![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2F314966d24678cff45c38ba97717cb156%2FSelection_042.png?generation=1608143377978583&alt=media\")
https://arxiv.org/pdf/1908.07170v1.pdf
\nEndotracheal Tube Detection and Segmentation in Chest Radiographs using Synthetic Data
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2F46fcba1068fd4398be7a8222adc09120%2FSelection_041.png?generation=1608143396052476&alt=media\")
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2F1187ae70550e4e0ae625e8769c8275fd%2FSelection_040.png?generation=1608143420410735&alt=media\")
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2F595d4e23eb393058f4260f1ba05393b5%2FSelection_053.png?generation=1608149267894616&alt=media\")
https://github.com/xinario/catheter_detection
", "messages": [ { "id": "1116007", "postDate": "12/16/2020 18:30:22", "content": "https://arxiv.org/pdf/1908.07170v1.pdf
\nEndotracheal Tube Detection and Segmentation in Chest Radiographs using Synthetic Data
https://github.com/xinario/catheter_detection
", "rawMarkdown": "https://arxiv.org/pdf/1908.07170v1.pdf\nEndotracheal Tube Detection and Segmentation in Chest Radiographs using Synthetic Data\n\n\n\n\n\n\n\n\n\n\nhttps://github.com/xinario/catheter_detection", "votes": null }, { "id": "1116018", "postDate": "12/16/2020 18:38:59", "content": "Computer-Aided Assessment of Catheters and Tubes on Radiographs: How Good is Artificial Intelligence for Assessment?
\nhttps://www.researchgate.net/publication/339164106_Computer-Aided_Assessment_of_Catheters_and_Tubes_on_Radiographs_How_Good_is_Artificial_Intelligence_for_Assessment
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2Fb684b720a2af17301d3cafb0533e4e94%2FSelection_043.png?generation=1608143921881550&alt=media\")
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2F8ecbdad5d8749f1e214a6c81666d6a06%2FSelection_044.png?generation=1608143937314046&alt=media\")
Accuracy of the 7-8-9 Rule for endotracheal tube placement in the neonate
\nhttps://www.nature.com/articles/7211503
PediatricXray100 dataset
\nhttps://github.com/xinario/PediatricXray1
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2F4b4fa5fd3167de0cf0c3e5e4e4900f11%2FSelection_045.png?generation=1608144897033962&alt=media\")
Deep Convolutional Neural Networks for Endotracheal Tube Position and X-ray Image Classification: Challenges and Opportunities
\nhttps://www.unboundmedicine.com/medline/citation/28600640/Deep_Convolutional_Neural_Networks_for_Endotracheal_Tube_Position_and_X_ray_Image_Classification:_Challenges_and_Opportunities_
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2Fd839d3e999e4865d51d10abd436ff38e%2FSelection_046.png?generation=1608145809731213&alt=media\")
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2F5c4d5c4573ac8c40b61d57987aeee885%2FSelection_047.png?generation=1608145829261278&alt=media\")
how to read Chest Radiology for ET Tube Position
\nhttps://www.youtube.com/watch?v=gwKwCARKYfw
", "rawMarkdown": "how to read Chest Radiology for ET Tube Position\n\nhttps://www.youtube.com/watch?v=gwKwCARKYfw", "votes": null }, { "id": "1116047", "postDate": "12/16/2020 19:28:13", "content": "https://healthcare-in-europe.com/en/news/deep-learning-software-helps-to-locate-the-carina.html
\n “Detecting Mal-Positioned Endotracheal Tubes in Portable Chest X-ray Images: Comparing Deep Learning Models with Hand-Engineered Approach on Carina Detection.”
quote \" Using this metric on 212 test cases, they foundd the manual algorithm identified the position of the carina in about 74% of cases whereas for the deep learning approach it was 89%.\"
", "rawMarkdown": "https://healthcare-in-europe.com/en/news/deep-learning-software-helps-to-locate-the-carina.html\n “Detecting Mal-Positioned Endotracheal Tubes in Portable Chest X-ray Images: Comparing Deep Learning Models with Hand-Engineered Approach on Carina Detection.”\n\nquote \" Using this metric on 212 test cases, they foundd the manual algorithm identified the position of the carina in about 74% of cases whereas for the deep learning approach it was 89%.\"", "votes": null }, { "id": "1116083", "postDate": "12/16/2020 20:19:44", "content": "Good to see that i am not alone who is already working in this way =)
", "rawMarkdown": "Good to see that i am not alone who is already working in this way =)", "votes": null }, { "id": "1116086", "postDate": "12/16/2020 20:31:45", "content": "\n\nEndotracheal Tube Detection and Segmentation in Chest Radiographs using Synthetic Data
\n
i missed a little, did you suggest a stage where we do segmentation before classification?
