pass value to transform in compose dynamically

[chrisiwu]

Is it somehow possible to pass a value to a transform in a compose dynamically? In my case I want to set the parameter "num_samples" in "RandCropByPosNegLabeld" dynamically. If the resolution of the image is higher it should be higher and vice versa.

Here the relevant part of my code:
train_files = [{"img": img, "seg": seg, "samples": samples} for img, seg, samples in data_list[:number_train]]

train_transforms = Compose([LoadImaged(keys = ["img", "seg"]), EnsureChannelFirstd(keys = ["img", "seg"]), ScaleIntensityd(keys = ["img", "seg"]), SpatialPadd(keys = ["img", "seg"], spatial_size = [96, 96, 96]), RandCropByPosNegLabeld(keys = ["img", "seg"], label_key = "seg", spatial_size = [96, 96, 96], pos = 1, neg = 1, num_samples = 4)])

train_ds = Dataset(data = train_files, transform = train_transforms)

train_loader = DataLoader(train_ds, batch_size = 2, shuffle = True, num_workers = 4, collate_fn = list_data_collate, pin_memory = torch.cuda.is_available())

[KumoLiu]

Hi @chrisiwu, we didn't support dynamically passing a value to a transform in a compose. For images with different resolutions maybe you could use Spacing first. Thanks!