sdc17/MEPDNet

[ICIP 2021] Multi-Encoder Parse-Decoder Network for Sequential Medical Image Segmentation.

Multi-Encoder Parse-Decoder Network for Sequential Medical Image Segmentation (MEPDNet)

Official implementation of Multi-Encoder Parse-Decoder Network for Sequential Medical Image Segmentation.

Supported Models

• MEPDNet(Ours)
• SegNet
• DeepLabv3+
• U-Net
• Attention U-Net
• R2U-Net
• Attention R2U-Net
• ScSE U-Net
• CE-Net
• UNet++

Training Script

python run.py --model $MODEL_NAME --mode train -l $LR -b $BATCH_SIZE -e $EPOCHS --gpu-id $GPU_ID

optional arguments:
  -h, --help            show this help message and exit
  --model MODEL
  --mode {train,test,use}
  -e EPOCHS, --epochs EPOCHS
                        Number of epochs
  -b BATCH_SIZE, --batch-size BATCH_SIZE
                        Batch size
  -l LR, --learning-rate LR
                        Learning rate

For example, to train MEPDNet:

python run.py --model mepdnet --mode train -l 0.00008 -b 2 -e 100 --gpu-id 0 1

Training scripts for other models are in train.sh.

Evaluation Script

python run.py --model $MODEL_NAME --mode test --state $MODEL_ID -b $BATCH_SIZE --gpu-ids $GPU_ID

optional arguments:
  -h, --help            show this help message and exit
  --model MODEL
  --mode {train,test,use}
  --gpu-ids GPU_IDS [GPU_IDS ...]
  --state STATE
  -b BATCH_SIZE, --batch-size BATCH_SIZE
                        Batch size

For example, to evaluate MEPDNet:

python run.py --model mepdnet --mode test --state 70 -b 4 --gpu-ids 0 1

Evaluation scripts for other models are in test.sh.

Cite

If you find this work useful, please consider citing the corresponding paper:

@inproceedings{shi2021multi,
  title={Multi-encoder parse-decoder network for sequential medical image segmentation},
  author={Shi, Dachuan and Liu, Ruiyang and Tao, Linmi and He, Zuoxiang and Huo, Li},
  booktitle={2021 IEEE international conference on image processing (ICIP)},
  pages={31--35},
  year={2021},
  organization={IEEE}
}