https://github.com/codingfisch/mriaug

Augment 3D images (e.g. MRIs) in PyTorch 🔥

Science Score: 13.0%

This score indicates how likely this project is to be science-related based on various indicators:

○
CITATION.cff file
✓
codemeta.json file
Found codemeta.json file
○
.zenodo.json file
○
DOI references
○
Academic publication links
○
Academic email domains
○
Institutional organization owner
○
JOSS paper metadata
○
Scientific vocabulary similarity
Low similarity (11.3%) to scientific vocabulary

Keywords

3d-images augment-images augmentation deep-learning gpu imaging machine-learning medical-imaging python pytorch

Last synced: 5 months ago · JSON representation

Repository

Augment 3D images (e.g. MRIs) in PyTorch 🔥

Basic Info

Host: GitHub
Owner: codingfisch
License: mit
Language: Python
Default Branch: main
Homepage:
Size: 4.92 MB

Statistics

Stars: 1
Watchers: 1
Forks: 0
Open Issues: 0
Releases: 0

Topics

3d-images augment-images augmentation deep-learning gpu imaging machine-learning medical-imaging python pytorch

Created about 1 year ago · Last pushed about 1 year ago

Metadata Files

Readme License

mriaug

pip install mriaug to use a 3D image library that is ~50x faster and simpler than torchio by

only using PyTorch → full GPU(+autograd) support 🔥
being tiny: ~200 lines of code → no room for bugs 🐛

while offering ~20 different augmentations (incl. MRI-specific operations) 🩻

👶 Normal users should use mriaug via niftiai, a deep learning framework for 3D images, since it - provides aug_transforms3d: A convenient function that compiles all mriaugmentations! - simplifies all the code needed for data loading, training, visualization...check it out here!

👴 Experienced users can build their own framework upon mriaug (use niftiai/augment.py as a cheat sheet)

Usage 💡

Let's create a 3D image tensor (with additional batch and channel dimension) and apply flip3d ```python import torch from mriaug import flip3d

shape = (1, 1, 4, 4, 4) x = torch.linspace(0, 1, 4*3).view(shape) xflipped = flip3d(x) print(x[..., 0, 0]) # tensor([[[0.0000, 0.2540, 0.5079, 0.7619]]]) print(xflipped[..., 0, 0]) # tensor([[[0.7619, 0.5079, 0.2540, 0.0000]]]) ``` Explore the gallery to understand the usage and effect of all ~20 augmentations!

Speed 💨

The popular libraries torchio and MONAI (utilizes torchio) often use ITK (CPU only) like this

PyTorch tensor → NumPy array → NiBabel image → ITK operation (C/C++) → NumPy array → PyTorch tensor

to augment a PyTorch tensor 😬 That's complicated and does not use the (for neural net training needed) GPU 🐌

Instead, mriaug directly uses PyTorch (CPU & GPU support) resulting in - ~50x fewer lines of code: torchio: ~10,000 LOC, mriaug: ~200 LOC 🤓 - ~50x speedup on GPU 🔥 based on the table below (run speed.py to reproduce) 💨

Click here, to see runtimes on a 256³ image in seconds (on AMD 5950X CPU and NVIDIA RTX 3090 GPU)

| Transformation | `torchio` | `mriaug` on CPU | `mriaug` on GPU | Speedup vs. torchio | |----------------|-----------|-----------------|-----------------|---------------------| | Flip | 0.014 | 0.012 | 0.002 | **7.5x** | | Affine | 0.297 | 0.608 | 0.011 | **27.9x** | | Warp | 0.951 | 0.850 | 0.009 | **103.3x** | | Bias Field | 3.258 | 0.081 | 0.002 | **1813.0x** | | Noise | 0.117 | 0.105 | 0.001 | **230.4x** | | Downsample | 0.282 | 0.013 | 0.000 | **592.3x** | | Ghosting | 0.241 | 0.170 | 0.003 | **78.3x** | | Spike | 0.265 | 0.172 | 0.003 | **88.8x** | | Motion | 0.696 | 0.540 | 0.009 | **78.6x** |

Gallery 🧠

Let's load an example 3D image x, show it with niftiview (used to create all images below)

define some arguments

python size = (160, 196, 160) zoom = torch.tensor([[-.2, 0, 0]]) rotate = torch.tensor([[0, .1, 0]]) translate = torch.tensor([[0, 0, .2]]) shear = torch.tensor([[0, .05, 0]])

and run all augmentations (see runall.py):