TractShapeNet/hcp.py at main · SlicerDMRI/TractShapeNet · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
from __future__ import print_function
import numpy
import torch.utils.data as data
import torch
import random
import h5py


class PointSet_pair(data.Dataset):
    def __init__(self, vec, gt, transform=None, n_points=[1024], val=False):
        self.vec = vec
        self.tpvt = gt
        self.transform = transform
        self.n_points = n_points
        self.val = val
        self.invalid_indices = []
        print(f"Dataset initialized with {len(self.vec)} items.")

    def __getitem__(self, index: int):
        # Process the first sample
        feat_1, label_1 = self._process_sample(index)

        # Process the second random sample
        index2 = random.randint(0, len(self.vec) - 1)
        feat_2, label_2 = self._process_sample(index2)

        return feat_1, label_1, feat_2, label_2

    def _process_sample(self, index: int):
        feat_points = self.vec[index]
        tracts = list(feat_points.keys())
        feat = None

        for i, tract in enumerate(tracts):
            feat_tract = feat_points[tract]
            if feat_tract.shape[0] > 0:
                id1 = numpy.random.randint(feat_tract.shape[0], size=2048)
                feat1 = feat_tract[id1]
                if feat is None:
                    feat = feat1
                else:
                    feat = numpy.concatenate((feat, feat1), 0)
            else:
                # Collect the index of invalid data
                self.invalid_indices.append(index)
                print(f"Sample {index} has no valid data in tract '{tract}'.")

        label = self.tpvt[index]
        if isinstance(label, numpy.ndarray):
            label = label.tolist()

        feat_tensor = torch.tensor(feat.T, dtype=torch.float)
        label_tensor = torch.tensor(label, dtype=torch.float)

        return feat_tensor, label_tensor

    def __len__(self) -> int:
        return len(self.vec)

class PointSet(data.Dataset):
    def __init__(self,vec,gt,transform=None,n_points=[1024],val=False):
        self.vec=vec
        self.tpvt=gt
        self.transform = transform
        self.n_points=n_points
        self.val = val

    def __getitem__(self, index: int):
        #print(index)
        feat_points=self.vec[index]
        tracts=list(feat_points.keys())
        #n_points_sub=int(self.n_points/len(tracts))
        feat=None
        for i,tract in enumerate(tracts):
            feat_tract=feat_points[tract]
            id1 = numpy.random.randint(feat_tract.shape[0], size=2048)
            feat1 = feat_tract[id1]
            if feat is None:
                feat=feat1
            else:
                feat=numpy.concatenate((feat,feat1),0)

        label = numpy.array(self.tpvt[index], dtype=numpy.float32)
        label = torch.tensor(label, dtype=torch.float32)
        feat = torch.tensor(feat.T, dtype=torch.float)
        label = label.clone().detach().float()


        return feat,label

    def __len__(self) -> int:
        return len(self.vec)


class Fiber_pair_vis(data.Dataset):
    def __init__(self,vec,gt,transform=None,n_points=1024):
        self.vec=vec
        self.tpvt=gt
        self.transform = transform
        self.n_points=n_points

    def __getitem__(self, index: int):
        feat_points=self.vec[index]
        feat=feat_points[self.n_points*index:self.n_points*(index+1),:]
        label = self.tpvt[index]
        feat = torch.tensor(feat.T, dtype=torch.float)
        label = torch.tensor(label, dtype=torch.float)
        # if self.transform is not None:
        #     img = self.transform(img)
        return feat,label

    def __len__(self) -> int:
        return len(self.vec)
class Fiber_sub(data.Dataset):
    def __init__(self,vec,gt,transform=None,n_points=[1024]):
        self.vec=vec
        self.tpvt=gt
        self.transform = transform
        self.n_points=n_points

    def __getitem__(self, index: int):

        #print(index)
        sub_id=self.vec[index]
        ff = h5py.File('/subjects/{}.h5'.format(sub_id))
        tracts=list(ff.keys())
        feat=None
        for i,tract in enumerate(tracts):
            feat_tract=ff[tract][:]
            id1 = numpy.random.randint(feat_tract.shape[0], size=20000)
            id = random.sample(list(numpy.unique(id1)), self.n_points[i])
            feat1 = feat_tract[id]
            if feat is None:
                feat=feat1
            else:
                feat=numpy.concatenate((feat,feat1),0)

        label = self.tpvt[index]
        feat = torch.tensor(feat.T, dtype=torch.float)
        label = torch.tensor(label, dtype=torch.float)
        # if self.transform is not None:
        #     img = self.transform(img)
        return feat,label

    def __len__(self) -> int:
        return len(self.vec)