On behalf of the Open Source Imaging Consortium and the far too many patients all over the world, we are grateful you have chosen to participate! Too often, Fibrotic Lung Diseases are overlooked and underfunded. It’s not right or fair to those battling the disease or to the many who have lost their fight. This disease kills as many people as breast cancer every year, but most have never heard of it. We truly believe your efforts will help us change that, by not only raising awareness, but more importantly, by bringing your expertise and innovative ideas to the problem at hand.
\n\nWhen you talk to Pulmonary Fibrosis experts, the one thing they all agree on is that this disease is heterogenous, meaning no patient experience is the same in disease progression. Currently, there is no way to help patients understand the path of their disease or what is to come. It’s beyond difficult for the patients, caregivers AND clinicians. You have a true opportunity with this challenge to create a deep learning algorithm based on data and image analysis.
\n\nThroughout the competition we will have Pulmonology Clinicians, Radiologists and Computer Science/Machine Learning experts monitoring the forums and standing by to answer any questions you may have. We want to help in any way we can.
\n\nWe have been told this may be perceived as a difficult challenge. For some, that may be true, for some, maybe not. However, we can guarantee your efforts will be truly meaningful to those battling this disease. Whether you win or you don’t, you will make a difference. We wish you the best of luck.
\n\nElizabeth Estes\nOSIC Executive Director (on behalf of the OSIC Team)
\n\n\n", "messages": [ { "id": 919234, "postDate": "2020-07-07T19:08:25.577Z", "content": "Remember: Kaggle is for everyone. Whether you're teaming up or sharing tips in the competition forum, we expect everyone to follow our Kaggle community guidelines.
\n
On behalf of the Open Source Imaging Consortium and the far too many patients all over the world, we are grateful you have chosen to participate! Too often, Fibrotic Lung Diseases are overlooked and underfunded. It’s not right or fair to those battling the disease or to the many who have lost their fight. This disease kills as many people as breast cancer every year, but most have never heard of it. We truly believe your efforts will help us change that, by not only raising awareness, but more importantly, by bringing your expertise and innovative ideas to the problem at hand.
\n\nWhen you talk to Pulmonary Fibrosis experts, the one thing they all agree on is that this disease is heterogenous, meaning no patient experience is the same in disease progression. Currently, there is no way to help patients understand the path of their disease or what is to come. It’s beyond difficult for the patients, caregivers AND clinicians. You have a true opportunity with this challenge to create a deep learning algorithm based on data and image analysis.
\n\nThroughout the competition we will have Pulmonology Clinicians, Radiologists and Computer Science/Machine Learning experts monitoring the forums and standing by to answer any questions you may have. We want to help in any way we can.
\n\nWe have been told this may be perceived as a difficult challenge. For some, that may be true, for some, maybe not. However, we can guarantee your efforts will be truly meaningful to those battling this disease. Whether you win or you don’t, you will make a difference. We wish you the best of luck.
\n\nElizabeth Estes\nOSIC Executive Director (on behalf of the OSIC Team)
\n\n\n", "rawMarkdown": "On behalf of the Open Source Imaging Consortium and the far too many patients all over the world, we are grateful you have chosen to participate! Too often, Fibrotic Lung Diseases are overlooked and underfunded. It’s not right or fair to those battling the disease or to the many who have lost their fight. This disease kills as many people as breast cancer every year, but most have never heard of it. We truly believe your efforts will help us change that, by not only raising awareness, but more importantly, by bringing your expertise and innovative ideas to the problem at hand. \n\nWhen you talk to Pulmonary Fibrosis experts, the one thing they all agree on is that this disease is heterogenous, meaning no patient experience is the same in disease progression. Currently, there is no way to help patients understand the path of their disease or what is to come. It’s beyond difficult for the patients, caregivers AND clinicians. You have a true opportunity with this challenge to create a deep learning algorithm based on data and image analysis.\n\nThroughout the competition we will have Pulmonology Clinicians, Radiologists and Computer Science/Machine Learning experts monitoring the forums and standing by to answer any questions you may have. We want to help in any way we can.\n\nWe have been told this may be perceived as a difficult challenge. For some, that may be true, for some, maybe not. However, we can guarantee your efforts will be truly meaningful to those battling this disease. Whether you win or you don’t, you will make a difference. We wish you the best of luck. \n\nElizabeth Estes\nOSIC Executive Director (on behalf of the OSIC Team)\n\n\n> **Remember**: Kaggle is for everyone. Whether you're teaming up or sharing tips in the competition forum, we expect everyone to follow our [Kaggle community guidelines](https://www.kaggle.com/community-guidelines).\n", "votes": 45 }, { "id": 1014256, "postDate": "2020-09-17T09:49:50.153Z", "content": "Remember: Kaggle is for everyone. Whether you're teaming up or sharing tips in the competition forum, we expect everyone to follow our Kaggle community guidelines.
\n
Hello, could you explain to me please how the amount of DCM images is related to the patients' weekly FVC scan?
