{ "id": 370436, "title": "Cropped datasets", "url": "/competitions/rsna-breast-cancer-detection/discussion/370436", "author_name": "FabienDaniel", "post_date": "2022-12-04T13:35:03.119000", "votes": 15, "comment_count": 2, "views": 0, "content": "

Using kernels from @theoviel and @davidroberts, I created cropped datasets of .png files, with sizes from 256 to 1024, where text from images are removed and images are cropped around the breast.

\n

The kernel used to perform the extraction extract is here
\nand the links to datasets are (indicating the longest image size):

\n\n
\n

EDIT: new versions were created, where the aspect ratio is now preserved and where VOI_LUT is applied. Links:

\n", "messages": [ { "id": 2054838, "postDate": "2022-12-04T13:35:03.120Z", "content": "

Using kernels from @theoviel and @davidroberts, I created cropped datasets of .png files, with sizes from 256 to 1024, where text from images are removed and images are cropped around the breast.

\n

The kernel used to perform the extraction extract is here
\nand the links to datasets are (indicating the longest image size):

\n\n
\n

EDIT: new versions were created, where the aspect ratio is now preserved and where VOI_LUT is applied. Links:

\n", "rawMarkdown": "Using kernels from @theoviel and @davidroberts, I created cropped datasets of .png files, with sizes from 256 to 1024, where text from images are removed and images are cropped around the breast.\n\nThe kernel used to perform the extraction extract is [here](https://www.kaggle.com/code/fabiendaniel/dicom-cropped-resized-png-jpg)\nand the links to datasets are (indicating the longest image size):\n- [256](https://www.kaggle.com/datasets/fabiendaniel/rsna-cropped-png-256) \n- [512](https://www.kaggle.com/datasets/fabiendaniel/rsna-cropped-png-512) \n- [768](https://www.kaggle.com/datasets/fabiendaniel/rsna-cropped-png-768)\n- [1024](https://www.kaggle.com/datasets/fabiendaniel/rsna-cropped-png-1024)\n\n___\n**EDIT**: new versions were created, where the aspect ratio is now preserved and where VOI_LUT is applied. Links:\n- [512](https://www.kaggle.com/datasets/fabiendaniel/rsna-cropped-voi-png-512)\n- [1024](https://www.kaggle.com/datasets/fabiendaniel/rsna-cropped-voi-png-1024)\n- [2048 part1](https://www.kaggle.com/datasets/fabiendaniel/rsna-cropped-voi-png-2048-part1) [2048 part2](https://www.kaggle.com/datasets/fabiendaniel/rsna-cropped-voi-png-2048-part2)", "votes": 15 }, { "id": 2138428, "postDate": "2023-02-10T18:54:38.810Z", "content": "

Thanks a lot

", "rawMarkdown": "Thanks a lot", "votes": 1 }, { "id": 2088137, "postDate": "2023-01-06T05:17:10.973Z", "content": "

Thanks! Very appreciated!

", "rawMarkdown": "Thanks! Very appreciated!", "votes": 1 } ], "comments": [ { "id": 2138428, "author_name": "Ches Charlemagne", "author_url": "", "post_date": "2023-02-10T18:54:38.810000", "content": "

Thanks a lot

", "votes": 1, "replies": [] }, { "id": 2088137, "author_name": "Nebil Ibrahim", "author_url": "", "post_date": "2023-01-06T05:17:10.973000", "content": "

Thanks! Very appreciated!

", "votes": 1, "replies": [] } ], "raw_markdown_by_id": { "2054838": "Using kernels from @theoviel and @davidroberts, I created cropped datasets of .png files, with sizes from 256 to 1024, where text from images are removed and images are cropped around the breast.\n\nThe kernel used to perform the extraction extract is [here](https://www.kaggle.com/code/fabiendaniel/dicom-cropped-resized-png-jpg)\nand the links to datasets are (indicating the longest image size):\n- [256](https://www.kaggle.com/datasets/fabiendaniel/rsna-cropped-png-256) \n- [512](https://www.kaggle.com/datasets/fabiendaniel/rsna-cropped-png-512) \n- [768](https://www.kaggle.com/datasets/fabiendaniel/rsna-cropped-png-768)\n- [1024](https://www.kaggle.com/datasets/fabiendaniel/rsna-cropped-png-1024)\n\n___\n**EDIT**: new versions were created, where the aspect ratio is now preserved and where VOI_LUT is applied. Links:\n- [512](https://www.kaggle.com/datasets/fabiendaniel/rsna-cropped-voi-png-512)\n- [1024](https://www.kaggle.com/datasets/fabiendaniel/rsna-cropped-voi-png-1024)\n- [2048 part1](https://www.kaggle.com/datasets/fabiendaniel/rsna-cropped-voi-png-2048-part1) [2048 part2](https://www.kaggle.com/datasets/fabiendaniel/rsna-cropped-voi-png-2048-part2)", "2138428": "Thanks a lot", "2088137": "Thanks! Very appreciated!" } }