{ "id": 203360, "title": "　English and Japanese", "url": "/competitions/ranzcr-clip-catheter-line-classification/discussion/203360", "author_name": "", "post_date": "2020-12-15T00:02:22.320587Z", "votes": 89, "comment_count": 7, "views": 0, "content": "

**RANZCR CLiP - Catheter and Line Position Challenge **

\n

First of all, I would like to express my gratitude to the medical professionals who work with COVID19 on a daily basis.

\n

I am a cardiologist who works with catheters every day.
\nNot only cardiologists, but also doctors and nurses from all other departments check the position of the inserted catheter with X-rays to make sure it is correct.
\nThis may be difficult to imagine, but it is because incorrect insertion of catheters can lead to serious medical accidents.

\n

Take a look at this
\nhttps://www.kaggle.com/ihelon/ranzcr-clip-exploratory-data-analysis

\n

There is an x-ray image of a catheter being inserted (with annotation), an abnormal catheter insertion image and a normal image.

\n

Probably the first thing you will all think is

\n

CVC (central venous catheter)
\nNGT (NG tube / gastric tube)
\nSGC (Swan-Ganz catheter)

\n

I think you may not be familiar with the difference between them, so here is a link to an easy-to-understand Youtube video.
\nIn this competition, CVCs inserted through the hand (called PICC) are also included, although CVCs are usually inserted through the neck or leg.

\n

<CVC
\nhttps://www.youtube.com/watch?v=mTBrCMn86cU

\n

<NGT
\nhttps://www.youtube.com/watch?v=j-TGmr_H3tY

\n

<SGC
\nhttps://www.youtube.com/watch?v=y241HEaBkLA

\n

The important points are

\n

Catheters are visible on X-rays, but they are covered with organs, so you need to change the brightness (at least to the human eye).

\n

There are some artifacts such as spinal bolts in several of the images, so we need to be careful not to misidentify them.

\n

I will update if I notice anything else.

\n
\n

**RANZCR CLiP - Catheter and Line Position Challenge **

\n

まずはCOVID19に日々従事する医療従事者に感謝の意を示したいと思います。

\n

私は、心臓内科医で日々カテーテル診療に従事する者です。
\n循環器内科医だけではなく、他のすべての診療科の医師・看護師は挿入されたカテーテルの位置が正しいかどうかについてレントゲンで確認しています。
\n想像が難しいかもしれませんが、カテーテルの誤挿入は重大な医療事故につながるからですね。
\nたびたび訴訟にもなっています。

\n

これを見てください
\nhttps://www.kaggle.com/ihelon/ranzcr-clip-exploratory-data-analysis

\n

カテーテルを挿入されたレントゲン画像（annotationあり）と、異常なカテーテル挿入画像と正常な画像があります。

\n

おそらくまず皆さんは

\n

・CVC（中心静脈カテーテル）：太い点滴と思ってください　
\n栄養剤やカテコラミン（死にそうな人に使う、循環作動薬・強心薬）のためには普通の点滴ではダメです
\n挿入部位は、首・鼠蹊部・手（珍しいが今回はあり）

\n

・NGT（NG tube /胃管）
\n鼻から突っ込む管
\n挿管患者でご飯食べれない人とかは、ドロドロにした栄養剤を胃に入れる
\nそのために用いる

\n

・SGC（スワンガンツカテーテル）
\nこれはわかりにくいですが、心臓の拍出量や、肺の血管の圧力を測定するために用います。

\n

の違いに馴染みがないと思うので、分かりやすいYoutube動画のリンクを張っておきます。
\n珍しいのはCVCは首や足から挿入するのが一般的ですが本コンペティションでは、手から入れるCVC（PICCと呼ばれます)も入っていますね。

