![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F3612572%2Fec93b41e4fda3abbb6de96f423905c36%2Fle-minh-phuong-niH7Z81S44g-unsplash.jpg?generation=1607987568127029&alt=media\")
\nPhoto by Le Minh Phuong on Unsplash
\nPhoto by Le Minh Phuong on Unsplash
Hi Kagglers !
\nYet another competition on image recognition in the medical imaging field… And OSIIC Pulmonary Fibrosis has just finished weeks ago 😄
\nFor those who want to get started, I propose here a small, non-exhaustive literature & resources review that should help give you some ideas.
\nPrevious Kaggle competitions in medical imaging classification :
\nMore specific works around this topic :
\nThese works are actually really specific, and quite complex for a novice. I would recommend starting with simpler models.
\nI know it is not exactly the same kind of competition, but yet I think it could be a good idea to take a look at what is happening in the (running) HuBMAP - Hacking the Kidney competition.
\nWhy ?
\nBecause I have a feeling (that's a mere suggestion, I didn't try anything yet) that in this competition you should build some model able to focus on a very specific part of the image - where the catheter is - and analyze that part. The rest of the image is actually of no importance in itself. That's the same procedure as in Hack the Kidney, where you have to identify and focus on some small part of the image.
\nGood luck to everyone, and happy kaggling !
", "messages": [ { "id": 1112826, "postDate": "2020-12-14T23:32:09.197Z", "content": "
\nPhoto by Le Minh Phuong on Unsplash
Hi Kagglers !
\nYet another competition on image recognition in the medical imaging field… And OSIIC Pulmonary Fibrosis has just finished weeks ago 😄
\nFor those who want to get started, I propose here a small, non-exhaustive literature & resources review that should help give you some ideas.
\nPrevious Kaggle competitions in medical imaging classification :
\nMore specific works around this topic :
\nThese works are actually really specific, and quite complex for a novice. I would recommend starting with simpler models.
\nI know it is not exactly the same kind of competition, but yet I think it could be a good idea to take a look at what is happening in the (running) HuBMAP - Hacking the Kidney competition.
\nWhy ?
\nBecause I have a feeling (that's a mere suggestion, I didn't try anything yet) that in this competition you should build some model able to focus on a very specific part of the image - where the catheter is - and analyze that part. The rest of the image is actually of no importance in itself. That's the same procedure as in Hack the Kidney, where you have to identify and focus on some small part of the image.
\nGood luck to everyone, and happy kaggling !
", "rawMarkdown": "\nPhoto by Le Minh Phuong on Unsplash\n\nHi Kagglers !\n\nYet another competition on image recognition in the medical imaging field... And [OSIIC Pulmonary Fibrosis](https://www.kaggle.com/c/osic-pulmonary-fibrosis-progression) has just finished weeks ago 😄\n\nFor those who want to get started, I propose here a small, non-exhaustive literature & resources review that should help give you some ideas.\n\n**Previous Kaggle competitions in medical imaging classification :**\n- [OSIIC Pulmonary Fibrosis](https://www.kaggle.com/c/osic-pulmonary-fibrosis-progression)\n- [RSNA Pneumonia Detection Challenge](https://www.kaggle.com/c/rsna-pneumonia-detection-challenge)\nThese 2 first ones are on lungs as well, what could be helpful although nice resources can also be found in further related topics.