Implementation of MedSegDiff in Pytorch - SOTA medical segmentation out of Baidu using DDPM and enhanced conditioning on the feature level, with filtering of features in fourier space.
-
StabilityAI for the generous sponsorship, as well as my other sponsors out there
-
Isamu and Daniel for adding a training script for a skin lesion dataset!
$ pip install med-seg-diff-pytorch
import torch
from med_seg_diff_pytorch import Unet, MedSegDiff
model = Unet(
dim = 64,
image_size = 128,
mask_channels = 1, # segmentation has 1 channel
input_img_channels = 3, # input images have 3 channels
dim_mults = (1, 2, 4, 8)
)
diffusion = MedSegDiff(
model,
timesteps = 1000
).cuda()
segmented_imgs = torch.rand(8, 1, 128, 128) # inputs are normalized from 0 to 1
input_imgs = torch.rand(8, 3, 128, 128)
loss = diffusion(segmented_imgs, input_imgs)
loss.backward()
# after a lot of training
pred = diffusion.sample(input_imgs) # pass in your unsegmented images
pred.shape # predicted segmented images - (8, 3, 128, 128)
Command to run
accelerate launch driver.py --mask_channels=1 --input_img_channels=3 --image_size=64 --data_path='./data' --dim=64 --epochs=100 --batch_size=1 --scale_lr --gradient_accumulation_steps=4
If you want to add in self condition where we condition with the mask we have so far, do --self_condition
- some basic training code, with Trainer taking in custom dataset tailored for medical image formats - thanks to @isamu-isozaki
- full blown transformer of any depth in the middle, as done in simple diffusion
@article{Wu2022MedSegDiffMI,
title = {MedSegDiff: Medical Image Segmentation with Diffusion Probabilistic Model},
author = {Junde Wu and Huihui Fang and Yu Zhang and Yehui Yang and Yanwu Xu},
journal = {ArXiv},
year = {2022},
volume = {abs/2211.00611}
}
@inproceedings{Hoogeboom2023simpleDE,
title = {simple diffusion: End-to-end diffusion for high resolution images},
author = {Emiel Hoogeboom and Jonathan Heek and Tim Salimans},
year = {2023}
}