synapse数据集预处理

[pandajok]

您好！非常感谢您在这篇论文中所做出的的贡献，能否分享一份Synapse数据集的预处理代码，我的邮箱：2914088431@qq.com，非常感谢！

[Dootmaan]

Hi @pandajok, we directly use the preprocessed Synapse dataset by contacting the authors of TransUnet so we do not have the processing code for Synapse. However, we have sent you an ACDC processing code.

[majixin-li]

Hi @pandajok, we directly use the preprocessed Synapse dataset by contacting the authors of TransUnet so we do not have the processing code for Synapse. However, we have sent you an ACDC processing code.

从transUnet下载的Synapse dataset是处理只好的数据集吧 我放到dataset目录下面，感觉不对啊 还差一些文件
下面是报错和目录结构：

[Dootmaan]

Hi @pandajok, we directly use the preprocessed Synapse dataset by contacting the authors of TransUnet so we do not have the processing code for Synapse. However, we have sent you an ACDC processing code.

从transUnet下载的Synapse dataset是处理只好的数据集吧 我放到dataset目录下面，感觉不对啊 还差一些文件 下面是报错和目录结构：

Hi @majixin-li. Thank you for raising this issue. We notice that the authors of TransUnet have released a new pack of their project and provided the preprocessed Synapse dataset directly (previously you will have to write them an email for the preprocessed dataset). However it seems like they forget to provide the 'list_Synapse' folder and the users have no clue what the 'list_dir' is in the parameters list (the third parameter) of class Synapse_dataset(Dataset):

class Synapse_dataset(Dataset):
    def __init__(self, base_dir, list_dir, split, transform=None):
        self.transform = transform  # using transform in torch!
        self.split = split
        self.sample_list = open(os.path.join(list_dir, self.split+'.txt')).readlines()
        self.data_dir = base_dir

The problem you encountered is also caused by the missing of 'lists_Synapse' folder which contains several text files describes the dataset split. You can provide us with you email address so we can send you the missing files, or you can directly contact the authors of TransUnet and ask them to upload the missing files. By the way, you will also have to rename the 'train_npz' and 'test_vol_h5' back into simple 'train' and 'test' to directly run our code if you have read the code of dataset_Synapse.py or dataset_ACDC.py.

[KYC890]

Hi @Dootmaan. I am also missing a "lists_Synapse" folder, could you provide one for me? Or you could directly provide a copy of the Synapse dataset you trained on before, my email: 1769838453@qq.com.