\n- [SIIM-ISIC Melanoma Classification](https://www.kaggle.com/c/siim-isic-melanoma-classification)\n- [Histopathologic Cancer Detection](https://www.kaggle.com/c/histopathologic-cancer-detection/overview/evaluation)\n\n**More specific works around this topic :**\n- [Deep-Learning System for Fully-Automated Peripherally Inserted Central Catheter Tip Detection](https://www.researchgate.net/publication/320249406_A_Deep-Learning_System_for_Fully-Automated_Peripherally_Inserted_Central_Catheter_PICC_Tip_Detection)\n- [Computer-Aided Assessment of Catheters and Tubes on Radiographs](https://arxiv.org/pdf/2002.03413.pdf)\n- [Real-time assessment of catheter contactand orientation using an integrated opticalcoherence tomography cardiac ablation catheter](https://www.osapublishing.org/DirectPDFAccess/E39F6818-F192-2F75-03818FB735B75F90_412217/ao-58-14-3823.pdf?da=1&id=412217&seq=0&mobile=no)\n\nThese works are actually really specific, and quite complex for a novice. I would recommend starting with simpler models.\n\nI know it is not exactly the same kind of competition, but yet I think it could be a good idea to take a look at **what is happening in the (running) [HuBMAP - Hacking the Kidney](https://www.kaggle.com/c/hubmap-kidney-segmentation) competition**. \n\nWhy ? \n\nBecause I have a feeling (that's a mere suggestion, I didn't try anything yet) that in this competition you should build **some model able to focus on a very specific part of the image** - where the catheter is - and analyze that part. The rest of the image is actually of no importance in itself. That's the same procedure as in Hack the Kidney, where you have to identify and focus on some small part of the image.\n\nGood luck to everyone, and happy kaggling !", "votes": 26 } ], "comments": [], "raw_markdown_by_id": { "1112826": "\nPhoto by Le Minh Phuong on Unsplash\n\nHi Kagglers !\n\nYet another competition on image recognition in the medical imaging field... And [OSIIC Pulmonary Fibrosis](https://www.kaggle.com/c/osic-pulmonary-fibrosis-progression) has just finished weeks ago 😄\n\nFor those who want to get started, I propose here a small, non-exhaustive literature & resources review that should help give you some ideas.\n\n**Previous Kaggle competitions in medical imaging classification :**\n- [OSIIC Pulmonary Fibrosis](https://www.kaggle.com/c/osic-pulmonary-fibrosis-progression)\n- [RSNA Pneumonia Detection Challenge](https://www.kaggle.com/c/rsna-pneumonia-detection-challenge)\nThese 2 first ones are on lungs as well, what could be helpful although nice resources can also be found in further related topics.\n- [SIIM-ISIC Melanoma Classification](https://www.kaggle.com/c/siim-isic-melanoma-classification)\n- [Histopathologic Cancer Detection](https://www.kaggle.com/c/histopathologic-cancer-detection/overview/evaluation)\n\n**More specific works around this topic :**\n- [Deep-Learning System for Fully-Automated Peripherally Inserted Central Catheter Tip Detection](https://www.researchgate.net/publication/320249406_A_Deep-Learning_System_for_Fully-Automated_Peripherally_Inserted_Central_Catheter_PICC_Tip_Detection)\n- [Computer-Aided Assessment of Catheters and Tubes on Radiographs](https://arxiv.org/pdf/2002.03413.pdf)\n- [Real-time assessment of catheter contactand orientation using an integrated opticalcoherence tomography cardiac ablation catheter](https://www.osapublishing.org/DirectPDFAccess/E39F6818-F192-2F75-03818FB735B75F90_412217/ao-58-14-3823.pdf?da=1&id=412217&seq=0&mobile=no)\n\nThese works are actually really specific, and quite complex for a novice. I would recommend starting with simpler models.\n\nI know it is not exactly the same kind of competition, but yet I think it could be a good idea to take a look at **what is happening in the (running) [HuBMAP - Hacking the Kidney](https://www.kaggle.com/c/hubmap-kidney-segmentation) competition**. \n\nWhy ? \n\nBecause I have a feeling (that's a mere suggestion, I didn't try anything yet) that in this competition you should build **some model able to focus on a very specific part of the image** - where the catheter is - and analyze that part. The rest of the image is actually of no importance in itself. That's the same procedure as in Hack the Kidney, where you have to identify and focus on some small part of the image.\n\nGood luck to everyone, and happy kaggling !" } }