![\"Selection-134\"](\"https://i.ibb.co/RDCdXJc/Selection-134.png\")
\n
![\"Selection-133\"](\"https://i.ibb.co/0XLYP51/Selection-133.png\")
\n
![\"Selection-136\"](\"https://i.ibb.co/wKkhkQg/Selection-136.png\")
\n
\n\n\n",
"votes": 21
},
{
"id": 1924333,
"postDate": "2022-09-03T01:02:51.403Z",
"content": "additional readings:
\n[1] Histological stroke clot analysis after thrombectomy: Technical aspects and recommendations
\nhttps://www.insist-h2020.eu/images/Dissemination/Staessens_-_2019_-_Int_J_Stroke.pdf
[2] Interpretable CNN for ischemic stroke subtype classification with active model adaptation
\nhttps://bmcmedinformdecismak.biomedcentral.com/track/pdf/10.1186/s12911-021-01721-5.pdf
\n(not image based method. it is using features like age, gender, hospital history, etc)
[3] Thrombus Histology Suggests Cardioembolic Cause in Cryptogenic Stroke
\nhttps://www.ahajournals.org/doi/epub/10.1161/STROKEAHA.116.013105
keywords for google search
\nwhat is CE and LAA?
\nAtherosclerosis = Thickening or hardening of the arteries
\nCardioembolic = the heart pumps unwanted materials into the brain circulation
\nsee: https://slideplayer.com/slide/10713149/37/images/3/SMALL+VESSEL+DISEASE+%2FPENETRATING+ARTERY.jpg
more infor:
\nhttps://www.kaggle.com/code/allunia/mayo-clinic-strip-ai-can-i-trust-my-model
It looks like resolution is important here. Some kind of selective sampling could be useful rather than resizing whole images. Anyone tried something similar to that?
", "rawMarkdown": "It looks like resolution is important here. Some kind of selective sampling could be useful rather than resizing whole images. Anyone tried something similar to that?", "votes": 1 }, { "id": 1925595, "postDate": "2022-09-04T05:52:33.487Z", "content": "area and no. of fragments
\nhttps://www.sciencedirect.com/science/article/pii/S1052305720308818
\n![\"Selection-139\"](\"https://i.ibb.co/71m291q/Selection-139.png\")
",
"votes": 2,
"replies": [
{
"id": 1940978,
"postDate": "2022-09-15T17:25:14.780Z",
"content": "Thanks for this interesting info! With average size of 109 mm2 for LAA vs. 52 mm2 for CE, it seems this would be a useful feature. However to calculate the area of the clots in mm, we need to know the dimensions of a pixel in object space. So far, I have been unable to find data such as sensor pixel size or system magnification used for the images in our data.
\nI did find some specs on a Leica AT-2 System from one of the papers which states dimension of .50um/pixel for 20x, and .25um/pixel for 40x, but different facilities (center_id) will most likely have different equipment and thus various sensor sizes and magnifications.
I looked into the XResolution and YResolution values in the Exif data. Both had values of 10 for the couple of images I sampled. I initially thought this would indicate pixel dimensions in um, but the ResolutionUnit was 3, which corresponds to 'cm'.
Being a large scale image collection project, I am sure this is recorded somewhere. Have you been able to determine any of these values?
", "rawMarkdown": "Thanks for this interesting info! With average size of **109 mm2** for **LAA** vs. **52 mm2** for **CE**, it seems this would be a useful feature. However to calculate the area of the clots in mm, we need to know the dimensions of a pixel in object space. So far, I have been unable to find data such as sensor pixel size or system magnification used for the images in our data.\n\nI did find some specs on a [Leica AT-2 System](https://www.leicabiosystems.com/sites/default/files/2020-10/Aperio_AT2_Brochure_USA.pdf) from one of the papers which states dimension of **.50um/pixel** for **20x**, and **.25um/pixel** for **40x**, but different facilities (`center_id`) will most likely have different equipment and thus various sensor sizes and magnifications.\n\nI looked into the `XResolution` and `YResolution` values in the Exif data. Both had values of 10 for the couple of images I sampled. I initially thought this would indicate pixel dimensions in um, but the `ResolutionUnit` was 3, which corresponds to 'cm'.\n\nBeing a large scale image collection project, I am sure this is recorded somewhere. Have you been able to determine any of these values? ", "votes": 1 } ] }, { "id": 1924327, "postDate": "2022-09-03T00:49:20.247Z", "content": "how to use photoshop and obrit
\nhttps://www.ncbi.nlm.nih.gov/pmc/articles/PMC6894878/
\npaper: Orbit image analysis machine learning software can be used for the histological quantification of acute ischemic stroke blood clots
this can be used as additional feature of target for training
\n[1] Orbit Image Analysis standard operating procedure for histological quantification.
