Datasets:
YongchengYAO
commited on
Commit
Β·
bf953a1
1
Parent(s):
63bd1b5
update doc
Browse files- doc/dataset_skm-tea.md +20 -20
doc/dataset_skm-tea.md
CHANGED
|
@@ -16,13 +16,13 @@ This tutorial is about:
|
|
| 16 |
```
|
| 17 |
# Files from SKM-TEA source
|
| 18 |
βββ v1-release
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
```
|
| 27 |
|
| 28 |
```
|
|
@@ -31,14 +31,14 @@ This tutorial is about:
|
|
| 31 |
βββ Masks
|
| 32 |
βββ MTR_001.nii.gz
|
| 33 |
βββ ...
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
```
|
| 38 |
|
| 39 |
## How to preprocess data?
|
| 40 |
|
| 41 |
-
β οΈ DO NOT change file or folder names. See why here
|
| 42 |
|
| 43 |
1. Go to the parent folder of `dicoms`, install [`dcm2niix`](https://github.com/rordenlab/dcm2niix), add processing script
|
| 44 |
|
|
@@ -46,8 +46,8 @@ This tutorial is about:
|
|
| 46 |
# local file structure
|
| 47 |
βββ [your folder]
|
| 48 |
βββ Masks
|
| 49 |
-
|
| 50 |
-
|
| 51 |
```
|
| 52 |
|
| 53 |
```bash
|
|
@@ -87,10 +87,10 @@ This tutorial is about:
|
|
| 87 |
# local file structure
|
| 88 |
βββ [your folder]
|
| 89 |
βββ Masks
|
| 90 |
-
|
| 91 |
-
|
| 92 |
-
|
| 93 |
-
|
| 94 |
```
|
| 95 |
|
| 96 |
2. Put these folders into `SKM-TEA-nii` and upload the **compressed** file to your HF dataset repo
|
|
@@ -98,9 +98,9 @@ This tutorial is about:
|
|
| 98 |
```
|
| 99 |
# DO NOT use other folder names
|
| 100 |
βββ SKM-TEA-nii
|
| 101 |
-
|
| 102 |
-
|
| 103 |
-
|
| 104 |
```
|
| 105 |
|
| 106 |
Your HF repo should contain `SKM-TEA-nii.zip`
|
|
|
|
| 16 |
```
|
| 17 |
# Files from SKM-TEA source
|
| 18 |
βββ v1-release
|
| 19 |
+
βββ dicoms
|
| 20 |
+
βββ MTR_001.tar.gz
|
| 21 |
+
βββ ...
|
| 22 |
+
βββ segmentation_masks
|
| 23 |
+
βββ raw-data-track
|
| 24 |
+
βββ MTR_001.nii.gz
|
| 25 |
+
βββ ...
|
| 26 |
```
|
| 27 |
|
| 28 |
```
|
|
|
|
| 31 |
βββ Masks
|
| 32 |
βββ MTR_001.nii.gz
|
| 33 |
βββ ...
|
| 34 |
+
βββ dicoms
|
| 35 |
+
βββ MTR_001.tar.gz
|
| 36 |
+
βββ ...
|
| 37 |
```
|
| 38 |
|
| 39 |
## How to preprocess data?
|
| 40 |
|
| 41 |
+
β οΈ DO NOT change file or folder names. See why [here](https://huggingface.co/datasets/YongchengYAO/MedVision/blob/main/src/medvision_ds/datasets/SKM_TEA/download.py).
|
| 42 |
|
| 43 |
1. Go to the parent folder of `dicoms`, install [`dcm2niix`](https://github.com/rordenlab/dcm2niix), add processing script
|
| 44 |
|
|
|
|
| 46 |
# local file structure
|
| 47 |
βββ [your folder]
|
| 48 |
βββ Masks
|
| 49 |
+
βββ dicoms
|
| 50 |
+
βββ prepare-skm-tea.sh # <-- add this
|
| 51 |
```
|
| 52 |
|
| 53 |
```bash
|
|
|
|
| 87 |
# local file structure
|
| 88 |
βββ [your folder]
|
| 89 |
βββ Masks
|
| 90 |
+
βββ dicoms
|
| 91 |
+
βββ Images-E1
|
| 92 |
+
βββ Images-E2
|
| 93 |
+
βββ prepare-skm-tea.sh
|
| 94 |
```
|
| 95 |
|
| 96 |
2. Put these folders into `SKM-TEA-nii` and upload the **compressed** file to your HF dataset repo
|
|
|
|
| 98 |
```
|
| 99 |
# DO NOT use other folder names
|
| 100 |
βββ SKM-TEA-nii
|
| 101 |
+
βββ Images-E1
|
| 102 |
+
βββ Images-E2
|
| 103 |
+
βββ Masks
|
| 104 |
```
|
| 105 |
|
| 106 |
Your HF repo should contain `SKM-TEA-nii.zip`
|