\n


\nhttps://www.youtube.com/watch?v=mTBrCMn86cU

\n


\nhttps://www.youtube.com/watch?v=j-TGmr_H3tY

\n


\nhttps://www.youtube.com/watch?v=y241HEaBkLA

\n

重要な点としては

\n

・カテーテルはレントゲンで見えるが、臓器と被っているので輝度を変更してみる必要があります（少なくとも人間の眼では）

\n

・本画像の複数枚には、脊椎ボルトのようなアーチファクトもありますので誤認識しないように注意が必要かと思いました

\n

・ペースメーカーが入っている患者がいるかはわかりませんが、いるとすると認識は困難です。（人間ならペースメーカーにつながれているので分かりますが、ペースメーカーから出ているリードはCVCと似ています。）

\n

・Abnormal画像をみてください
\n重症患者では、心電図モニターのリードなどが体外にくっついています　しかしこれは体内に入れられてたカテーテルかどうかは、医師の眼なら容易ですが、一見すると機械には判断が難しいでしょう。
\n（入口で説明や判断するのもありですが、下肢からのCVCだと説明難しいですね）

\n

他に気づいたことがあればアップデートします。

", "messages": [ { "id": "1112840", "postDate": "12/15/2020 00:02:22", "content": "

**RANZCR CLiP - Catheter and Line Position Challenge **

\n

First of all, I would like to express my gratitude to the medical professionals who work with COVID19 on a daily basis.

\n

I am a cardiologist who works with catheters every day.
\nNot only cardiologists, but also doctors and nurses from all other departments check the position of the inserted catheter with X-rays to make sure it is correct.
\nThis may be difficult to imagine, but it is because incorrect insertion of catheters can lead to serious medical accidents.

\n

Take a look at this
\nhttps://www.kaggle.com/ihelon/ranzcr-clip-exploratory-data-analysis

\n

There is an x-ray image of a catheter being inserted (with annotation), an abnormal catheter insertion image and a normal image.

\n

Probably the first thing you will all think is

\n

CVC (central venous catheter)
\nNGT (NG tube / gastric tube)
\nSGC (Swan-Ganz catheter)

\n

I think you may not be familiar with the difference between them, so here is a link to an easy-to-understand Youtube video.
\nIn this competition, CVCs inserted through the hand (called PICC) are also included, although CVCs are usually inserted through the neck or leg.

\n

<CVC
\nhttps://www.youtube.com/watch?v=mTBrCMn86cU

\n

<NGT
\nhttps://www.youtube.com/watch?v=j-TGmr_H3tY

\n

<SGC
\nhttps://www.youtube.com/watch?v=y241HEaBkLA

\n

The important points are

\n

Catheters are visible on X-rays, but they are covered with organs, so you need to change the brightness (at least to the human eye).

\n

There are some artifacts such as spinal bolts in several of the images, so we need to be careful not to misidentify them.

\n

I will update if I notice anything else.

\n
\n

**RANZCR CLiP - Catheter and Line Position Challenge **

\n

まずはCOVID19に日々従事する医療従事者に感謝の意を示したいと思います。

\n

私は、心臓内科医で日々カテーテル診療に従事する者です。
\n循環器内科医だけではなく、他のすべての診療科の医師・看護師は挿入されたカテーテルの位置が正しいかどうかについてレントゲンで確認しています。
\n想像が難しいかもしれませんが、カテーテルの誤挿入は重大な医療事故につながるからですね。
\nたびたび訴訟にもなっています。

\n

これを見てください
\nhttps://www.kaggle.com/ihelon/ranzcr-clip-exploratory-data-analysis

\n

カテーテルを挿入されたレントゲン画像（annotationあり）と、異常なカテーテル挿入画像と正常な画像があります。

\n

おそらくまず皆さんは

\n

・CVC（中心静脈カテーテル）：太い点滴と思ってください　
\n栄養剤やカテコラミン（死にそうな人に使う、循環作動薬・強心薬）のためには普通の点滴ではダメです
\n挿入部位は、首・鼠蹊部・手（珍しいが今回はあり）

