matthewcheng222 / lapmmirf

Multi-Modal Image Registration Framework using Laplacian Pyramid

LapMMIRF : Multi-Modal Image Registration Framework using Laplacian Pyramid

Multi-Modal Medical Image Registration Framework using Laplacian Pyramid for precise image alignment and analysis.

Prerequisites

• Python 3.9.16+
• Pytorch 1.7.0 - 2.0.0
• NumPy
• NiBabel

This code has been tested on Google Colaboraty with Pytorch 2.0.0+cu118 and NVIDIA A100-SXM4-40GB GPU.

Instructions

All required imports and functions are contained in the LapMMIRF.ipynb file, running all cells would do the following.

• download necessary datasets from the IXI dataset
• install and import standard and third part libraries
• train the model using the specified hyperparameters
• save the trained model to the output folder as model.pth
• test the model with random images and saving the registered images

Dataset

All fixed and moving images in this directory is obtained from the IXI Dataset.

• Fixed Image: PD MRI Scans
• Moving Image: T2-Weighted MRI Scans

Keywords

Keywords: Deep Learning, Multi-Modal, Image Registration, Large Deformation, Laplacian Pyramid Networks, Convolutional Neural Networks