关于BRATSDataset3D问题

[NickWang1122]

BRATSDataset3D为什么会有两个len函数
def len(self):
return len(self.database) * 155
def len(self):
return len(self. Database)

[NickWang1122]

按照您源代码的想法应该是得到所有样本数*155张切片送入网络，但是实际上送入网络的只有样本数个数的数据，应该是后面的len函数覆盖了前面len(self. Database) * 155

[WuJunde]

It should be len(self.database) * 155. will be revised in the coming version

[saisusmitha]

@WuJunde Kindly let me know how many images are in the training set, also in bratsdataloader.py
n = x // 155 #quotient
slice = x % 155 #remainder
o = torch.tensor(nib_img.get_fdata())[:,:,slice]

what does this slice usage say? only one slice out of 155 is used for training. Kindly explain the data loader as using the given loader once after data is loaded I saved the images all 5 images(t1,t2...) as png, they look wrong. kindly explain about how you used 240x240x155 original image shape, what pre-processing is done and where and how to find the hd95, jaccard,dice scores?

Kindly let me know where I am going wrong