\n

・NGT（NG tube /胃管）
\n鼻から突っ込む管
\n挿管患者でご飯食べれない人とかは、ドロドロにした栄養剤を胃に入れる
\nそのために用いる

\n

・SGC（スワンガンツカテーテル）
\nこれはわかりにくいですが、心臓の拍出量や、肺の血管の圧力を測定するために用います。

\n

の違いに馴染みがないと思うので、分かりやすいYoutube動画のリンクを張っておきます。
\n珍しいのはCVCは首や足から挿入するのが一般的ですが本コンペティションでは、手から入れるCVC（PICCと呼ばれます)も入っていますね。

\n


\nhttps://www.youtube.com/watch?v=mTBrCMn86cU

\n


\nhttps://www.youtube.com/watch?v=j-TGmr_H3tY

\n


\nhttps://www.youtube.com/watch?v=y241HEaBkLA

\n

重要な点としては

\n

・カテーテルはレントゲンで見えるが、臓器と被っているので輝度を変更してみる必要があります（少なくとも人間の眼では）

\n

・本画像の複数枚には、脊椎ボルトのようなアーチファクトもありますので誤認識しないように注意が必要かと思いました

\n

・ペースメーカーが入っている患者がいるかはわかりませんが、いるとすると認識は困難です。（人間ならペースメーカーにつながれているので分かりますが、ペースメーカーから出ているリードはCVCと似ています。）

\n

・Abnormal画像をみてください
\n重症患者では、心電図モニターのリードなどが体外にくっついています　しかしこれは体内に入れられてたカテーテルかどうかは、医師の眼なら容易ですが、一見すると機械には判断が難しいでしょう。
\n（入口で説明や判断するのもありですが、下肢からのCVCだと説明難しいですね）

\n

他に気づいたことがあればアップデートします。

", "rawMarkdown": "**RANZCR CLiP - Catheter and Line Position Challenge **\n\nFirst of all, I would like to express my gratitude to the medical professionals who work with COVID19 on a daily basis.\n\nI am a cardiologist who works with catheters every day.\nNot only cardiologists, but also doctors and nurses from all other departments check the position of the inserted catheter with X-rays to make sure it is correct.\nThis may be difficult to imagine, but it is because incorrect insertion of catheters can lead to serious medical accidents.\n\nTake a look at this\nhttps://www.kaggle.com/ihelon/ranzcr-clip-exploratory-data-analysis\n\n\nThere is an x-ray image of a catheter being inserted (with annotation), an abnormal catheter insertion image and a normal image.\n\nProbably the first thing you will all think is\n\nCVC (central venous catheter)\nNGT (NG tube / gastric tube)\nSGC (Swan-Ganz catheter)\n\nI think you may not be familiar with the difference between them, so here is a link to an easy-to-understand Youtube video.\nIn this competition, CVCs inserted through the hand (called PICC) are also included, although CVCs are usually inserted through the neck or leg.\n\n\nhttps://www.youtube.com/watch?v=mTBrCMn86cU\n\n\nhttps://www.youtube.com/watch?v=j-TGmr_H3tY\n\n\nhttps://www.youtube.com/watch?v=y241HEaBkLA\n\n\n重要な点としては\n\n・カテーテルはレントゲンで見えるが、臓器と被っているので輝度を変更してみる必要があります（少なくとも人間の眼では）\n\n・本画像の複数枚には、脊椎ボルトのようなアーチファクトもありますので誤認識しないように注意が必要かと思いました\n\n・ペースメーカーが入っている患者がいるかはわかりませんが、いるとすると認識は困難です。（人間ならペースメーカーにつながれているので分かりますが、ペースメーカーから出ているリードはCVCと似ています。）\n\n・Abnormal画像をみてください\n重症患者では、心電図モニターのリードなどが体外にくっついています　しかしこれは体内に入れられてたカテーテルかどうかは、医師の眼なら容易ですが、一見すると機械には判断が難しいでしょう。\n（入口で説明や判断するのもありですが、下肢からのCVCだと説明難しいですね）\n\n他に気づいたことがあればアップデートします。", "votes": null }, { "id": "1112858", "postDate": "12/15/2020 00:31:15", "content": "

I'd also like to thank the medical professionals all over the world (including you).

