Scripts to download image metadata from the ISIC Archive RESTful API
The International Skin Imaging Collaboration (ISIC) Archive is a partnership among multiple institutions to create an image repository of skin lesions. One of its goals is to improve clinical support in the identification of skin cancer.
The website provides a straight-forward method of downloading the archive's images, but their metadata is only available through their RESTful API. Unfortunately they have not yet implemented a batch-download option for all the images' details. This is why I have written these python scripts.
- ISIC_metadata_importer Downloads metadata for images into a CSV Files.
usage: ISIC_metadata_importer.py [-h] [-c COUNT] [-o OUTPUT] [-k KEYS]
optional arguments:
-h, --help show this help message and exit
-c COUNT, --count COUNT
number of entries requested
-o OUTPUT, --output OUTPUT
path+name of output file
-k KEYS, --keys KEYS CSV file containing kunstructured keys of interest
- ISIC_unstructured_fields.py Returns all unstructured fields available into a comma-separated list. Not all image entries have all these fields, so you might need these for consistency.
- unstructured_keys.csv Input file that defines the unstructured fields of interest. Use the outout from ISIC_unstructured_fields.py to create a CSV file as shown in the table belos. Make sure your internal field names are unique.
| Internal Field Names | ISIC Field Names |
|---|---|
| age | age_approx |
| img_name | name |
| ... | ... |
- all_unstructured_keys.txt A list of unstructured keys extracted from all the images available as of 6/9/2017.
If you agree with the following:
- You want to get the metadata from all the images (max 100000)
- You think the fields I have defined look appropriate (or don't care about them)
Then clone this repo and execute this line:
python ISIC_metadata_importer.py -k unstructured_keys.csv