File size: 5,682 Bytes
b8597df
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
147
{
    "schema": "https://github.com/Project-MONAI/MONAI-extra-test-data/releases/download/0.8.1/meta_schema_20220324.json",
    "version": "0.0.1",
    "changelog": {
        "0.0.1": "Initial version"
    },
    "monai_version": "1.5.0",
    "pytorch_version": "2.6.0",
    "numpy_version": "1.26.4",
    "optional_packages_version": {},
    "required_packages_version": {
        "setuptools": "75.8.0",
        "opencv-python-headless": "4.11.0.86",
        "pandas": "2.3.0",
        "seaborn": "0.13.2",
        "scikit-learn": "1.6.1",
        "progressbar": "2.5",
        "pydicom": "3.0.1",
        "fire": "0.7.0",
        "torchvision": "0.21.0",
        "detectron2": "0.6",
        "lxml": "5.4.0",
        "pillow": "11.2.1"
    },
    "name": "retinalOCT_RPD_segmentation",
    "task": "Reticular Pseudodrusen (RPD) instance segmentation.",
    "description": "This network detects and segments Reticular Pseudodrusen (RPD) instances in Optical Coherence Tomography (OCT) B-scans which can be presented in a vol or dicom format.",
    "authors": "Yelena Bagdasarova, Scott Song",
    "copyright": "Copyright (c) 2022, uw-biomedical-ml",
    "network_data_format": {
        "inputs": {
            "image": {
                "type": "image",
                "format": "magnitude",
                "modality": "OCT",
                "num_channels": 1,
                "spatial_shape": [
                    496,
                    1024
                ],
                "dtype": "int16",
                "value_range": [
                    0,
                    256
                ],
                "is_patch_data": false,
                "channel_def": {
                    "0": "image"
                }
            }
        },
        "preprocessed_data_sources": {
            "vol_file": {
                "type": "image",
                "format": "magnitude",
                "modality": "OCT",
                "num_channels": 1,
                "spatial_shape": [
                    496,
                    1024,
                    "D"
                ],
                "dtype": "int16",
                "value_range": [
                    0,
                    256
                ],
                "description": "The pixel array of each OCT slice is extracted with volreader and the png files saved to <extracted_dir>/<some>/<file>/<name>/<some_file_name>_oct_<DDD>.png on disk, where <DDD> is the slice number and a nested hierarchy of folders is created using the underscores in the original filename. "
            },
            "dicom_series": {
                "type": "image",
                "format": "magnitude",
                "modality": "OCT",
                "SOP class UID": "1.2.840.10008.5.1.4.1.1.77.1.5.4",
                "num_channels": 1,
                "spatial_shape": [
                    496,
                    1024,
                    "D"
                ],
                "dtype": "int16",
                "value_range": [
                    0,
                    256
                ],
                "description": "The pixel array of each OCT slice is extracted with pydicom and the png files saved to <extracted_dir>/<SOPInstanceUID>/<SOPInstanceUID>_oct_<DDD>.png on disk, where <DDD> is the slice number. "
            }
        },
        "outputs": {
            "pred": {
                "dtype": "dictionary",
                "type": "dictionary",
                "format": "COCO",
                "modality": "n/a",
                "value_range": [
                    0,
                    1
                ],
                "num_channels": 1,
                "spatial_shape": [
                    496,
                    1024
                ],
                "channel_def": {
                    "0": "RPD"
                },
                "description": "This output is a JSON file in COCO Instance Segmentation format, containing bounding boxes, segmentation masks, and output probabilities for detected instances."
            }
        },
        "post_processed_outputs": {
            "binary segmentation": {
                "type": "image",
                "format": "TIFF",
                "modality": "OCT",
                "num_channels": 3,
                "spatial_shape": [
                    496,
                    1024
                ],
                "description": "This output is a multi-page TIFF file. Each page of the TIFF image corresponds to a binary segmentation mask for a single OCT slice from the input volume. The segmentation masks are stacked in the same order as the original OCT slices."
            },
            "binary segmentation overlay": {
                "type": "image",
                "format": "TIFF",
                "modality": "OCT",
                "num_channels": 3,
                "spatial_shape": [
                    496,
                    1024
                ],
                "description": "This output is a multi-page TIFF file. Each page of the TIFF image corresponds to a single OCT slice from the input volume overlayed with the detected binary segmentation mask."
            },
            "instance segmentation overlay": {
                "type": "image",
                "format": "TIFF",
                "modality": "OCT",
                "num_channels": 3,
                "spatial_shape": [
                    496,
                    1024
                ],
                "description": "This output is a multi-page TIFF file. Each page of the TIFF image corresponds to a single OCT slice from the input volume overlayed with the detected binary segmentation mask."
            }
        }
    }
}