", "rawMarkdown": "> Endotracheal Tube Detection and Segmentation in Chest Radiographs using Synthetic Data\n\ni missed a little, did you suggest a stage where we do **segmentation** before **classification**?", "votes": null }, { "id": "1116092", "postDate": "12/16/2020 20:42:04", "content": "one possibilty:
\nencoder \n |--->segemntation head\n |--->classification head\nsegmentation head is used only for training. it is to force the model to learn meaningfully encoding to improve generalisation. segmentation head can be removed at inference.
\n(e.g. mask-rcnn improves results of object detection compared to rcnn only)
of course, there are other possible configurations:
\nencoder ---> segmentation head --->classification head\nin this case classification uses results of segmentation. but has to be used in training and inference
", "rawMarkdown": "one possibilty:\n\n```\nencoder \n |--->segemntation head\n |--->classification head\n\n```\n\nsegmentation head is used only for training. it is to force the model to learn meaningfully encoding to improve generalisation. segmentation head can be removed at inference.\n(e.g. mask-rcnn improves results of object detection compared to rcnn only)\n\nof course, there are other possible configurations:\n\n```\nencoder ---> segmentation head --->classification head\n\n```\n\nin this case classification uses results of segmentation. but has to be used in training and inference", "votes": null }, { "id": "1119387", "postDate": "12/20/2020 02:12:35", "content": "ETT-Net looks alike like U-NET with extra Upsamplings, Global Average Pooling.
", "rawMarkdown": "ETT-Net looks alike like U-NET with extra Upsamplings, Global Average Pooling.", "votes": null }, { "id": "1134160", "postDate": "12/31/2020 22:59:23", "content": "Really glad to see that people have discovered our publications!
\nBTW, the link to 100 pediatric X-rays with pixel wise annotation can be found here https://github.com/xinario/PediatricXray100
\nanother link to a set of adult X-rays (~2000) with synthetic tubes (with corresponding masks) can be found here
\nhttps://github.com/xinario/catheter_detection
Hope these resources could help!
", "rawMarkdown": "Really glad to see that people have discovered our publications!\n\nBTW, the link to 100 pediatric X-rays with pixel wise annotation can be found here https://github.com/xinario/PediatricXray100\n\nanother link to a set of adult X-rays (~2000) with synthetic tubes (with corresponding masks) can be found here \nhttps://github.com/xinario/catheter_detection\n\nHope these resources could help!", "votes": null } ], "comments": [ { "id": 1116018, "author_name": "hengck23", "author_url": "", "post_date": "12/16/2020 18:38:59", "content": "Computer-Aided Assessment of Catheters and Tubes on Radiographs: How Good is Artificial Intelligence for Assessment?
\nhttps://www.researchgate.net/publication/339164106_Computer-Aided_Assessment_of_Catheters_and_Tubes_on_Radiographs_How_Good_is_Artificial_Intelligence_for_Assessment
Accuracy of the 7-8-9 Rule for endotracheal tube placement in the neonate
\nhttps://www.nature.com/articles/7211503
PediatricXray100 dataset
\nhttps://github.com/xinario/PediatricXray1
Deep Convolutional Neural Networks for Endotracheal Tube Position and X-ray Image Classification: Challenges and Opportunities
\nhttps://www.unboundmedicine.com/medline/citation/28600640/Deep_Convolutional_Neural_Networks_for_Endotracheal_Tube_Position_and_X_ray_Image_Classification:_Challenges_and_Opportunities_
how to read Chest Radiology for ET Tube Position
\nhttps://www.youtube.com/watch?v=gwKwCARKYfw
", "votes": null, "replies": [] }, { "id": 1116047, "author_name": "hengck23", "author_url": "", "post_date": "12/16/2020 19:28:13", "content": "https://healthcare-in-europe.com/en/news/deep-learning-software-helps-to-locate-the-carina.html
\n “Detecting Mal-Positioned Endotracheal Tubes in Portable Chest X-ray Images: Comparing Deep Learning Models with Hand-Engineered Approach on Carina Detection.”
quote \" Using this metric on 212 test cases, they foundd the manual algorithm identified the position of the carina in about 74% of cases whereas for the deep learning approach it was 89%.\"
", "votes": null, "replies": [] }, { "id": 1116083, "author_name": "sandorkonya", "author_url": "", "post_date": "12/16/2020 20:19:44", "content": "Good to see that i am not alone who is already working in this way =)
", "votes": null, "replies": [] }, { "id": 1116086, "author_name": "hiramcho", "author_url": "", "post_date": "12/16/2020 20:31:45", "content": "\n\nEndotracheal Tube Detection and Segmentation in Chest Radiographs using Synthetic Data
\n
i missed a little, did you suggest a stage where we do segmentation before classification?