\nhttps://www.ncbi.nlm.nih.gov/pmc/articles/PMC6894878/bin/pone.0225841.s001.pdf
\nhttps://journals.plos.org/plosone/article?id=10.1371/journal.pone.0225841
![\"Selection-135\"](\"https://i.ibb.co/7X0STKm/Selection-135.png\")
experimental results using Orbit procedure[1]:
\n[2] Association between clot composition and stroke origin in mechanical thrombectomy patients: analysis of the Stroke Thromboembolism Registry of Imaging and Pathology
\nhttps://jnis.bmj.com/content/neurintsurg/13/7/594.full.pdf
\n(i think out data comes from this mayo clinic paper)
\"Machine learning classification algorithm results. The stacked ensemble for the classification algorithm differentiating CE from LAA had a fivefold cross-validated AUC of 0.55 (area under the precision-recall curve of 0.33).\"
\nanyone to manually segment image into RBC, WBC, Fibrin, and Platelets and upload to public database?
\nalternative composition determination using qupath software:
\nClot composition characterization using diffuse reflectance spectroscopy in acute ischemic stroke
\nhttps://opg.optica.org/DirectPDFAccess/547E1310-C6BD-4C8B-B9FE31E022E455E1_472721/boe-13-6-3311.pdf?da=1&id=472721&seq=0&mobile=no
Supplementary document for Clot composition characterization using diffuse reflectance spectroscopy in acute ischemic stroke
\nhttps://s3-eu-west-1.amazonaws.com/pstorage-osa-2151918321/35158285/5825798.pdf?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAIC5UZOEV5G5EP3CQ/20220904/eu-west-1/s3/aws4_request&X-Amz-Date=20220904T195023Z&X-Amz-Expires=10&X-Amz-SignedHeaders=host&X-Amz-Signature=6ac79f05b8eadaac9d4a3c90732502493029ce3c3bcffdb077d8f996c677aaed
\n\nexperimental results using Orbit procedure[1]: \n\n[2] Association between clot composition and stroke origin in mechanical thrombectomy patients: analysis of the Stroke Thromboembolism Registry of Imaging and Pathology \n\nhttps://jnis.bmj.com/content/neurintsurg/13/7/594.full.pdf\n(i think out data comes from this mayo clinic paper)\n\n\"Machine learning classification algorithm results. The stacked ensemble for the classification algorithm differentiating CE from LAA had a fivefold cross-validated AUC of 0.55 (area under the precision-recall curve of 0.33).\"\n\n---\n\nanyone to manually segment image into RBC, WBC, Fibrin, and Platelets and upload to public database?\n\n---\n\nalternative composition determination using qupath software:\n\n Clot composition characterization using diffuse reflectance spectroscopy in acute ischemic stroke \nhttps://opg.optica.org/DirectPDFAccess/547E1310-C6BD-4C8B-B9FE31E022E455E1_472721/boe-13-6-3311.pdf?da=1&id=472721&seq=0&mobile=no\n\nSupplementary document for Clot composition characterization using diffuse reflectance spectroscopy in acute ischemic stroke \nhttps://s3-eu-west-1.amazonaws.com/pstorage-osa-2151918321/35158285/5825798.pdf?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAIC5UZOEV5G5EP3CQ/20220904/eu-west-1/s3/aws4_request&X-Amz-Date=20220904T195023Z&X-Amz-Expires=10&X-Amz-SignedHeaders=host&X-Amz-Signature=6ac79f05b8eadaac9d4a3c90732502493029ce3c3bcffdb077d8f996c677aaed",
"votes": 2
}
],
"comments": [
{
"id": 1924333,
"author_name": "hengck23",
"author_url": "",
"post_date": "2022-09-03T01:02:51.403000",
"content": "additional readings:
\n[1] Histological stroke clot analysis after thrombectomy: Technical aspects and recommendations
\nhttps://www.insist-h2020.eu/images/Dissemination/Staessens_-_2019_-_Int_J_Stroke.pdf
[2] Interpretable CNN for ischemic stroke subtype classification with active model adaptation
\nhttps://bmcmedinformdecismak.biomedcentral.com/track/pdf/10.1186/s12911-021-01721-5.pdf
\n(not image based method. it is using features like age, gender, hospital history, etc)
[3] Thrombus Histology Suggests Cardioembolic Cause in Cryptogenic Stroke
\nhttps://www.ahajournals.org/doi/epub/10.1161/STROKEAHA.116.013105
keywords for google search
\nwhat is CE and LAA?
