate

Launch the notebook in Jupyter:

jupyter notebook QuantumSegnet.ipynb

Model Optimization

Intel optimisations are used in this project to improve performance:

PyTorch Intel Extension:

import intel_extension_for_pytorch as ipex
model = ipex.optimize(model)

Inference with OpenVINO:

from openvino.runtime import Core

def load_openvino_model(model_path):
    ie = Core()
    model = ie.read_model(model=model_path)
    compiled_model = ie.compile_model(model=model, device_name="CPU")
    return compiled_model