search

yeyuxmf / retinal_vascular_segmentation

Retinal vascular segmentation，Transformer，CNN， Pytorch, Python
5 stars 1 forks source link

readme

retinal_vascular_segmentation

Retinal vascular segmentation，Transformer，CNN， Pytorch, Python， Date:2023.09.01 Achieve state-of-the-art results.

Introduce

  1. This is a retinal vessel segmentation, although its accuracy surpasses all current papers on the HRF, DRIVE, STARE, and CHASEDB1 datasets, it is not innovative and is a semi-finished paper.

  2. The paper only wrote the experimental part and recorded some experimental data in the Word document in the Comparative Literature folder.

    The previous code had some issues, but now it has been re-submitted after being trained and verified for correctness. (以前的代码有点问题，现已经过训练验证正确性后从新提交。)

Method

Reference 3D medical image segmentation model:

  1. Shaker A, Maaz M, Rasheed H, et al. UNETR++: delving into efficient and accurate 3D medical image segmentation[J]. arXiv preprint arXiv:2212.04497, 2022.
    企业微信截图_17018485728081

This is the model structure diagram in the UNETR++paper. I directly modified it from 3D to 2D image application.

Train

python main.py Note: The batch_size=4 during training cannot be easily changed, as it will affect the accuracy.

Test

python test.py and python test_HRF.py HRF is a high-resolution dataset with different testing methods.

Experimental data

1

2

3

4

License permissions

  1. Please comply with the relevant medical image data usage license for usage permissions.
  2. Anyone can continue mathematical research on this.