Table of contents
- foreword
- installation steps
- MONAI-based MedMNIST dataset classification task
- import dependencies
- download data
- read image information
- MONAI transforms
- Define Dataset, Network and Optimizer
- train
- test
- Summarize
- reference link
foreword
Recently, I was reading a paper related to the direction of medical images on CVPR 2023, and found that the Label-Free Liver Tumor Segmentation paper used the MONAI framework. In some journal papers on medical images that I have paid attention to before, MONAI also appeared, and the previous tutor had recommended it, so I learned about it. After a simple search, I found that there are no Chinese tutorials about MONAI on the Internet. There will be a series of very detailed tutorials about MONAI, which will be published on GiantPandaCV.
MONAI has three main warehouses, MONAI Core, MONAI Label and MONAI Deploy (SDK), which are used for model training, medical image labeling and model deployment respectively, and we will introduce them one by one later. The addresses are as follows:
- https://github.com/Project-MONAI/MONAI
- https://github.com/Project-MONAI/MONAILabel
- https://docs.monai.io/projects/monai-deploy-app-sdk/en/stable/index.html
Regarding the torch-based MONAI Core, which is what we focus on in the first part, it mainly has the following advantages:
- Flexible multi-dimensional medical image data preprocessing function &