应该是v1的，但是model_restore.py对不上

I found your poster "Pelvic Fracture Segmentation Using a Multi-scale Distance-weighted Neural Network" at MICCAI23 where you reference this repository.

Unfortunately, it seems there is nothing (yet?) shared in this repo. Do you plan to update the code and can you give the community an idea of the timeline?

I'd like to highlight the importance to share these resources with the community to allow us to independently evaluate the results and use the model for future research - which ultimately will led to citations of your paper, a win-win situation ;)

'model_best.model.pkl' is not provided，you just provide ' FractureSegModel.model'，Can you provide the file of 'model_best.model.pkl'，thanks

So before training nnUNet, you first prepare the data using utils.py script inside (Training/fracSegNet/dataset_conversion) folder. But, as I know the dataset is in the format of 001.mha .....100.mha. How come you guys read "_RI_frac.nii.gz" , "_LI_frac.nii.gz" , "_SA_frac.nii.gz" nifti files. Can you provide information about which code you've ran to get nii.gz files from the .mha files?

Hi, thanks for you nice work.
I want to ask can you share your code for training?
Because now you only release the testing code and the pre-trained model.
I cannot re-implement the training process by myself.

Thanks!

Hello, thank you for your efforts and great work. Do you plan to make a version to work with the nnunetV2? It would be great to be able to use apple MPS.

I am looking forward to using your program when it becomes available with the apple gpu!

Hello, when I try to run this code : nnUNet_plan_and_preprocess -t 600 --verify_dataset_integrity
It shows the error:
ModuleNotFoundError: No module named 'matplotlib'
ModuleNotFoundError: No module named 'vtkmodules'
Well I have installed the latest version of the modules...but
Can you please provide the requirements.txt file so that all the libraries will have correct version.

Hi, thanks to your great work.I'm having some issues running your code. Can you give me some help？@YzzLiu
in function get_tp_fp_fn_tn（）
‘’‘
def get_tp_fp_fn_tn(net_output, gt, disMap = None, axes=None, mask=None, square=False ,current_epoch = None):
smooth_trans = True
Tauo_st = 0
st_epoch = 1000

if smooth_trans == False:
    if disMap != None:
        disMap = disMap / torch.mean(disMap)
        temp_disMap_value = disMap
if smooth_trans == True:
    if disMap != None:
        disMap = disMap / torch.mean(disMap)
        if current_epoch <Tauo_st:
            temp_disMap_value = torch.ones_like(disMap)
        elif Tauo_st <= current_epoch < Tauo_st + st_epoch :
            warm_start_matrix = torch.ones_like(disMap)
            warm_para = float(Tauo_st + st_epoch - current_epoch) / st_epoch
            temp_disMap_value = warm_para * warm_start_matrix + (1 - warm_para) * disMap
        elif current_epoch >= Tauo_st + st_epoch:
            temp_disMap_value = disMap

disMap = temp_disMap_value

if axes is None:
    axes = tuple(range(2, len(net_output.size())))

shp_x = net_output.shape
shp_y = gt.shape
shp_disMap = disMap.shape

with torch.no_grad():
    if len(shp_x) != len(shp_y):
        gt = gt.view((shp_y[0], 1, *shp_y[1:]))
        disMap = disMap.view((shp_disMap[0], 1, *shp_disMap[1:]))
    if all([i == j for i, j in zip(net_output.shape, gt.shape)]):
        # if this is the case then gt is probably already a one hot encoding
        y_onehot = gt
    else:
        gt = gt.long()
        y_onehot = torch.zeros(shp_x)
        if net_output.device.type == "cuda":
            y_onehot = y_onehot.cuda(net_output.device.index)
        y_onehot.scatter_(1, gt, 1)
disMap2onehot = disMap.repeat_interleave(shp_x[1],dim = 1)

disMap2onehot = disMap2onehot.cuda(net_output.device.index)

tp = net_output * y_onehot
fp = net_output * (1 - y_onehot)
fn = (1 - net_output) * y_onehot
tn = (1 - net_output) * (1 - y_onehot)

tp = torch.mul(tp, disMap2onehot)
fp = torch.mul(fp, disMap2onehot)
fn = torch.mul(fn, disMap2onehot)
tn = torch.mul(tn, disMap2onehot)

if mask is not None:
    tp = torch.stack(tuple(x_i * mask[:, 0] for x_i in torch.unbind(tp, dim=1)), dim=1)
    fp = torch.stack(tuple(x_i * mask[:, 0] for x_i in torch.unbind(fp, dim=1)), dim=1)
    fn = torch.stack(tuple(x_i * mask[:, 0] for x_i in torch.unbind(fn, dim=1)), dim=1)
    tn = torch.stack(tuple(x_i * mask[:, 0] for x_i in torch.unbind(tn, dim=1)), dim=1)

if square:
    tp = tp ** 2
    fp = fp ** 2
    fn = fn ** 2
    tn = tn ** 2

if len(axes) > 0:
    tp = sum_tensor(tp, axes, keepdim=False)
    fp = sum_tensor(fp, axes, keepdim=False)
    fn = sum_tensor(fn, axes, keepdim=False)
    tn = sum_tensor(tn, axes, keepdim=False)

return tp, fp, fn, tn

‘’‘
tp = torch.mul(tp, disMap2onehot) there is a bug：The size of tensor a (128) must match the size of tensor b (218) at non-singleton dimension 4.
i add ' disMap2onehot = torch.nn.functional.interpolate(disMap2onehot, size=tp.shape[2:])'
I don't know if it's the right thing to do.

.