\nAtherosclerosis = Thickening or hardening of the arteries
\nCardioembolic = the heart pumps unwanted materials into the brain circulation
\nsee: https://slideplayer.com/slide/10713149/37/images/3/SMALL+VESSEL+DISEASE+%2FPENETRATING+ARTERY.jpg
more infor:
\nhttps://www.kaggle.com/code/allunia/mayo-clinic-strip-ai-can-i-trust-my-model
It looks like resolution is important here. Some kind of selective sampling could be useful rather than resizing whole images. Anyone tried something similar to that?
", "votes": 1, "replies": [] }, { "id": 1925595, "author_name": "hengck23", "author_url": "", "post_date": "2022-09-04T05:52:33.487000", "content": "area and no. of fragments
\nhttps://www.sciencedirect.com/science/article/pii/S1052305720308818
\n
Thanks for this interesting info! With average size of 109 mm2 for LAA vs. 52 mm2 for CE, it seems this would be a useful feature. However to calculate the area of the clots in mm, we need to know the dimensions of a pixel in object space. So far, I have been unable to find data such as sensor pixel size or system magnification used for the images in our data.
\nI did find some specs on a Leica AT-2 System from one of the papers which states dimension of .50um/pixel for 20x, and .25um/pixel for 40x, but different facilities (center_id) will most likely have different equipment and thus various sensor sizes and magnifications.
I looked into the XResolution and YResolution values in the Exif data. Both had values of 10 for the couple of images I sampled. I initially thought this would indicate pixel dimensions in um, but the ResolutionUnit was 3, which corresponds to 'cm'.
Being a large scale image collection project, I am sure this is recorded somewhere. Have you been able to determine any of these values?
", "votes": 1, "replies": [] } ] }, { "id": 1924327, "author_name": "hengck23", "author_url": "", "post_date": "2022-09-03T00:49:20.247000", "content": "how to use photoshop and obrit
\nhttps://www.ncbi.nlm.nih.gov/pmc/articles/PMC6894878/
\npaper: Orbit image analysis machine learning software can be used for the histological quantification of acute ischemic stroke blood clots
this can be used as additional feature of target for training
\n[1] Orbit Image Analysis standard operating procedure for histological quantification.
\nhttps://www.ncbi.nlm.nih.gov/pmc/articles/PMC6894878/bin/pone.0225841.s001.pdf
\nhttps://journals.plos.org/plosone/article?id=10.1371/journal.pone.0225841
experimental results using Orbit procedure[1]:
\n[2] Association between clot composition and stroke origin in mechanical thrombectomy patients: analysis of the Stroke Thromboembolism Registry of Imaging and Pathology
\nhttps://jnis.bmj.com/content/neurintsurg/13/7/594.full.pdf
\n(i think out data comes from this mayo clinic paper)
\"Machine learning classification algorithm results. The stacked ensemble for the classification algorithm differentiating CE from LAA had a fivefold cross-validated AUC of 0.55 (area under the precision-recall curve of 0.33).\"
\nanyone to manually segment image into RBC, WBC, Fibrin, and Platelets and upload to public database?
