![\"Staining-intensity-for-H-E-rises-as-section-thickness-increases-Panel-A-shows\"](\"https://i.ibb.co/st9jk2T/Staining-intensity-for-H-E-rises-as-section-thickness-increases-Panel-A-shows.jpg\")
Good visual showing how tissue section thickness can impact staining intensity.
\n
Paper Reference Link - https://www.researchgate.net/publication/346470269_Impact_of_Preanalytical_Factors_During_Histology_Processing_on_Section_Suitability_for_Digital_Image_Analysis
\nPicture Reference Link - https://www.researchgate.net/figure/Staining-intensity-for-H-E-rises-as-section-thickness-increases-Panel-A-shows_fig2_346470269
", "messages": [ { "id": "1831996", "postDate": "06/24/2022 15:15:10", "content": "Good visual showing how tissue section thickness can impact staining intensity.
\n
Paper Reference Link - https://www.researchgate.net/publication/346470269_Impact_of_Preanalytical_Factors_During_Histology_Processing_on_Section_Suitability_for_Digital_Image_Analysis
\nPicture Reference Link - https://www.researchgate.net/figure/Staining-intensity-for-H-E-rises-as-section-thickness-increases-Panel-A-shows_fig2_346470269
", "rawMarkdown": "**Good visual showing how tissue section thickness can impact staining intensity.** \n* This image is for H&E staining. I'm not certain but this finding may generalize to other staining techniques\n* Might indicate that some augmentation surrounding intensity/contrast could be helpful.\n\n\n\nPaper Reference Link - https://www.researchgate.net/publication/346470269_Impact_of_Preanalytical_Factors_During_Histology_Processing_on_Section_Suitability_for_Digital_Image_Analysis\n\nPicture Reference Link - https://www.researchgate.net/figure/Staining-intensity-for-H-E-rises-as-section-thickness-increases-Panel-A-shows_fig2_346470269", "votes": null }, { "id": "1833170", "postDate": "06/25/2022 17:44:06", "content": "Nice visual. I wonder if any of the test images use tissue_thickness not equal to 4. All train and provided test info use 4 but if test uses something other than 4, we will need to lighten and darken the images based on your plot.
Hi Chris, the test images do use different slice thicknesses for different organs. Note that training images come from HPA and test come from a combo of HuBMAP and HPA.
\nHere’s what the hosts said in the data section:
\n\n", "rawMarkdown": "Hi Chris, the test images do use different slice thicknesses for different organs. Note that training images come from HPA and test come from a combo of HuBMAP and HPA.\n\nHere’s what the hosts said in the data section:\n\n> `tissue_thickness` - The thickness of the biopsy sample in micrometers. All HPA images have a thickness of 4 µm. The Hubmap samples have tissue slice thicknesses 10 µm for kidney, 8 µm for large intestine, 4 µm for spleen, 5 µm for lung, and 5 µm for prostate.", "votes": null }, { "id": "1833199", "postDate": "06/25/2022 18:07:11", "content": "\n
tissue_thickness- The thickness of the biopsy sample in micrometers. All HPA images have a thickness of 4 µm. The Hubmap samples have tissue slice thicknesses 10 µm for kidney, 8 µm for large intestine, 4 µm for spleen, 5 µm for lung, and 5 µm for prostate.
Here’s the relevant quote regarding dataset distributions:
\n\n", "rawMarkdown": "Here’s the relevant quote regarding dataset distributions:\n\n> **The training dataset consists of data from public HPA data, the public test set is a combination of private HPA data and HuBMAP data, and the private test set contains only HuBMAP data.** Adapting models to function properly when presented with data that was prepared using a different protocol will be one of the core challenges of this competition. While this is expected to make the problem more difficult, developing models that generalize is a key goal of this endeavor.", "votes": null } ], "comments": [ { "id": 1833170, "author_name": "cdeotte", "author_url": "", "post_date": "06/25/2022 17:44:06", "content": "The training dataset consists of data from public HPA data, the public test set is a combination of private HPA data and HuBMAP data, and the private test set contains only HuBMAP data. Adapting models to function properly when presented with data that was prepared using a different protocol will be one of the core challenges of this competition. While this is expected to make the problem more difficult, developing models that generalize is a key goal of this endeavor.
\n
Nice visual. I wonder if any of the test images use tissue_thickness not equal to 4. All train and provided test info use 4 but if test uses something other than 4, we will need to lighten and darken the images based on your plot.
Hi Chris, the test images do use different slice thicknesses for different organs. Note that training images come from HPA and test come from a combo of HuBMAP and HPA.
\nHere’s what the hosts said in the data section:
\n\n", "votes": null, "replies": [] }, { "id": 1833199, "author_name": "dschettler8845", "author_url": "", "post_date": "06/25/2022 18:07:11", "content": "\n
tissue_thickness- The thickness of the biopsy sample in micrometers. All HPA images have a thickness of 4 µm. The Hubmap samples have tissue slice thicknesses 10 µm for kidney, 8 µm for large intestine, 4 µm for spleen, 5 µm for lung, and 5 µm for prostate.
Here’s the relevant quote regarding dataset distributions:
\n\n", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "1831996": "**Good visual showing how tissue section thickness can impact staining intensity.** \n* This image is for H&E staining. I'm not certain but this finding may generalize to other staining techniques\n* Might indicate that some augmentation surrounding intensity/contrast could be helpful.\n\n\n\nPaper Reference Link - https://www.researchgate.net/publication/346470269_Impact_of_Preanalytical_Factors_During_Histology_Processing_on_Section_Suitability_for_Digital_Image_Analysis\n\nPicture Reference Link - https://www.researchgate.net/figure/Staining-intensity-for-H-E-rises-as-section-thickness-increases-Panel-A-shows_fig2_346470269", "1833170": "Nice visual. I wonder if any of the test images use `tissue_thickness` not equal to 4. All train and provided test info use 4 but if test uses something other than 4, we will need to lighten and darken the images based on your plot.", "1833196": "Hi Chris, the test images do use different slice thicknesses for different organs. Note that training images come from HPA and test come from a combo of HuBMAP and HPA.\n\nHere’s what the hosts said in the data section:\n\n> `tissue_thickness` - The thickness of the biopsy sample in micrometers. All HPA images have a thickness of 4 µm. The Hubmap samples have tissue slice thicknesses 10 µm for kidney, 8 µm for large intestine, 4 µm for spleen, 5 µm for lung, and 5 µm for prostate.", "1833199": "Here’s the relevant quote regarding dataset distributions:\n\n> **The training dataset consists of data from public HPA data, the public test set is a combination of private HPA data and HuBMAP data, and the private test set contains only HuBMAP data.** Adapting models to function properly when presented with data that was prepared using a different protocol will be one of the core challenges of this competition. While this is expected to make the problem more difficult, developing models that generalize is a key goal of this endeavor." }, "source": "meta" }The training dataset consists of data from public HPA data, the public test set is a combination of private HPA data and HuBMAP data, and the private test set contains only HuBMAP data. Adapting models to function properly when presented with data that was prepared using a different protocol will be one of the core challenges of this competition. While this is expected to make the problem more difficult, developing models that generalize is a key goal of this endeavor.
\n