search

jackjacktang / LG-Net

Official repository for MICCAI2021-LG_Net
7 stars 1 forks source link

Question on the data split used #1

Open jackhu-bme opened 4 months ago

jackhu-bme commented 4 months ago

We aim to use the same data split as yours for fair comparison. Please let ue know what your data split is in the paper. "We used only T1-weighted MR brain images in our experiment, however, our framework can be extended to fit other imaging sequences. In order to perform volumetric evaluation, we removed all the cases that failed quality control after automatic tissue segmentation, focusing our analysis on 529 cases. 449 images were randomly selected for the training (including 9 images for evaluation), totalling 42,480 pseudo 3D slices." We need to know what 449 images your paper chose, and what is your 440 training images, 9 validation images. Just need the name of the images is sufficient. This helps us to fairly compare with your results. (just the MAE, MSE, PSNR, nothing related to segmentation). If this can not be known, we may fail to compare ours with your method.

jackjacktang commented 4 months ago

Hi Jack Hu,

Thank you for your interest in our work. Please kindly find the attached file for our experiment split and let us know if you have any further questions!

Regards, Zihao (Jack)

data_split_info.txt

jackhu-bme commented 4 months ago

wow!!!! That really helps!!! Thanks a LOT!!!