", "votes": null, "replies": [ { "id": 1116092, "author_name": "hengck23", "author_url": "", "post_date": "12/16/2020 20:42:04", "content": "one possibilty:
\nencoder \n |--->segemntation head\n |--->classification head\nsegmentation head is used only for training. it is to force the model to learn meaningfully encoding to improve generalisation. segmentation head can be removed at inference.
\n(e.g. mask-rcnn improves results of object detection compared to rcnn only)
of course, there are other possible configurations:
\nencoder ---> segmentation head --->classification head\nin this case classification uses results of segmentation. but has to be used in training and inference
", "votes": null, "replies": [] } ] }, { "id": 1119387, "author_name": "alincijov", "author_url": "", "post_date": "12/20/2020 02:12:35", "content": "ETT-Net looks alike like U-NET with extra Upsamplings, Global Average Pooling.
", "votes": null, "replies": [] }, { "id": 1134160, "author_name": "xinario", "author_url": "", "post_date": "12/31/2020 22:59:23", "content": "Really glad to see that people have discovered our publications!
\nBTW, the link to 100 pediatric X-rays with pixel wise annotation can be found here https://github.com/xinario/PediatricXray100
\nanother link to a set of adult X-rays (~2000) with synthetic tubes (with corresponding masks) can be found here
\nhttps://github.com/xinario/catheter_detection
Hope these resources could help!
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1116007": "https://arxiv.org/pdf/1908.07170v1.pdf\nEndotracheal Tube Detection and Segmentation in Chest Radiographs using Synthetic Data\n\n\n\n\n\n\n\n\n\n\nhttps://github.com/xinario/catheter_detection", "1116018": "Computer-Aided Assessment of Catheters and Tubes on Radiographs: How Good is Artificial Intelligence for Assessment?\nhttps://www.researchgate.net/publication/339164106_Computer-Aided_Assessment_of_Catheters_and_Tubes_on_Radiographs_How_Good_is_Artificial_Intelligence_for_Assessment\n\n\n\n", "1116020": "Accuracy of the 7-8-9 Rule for endotracheal tube placement in the neonate\nhttps://www.nature.com/articles/7211503", "1116027": "PediatricXray100 dataset\nhttps://github.com/xinario/PediatricXray1\n\n", "1116033": "Deep Convolutional Neural Networks for Endotracheal Tube Position and X-ray Image Classification: Challenges and Opportunities\nhttps://www.unboundmedicine.com/medline/citation/28600640/Deep_Convolutional_Neural_Networks_for_Endotracheal_Tube_Position_and_X_ray_Image_Classification:_Challenges_and_Opportunities_\n\n\n\n", "1116035": "how to read Chest Radiology for ET Tube Position\n\nhttps://www.youtube.com/watch?v=gwKwCARKYfw", "1116047": "https://healthcare-in-europe.com/en/news/deep-learning-software-helps-to-locate-the-carina.html\n “Detecting Mal-Positioned Endotracheal Tubes in Portable Chest X-ray Images: Comparing Deep Learning Models with Hand-Engineered Approach on Carina Detection.”\n\nquote \" Using this metric on 212 test cases, they foundd the manual algorithm identified the position of the carina in about 74% of cases whereas for the deep learning approach it was 89%.\"", "1116083": "Good to see that i am not alone who is already working in this way =)", "1116086": "> Endotracheal Tube Detection and Segmentation in Chest Radiographs using Synthetic Data\n\ni missed a little, did you suggest a stage where we do **segmentation** before **classification**?", "1116092": "one possibilty:\n\n```\nencoder \n |--->segemntation head\n |--->classification head\n\n```\n\nsegmentation head is used only for training. it is to force the model to learn meaningfully encoding to improve generalisation. segmentation head can be removed at inference.\n(e.g. mask-rcnn improves results of object detection compared to rcnn only)\n\nof course, there are other possible configurations:\n\n```\nencoder ---> segmentation head --->classification head\n\n```\n\nin this case classification uses results of segmentation. but has to be used in training and inference", "1119387": "ETT-Net looks alike like U-NET with extra Upsamplings, Global Average Pooling.", "1134160": "Really glad to see that people have discovered our publications!\n\nBTW, the link to 100 pediatric X-rays with pixel wise annotation can be found here https://github.com/xinario/PediatricXray100\n\nanother link to a set of adult X-rays (~2000) with synthetic tubes (with corresponding masks) can be found here \nhttps://github.com/xinario/catheter_detection\n\nHope these resources could help!" }, "source": "meta" }