\n

Thank you for sharing great knowledge.

", "rawMarkdown": "I'd also like to thank the medical professionals all over the world (including you).\n\nThank you for sharing great knowledge.", "votes": null }, { "id": "1113447", "postDate": "12/15/2020 12:49:16", "content": "

Thank you so much. Also.. Probably it's better if Kaggle adds a page named \"Domain Knowledge\" or something similar that could be a one-stop for domain knowledge for competitions that are usually held on kaggle with an option for the community to contribute

", "rawMarkdown": "Thank you so much. Also.. Probably it's better if Kaggle adds a page named \"Domain Knowledge\" or something similar that could be a one-stop for domain knowledge for competitions that are usually held on kaggle with an option for the community to contribute", "votes": null }, { "id": "1113765", "postDate": "12/15/2020 17:07:29", "content": "

Thank you Sahashi sama. Any kind of domain help we get in this competition would be really really useful!

", "rawMarkdown": "Thank you Sahashi sama. Any kind of domain help we get in this competition would be really really useful!", "votes": null }, { "id": "1113846", "postDate": "12/15/2020 18:16:18", "content": "

Thank you so much for this information.

", "rawMarkdown": "Thank you so much for this information.", "votes": null }, { "id": "1118337", "postDate": "12/19/2020 01:04:49", "content": "

Thanks for the great info! You mentioned changing brightness for the human eye usually works better. I thought the same when passing the images to my image data generator. If the human eye can detect it better, I guess it can help the model recognize it better too and results in better generalization ability. I am playing with this parameter in my notebook here: https://www.kaggle.com/sinamhd9/keras-models-image-data-generator-part1/edit/run/49542288.

", "rawMarkdown": "Thanks for the great info! You mentioned changing brightness for the human eye usually works better. I thought the same when passing the images to my image data generator. If the human eye can detect it better, I guess it can help the model recognize it better too and results in better generalization ability. I am playing with this parameter in my notebook here: https://www.kaggle.com/sinamhd9/keras-models-image-data-generator-part1/edit/run/49542288.", "votes": null }, { "id": "1229425", "postDate": "03/07/2021 10:44:01", "content": "

Thank you for useful information.
\nI'm convinced of poor CV score in CVC class.

", "rawMarkdown": "Thank you for useful information.\nI'm convinced of poor CV score in CVC class.", "votes": null }, { "id": "2473571", "postDate": "10/08/2023 11:49:14", "content": "

Thanks for sharing!!

", "rawMarkdown": "Thanks for sharing!!", "votes": null } ], "comments": [ { "id": 1112858, "author_name": "ttahara", "author_url": "", "post_date": "12/15/2020 00:31:15", "content": "

I'd also like to thank the medical professionals all over the world (including you).

\n

Thank you for sharing great knowledge.

", "votes": null, "replies": [] }, { "id": 1113447, "author_name": "rawwar", "author_url": "", "post_date": "12/15/2020 12:49:16", "content": "

Thank you so much. Also.. Probably it's better if Kaggle adds a page named \"Domain Knowledge\" or something similar that could be a one-stop for domain knowledge for competitions that are usually held on kaggle with an option for the community to contribute

", "votes": null, "replies": [] }, { "id": 1113765, "author_name": "allohvk", "author_url": "", "post_date": "12/15/2020 17:07:29", "content": "

Thank you Sahashi sama. Any kind of domain help we get in this competition would be really really useful!

", "votes": null, "replies": [] }, { "id": 1113846, "author_name": "amit209parmar", "author_url": "", "post_date": "12/15/2020 18:16:18", "content": "

Thank you so much for this information.

", "votes": null, "replies": [] }, { "id": 1118337, "author_name": "sinamhd9", "author_url": "", "post_date": "12/19/2020 01:04:49", "content": "

Thanks for the great info! You mentioned changing brightness for the human eye usually works better. I thought the same when passing the images to my image data generator. If the human eye can detect it better, I guess it can help the model recognize it better too and results in better generalization ability. I am playing with this parameter in my notebook here: https://www.kaggle.com/sinamhd9/keras-models-image-data-generator-part1/edit/run/49542288.