\nalternative composition determination using qupath software:
\nClot composition characterization using diffuse reflectance spectroscopy in acute ischemic stroke
\nhttps://opg.optica.org/DirectPDFAccess/547E1310-C6BD-4C8B-B9FE31E022E455E1_472721/boe-13-6-3311.pdf?da=1&id=472721&seq=0&mobile=no
Supplementary document for Clot composition characterization using diffuse reflectance spectroscopy in acute ischemic stroke
\nhttps://s3-eu-west-1.amazonaws.com/pstorage-osa-2151918321/35158285/5825798.pdf?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAIC5UZOEV5G5EP3CQ/20220904/eu-west-1/s3/aws4_request&X-Amz-Date=20220904T195023Z&X-Amz-Expires=10&X-Amz-SignedHeaders=host&X-Amz-Signature=6ac79f05b8eadaac9d4a3c90732502493029ce3c3bcffdb077d8f996c677aaed
\n\n\n",
"1924333": "additional readings: \n\n[1] Histological stroke clot analysis after thrombectomy: Technical aspects and recommendations \nhttps://www.insist-h2020.eu/images/Dissemination/Staessens_-_2019_-_Int_J_Stroke.pdf \n\n[2] Interpretable CNN for ischemic stroke subtype classification with active model adaptation \nhttps://bmcmedinformdecismak.biomedcentral.com/track/pdf/10.1186/s12911-021-01721-5.pdf \n(not image based method. it is using features like age, gender, hospital history, etc)\n\n\n[3] Thrombus Histology Suggests Cardioembolic Cause in Cryptogenic Stroke \nhttps://www.ahajournals.org/doi/epub/10.1161/STROKEAHA.116.013105\n\n---\n\nkeywords for google search\n\n- RBC, WBC, Fibrin, and Platelets\n- etiology\n- histology\n- CE (Cardioembolic) or LAA (Large Artery Atherosclerosis)\n- Thrombosis, thrombus\n- acute ischaemic stroke (AIS) \n- Oribit (software)\n- Mechanical Thrombectomy\n\n---\n\nwhat is CE and LAA?\nAtherosclerosis = Thickening or hardening of the arteries\nCardioembolic = the heart pumps unwanted materials into the brain circulation \nsee: https://slideplayer.com/slide/10713149/37/images/3/SMALL+VESSEL+DISEASE+%2FPENETRATING+ARTERY.jpg \n\n\nmore infor:\nhttps://www.kaggle.com/code/allunia/mayo-clinic-strip-ai-can-i-trust-my-model",
"1942548": "It looks like resolution is important here. Some kind of selective sampling could be useful rather than resizing whole images. Anyone tried something similar to that?",
"1925595": "area and no. of fragments\nhttps://www.sciencedirect.com/science/article/pii/S1052305720308818\n",
"1924327": "how to use photoshop and obrit\nhttps://www.ncbi.nlm.nih.gov/pmc/articles/PMC6894878/\npaper: Orbit image analysis machine learning software can be used for the histological quantification of acute ischemic stroke blood clots\n\nthis can be used as additional feature of target for training\n\n[1] Orbit Image Analysis standard operating procedure for histological quantification. \nhttps://www.ncbi.nlm.nih.gov/pmc/articles/PMC6894878/bin/pone.0225841.s001.pdf\nhttps://journals.plos.org/plosone/article?id=10.1371/journal.pone.0225841\n\n\n\nexperimental results using Orbit procedure[1]: \n\n[2] Association between clot composition and stroke origin in mechanical thrombectomy patients: analysis of the Stroke Thromboembolism Registry of Imaging and Pathology \n\nhttps://jnis.bmj.com/content/neurintsurg/13/7/594.full.pdf\n(i think out data comes from this mayo clinic paper)\n\n\"Machine learning classification algorithm results. The stacked ensemble for the classification algorithm differentiating CE from LAA had a fivefold cross-validated AUC of 0.55 (area under the precision-recall curve of 0.33).\"\n\n---\n\nanyone to manually segment image into RBC, WBC, Fibrin, and Platelets and upload to public database?\n\n---\n\nalternative composition determination using qupath software:\n\n Clot composition characterization using diffuse reflectance spectroscopy in acute ischemic stroke \nhttps://opg.optica.org/DirectPDFAccess/547E1310-C6BD-4C8B-B9FE31E022E455E1_472721/boe-13-6-3311.pdf?da=1&id=472721&seq=0&mobile=no\n\nSupplementary document for Clot composition characterization using diffuse reflectance spectroscopy in acute ischemic stroke \nhttps://s3-eu-west-1.amazonaws.com/pstorage-osa-2151918321/35158285/5825798.pdf?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAIC5UZOEV5G5EP3CQ/20220904/eu-west-1/s3/aws4_request&X-Amz-Date=20220904T195023Z&X-Amz-Expires=10&X-Amz-SignedHeaders=host&X-Amz-Signature=6ac79f05b8eadaac9d4a3c90732502493029ce3c3bcffdb077d8f996c677aaed"
}
}