", "votes": null, "replies": [] }, { "id": 1229425, "author_name": "kotanoda", "author_url": "", "post_date": "03/07/2021 10:44:01", "content": "

Thank you for useful information.
\nI'm convinced of poor CV score in CVC class.

", "votes": null, "replies": [] }, { "id": 2473571, "author_name": "kuroco", "author_url": "", "post_date": "10/08/2023 11:49:14", "content": "

Thanks for sharing!!

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1112840": "**RANZCR CLiP - Catheter and Line Position Challenge **\n\nFirst of all, I would like to express my gratitude to the medical professionals who work with COVID19 on a daily basis.\n\nI am a cardiologist who works with catheters every day.\nNot only cardiologists, but also doctors and nurses from all other departments check the position of the inserted catheter with X-rays to make sure it is correct.\nThis may be difficult to imagine, but it is because incorrect insertion of catheters can lead to serious medical accidents.\n\nTake a look at this\nhttps://www.kaggle.com/ihelon/ranzcr-clip-exploratory-data-analysis\n\n\nThere is an x-ray image of a catheter being inserted (with annotation), an abnormal catheter insertion image and a normal image.\n\nProbably the first thing you will all think is\n\nCVC (central venous catheter)\nNGT (NG tube / gastric tube)\nSGC (Swan-Ganz catheter)\n\nI think you may not be familiar with the difference between them, so here is a link to an easy-to-understand Youtube video.\nIn this competition, CVCs inserted through the hand (called PICC) are also included, although CVCs are usually inserted through the neck or leg.\n\n\nhttps://www.youtube.com/watch?v=mTBrCMn86cU\n\n\nhttps://www.youtube.com/watch?v=j-TGmr_H3tY\n\n\nhttps://www.youtube.com/watch?v=y241HEaBkLA\n\n\n重要な点としては\n\n・カテーテルはレントゲンで見えるが、臓器と被っているので輝度を変更してみる必要があります（少なくとも人間の眼では）\n\n・本画像の複数枚には、脊椎ボルトのようなアーチファクトもありますので誤認識しないように注意が必要かと思いました\n\n・ペースメーカーが入っている患者がいるかはわかりませんが、いるとすると認識は困難です。（人間ならペースメーカーにつながれているので分かりますが、ペースメーカーから出ているリードはCVCと似ています。）\n\n・Abnormal画像をみてください\n重症患者では、心電図モニターのリードなどが体外にくっついています　しかしこれは体内に入れられてたカテーテルかどうかは、医師の眼なら容易ですが、一見すると機械には判断が難しいでしょう。\n（入口で説明や判断するのもありですが、下肢からのCVCだと説明難しいですね）\n\n他に気づいたことがあればアップデートします。", "1112858": "I'd also like to thank the medical professionals all over the world (including you).\n\nThank you for sharing great knowledge.", "1113447": "Thank you so much. Also.. Probably it's better if Kaggle adds a page named \"Domain Knowledge\" or something similar that could be a one-stop for domain knowledge for competitions that are usually held on kaggle with an option for the community to contribute", "1113765": "Thank you Sahashi sama. Any kind of domain help we get in this competition would be really really useful!", "1113846": "Thank you so much for this information.", "1118337": "Thanks for the great info! You mentioned changing brightness for the human eye usually works better. I thought the same when passing the images to my image data generator. If the human eye can detect it better, I guess it can help the model recognize it better too and results in better generalization ability. I am playing with this parameter in my notebook here: https://www.kaggle.com/sinamhd9/keras-models-image-data-generator-part1/edit/run/49542288.", "1229425": "Thank you for useful information.\nI'm convinced of poor CV score in CVC class.", "2473571": "Thanks for sharing!!" }, "source": "meta" }