\nI mean how many dcm images per time are taken? for example, patient ID00007637202177411956430 FVC scans start at week -4 up to week 57 (total weeks = 62) with 9 scans and apparent random intervals (week: -4, 5, 7, 9 11, 17….) and it has a total of 30 dcm images(.5 images per week, or 3.3 images per scan), whereas next patient ID00009637202177434476278 FVC scans start at week 8 up to week 60 (total weeks = 52)with 9 scans and random intervals as well(like all the patients) and has a total of 394 dcm images(7.6 images per week or 43.7 images per scan ), as far as I know, this random pattern continues and I don't know how to relate the images to the weekly FVC scan. If you could give me an insight of this would be of great help.
\nthanks
Multiple dcm images represent slices. CT imaging produces a 3D volume for each scan, this volume consists of 2D slices, each slice is a dcm image in our case. In other words, by stacking 2D images, you get the volume (case or patient). They're all taken at the same time. I hope this helps.
\nCT scan has been done once (at week 0), while FVC measurements are recorded on several weeks. We assume that the week when the CT scan has been taken is week 0. So, positive week value in the CSV file indicates that this FVC has been recorded after the CT scan. Negative values mean that this FVC has been recorded before the scan. For example: a week value of -2 means that the corresponding FVC value has been recorded two weeks before the CT scan. Hope that helps.
Not really, the algorithm needs to be robust against it as it is likely to happen at test time (having a variable no of slices).
", "rawMarkdown": "Not really, the algorithm needs to be robust against it as it is likely to happen at test time (having a variable no of slices)." } ] }, { "id": 994945, "postDate": "2020-09-02T04:51:24.963Z", "content": "What is the specific meaning of Percent in train.csv? Can you give me an example? Thank you.
\nMy understanding is the proportion of FVC of healthy people of the same age as the patient. For example, the FVC of the patient with ID00007637202177411956430 is 2315, Percent = the FVC of this patient divided by the FVC of healthy people with the same conditions = 58.2536487166583.
\nIs this understanding correct?
I can see that the minimum week number is -12 and the maximum is 133 in the sample_submission.csv. So, are the week numbers guaranteed to be in this range only in the private test set too ? In other words, do we need to predict for this range only ?
Yes, you provide predictions for this range only. This range covers all possible weeks in the hidden test set.
", "rawMarkdown": "Yes, you provide predictions for this range only. This range covers all possible weeks in the hidden test set.", "votes": 2 }, { "id": 995415, "postDate": "2020-09-02T12:30:16.660Z", "content": "Okay Thanks!
\nJust to make sure, I'd additionally like to ask that the Sex column will contain Male or Female values only right ?
Yes. You are welcome.
", "rawMarkdown": "Yes. You are welcome.", "votes": 1 }, { "id": 1019218, "postDate": "2020-09-20T09:28:56.260Z", "content": "@ahmedhshahin
\ncan it be fare assumption that final three weeks of patients visit may be either after initial visit at which FVC measured called Base Week or including the Base Week ?
\nSay suppose Base week in Test is 6 ,its unlikely that it could be the finial visit for a patient and he would no more have need to visit..
Yes, all patients had more than 4 visits.
", "rawMarkdown": "Yes, all patients had more than 4 visits." } ] }, { "id": 994280, "postDate": "2020-09-01T13:54:12.330Z", "content": "Are all patients diagnosed IPF by biopsy??
", "rawMarkdown": "Are all patients diagnosed IPF by biopsy??", "votes": 1 }, { "id": 951936, "postDate": "2020-07-30T13:59:59.790Z", "content": "Can we train the model on Notebooks in kaggle? The dataset is very large!
", "rawMarkdown": "Can we train the model on Notebooks in kaggle? The dataset is very large!", "votes": 1, "replies": [ { "id": 951971, "postDate": "2020-07-30T14:26:04.037Z", "content": "yes
", "rawMarkdown": "yes", "votes": 1 } ] }, { "id": 944108, "postDate": "2020-07-24T20:48:47.500Z", "content": "Very interesting initiative!\nIndeed, the importance of Fibrotic Lung Diseases is even more obvious in these challenging times when the heterogeneous group of diseases has been supplemented with COVID 19. Good luck to all competitors!
", "rawMarkdown": "Very interesting initiative!\nIndeed, the importance of Fibrotic Lung Diseases is even more obvious in these challenging times when the heterogeneous group of diseases has been supplemented with COVID 19. Good luck to all competitors!", "votes": 2 }, { "id": 921642, "postDate": "2020-07-09T12:51:22.127Z", "content": "Can you elaborate more about confidence in submission file
", "rawMarkdown": "Can you elaborate more about confidence in submission file", "votes": 2, "replies": [ { "id": 921664, "postDate": "2020-07-09T13:10:59.597Z", "content": "Hi Yasr,\nI would advise you to check the evaluation page (if you haven't seen it already):\nhttps://www.kaggle.com/c/osic-pulmonary-fibrosis-progression/overview/evaluation
\n\nThe idea is basically that you are asked to predict, for each week, the FVC prediction and model confidence. Model confidence is the standard deviation of the prediction that reflects how confident is the model about its prediction. The reason why we incorporate confidence in the cycle is that medical applications are quite critical and we need to know the model confidence about its prediction in such applications.\nI hope this helps.
", "rawMarkdown": "Hi Yasr,\nI would advise you to check the evaluation page (if you haven't seen it already):\nhttps://www.kaggle.com/c/osic-pulmonary-fibrosis-progression/overview/evaluation\n\nThe idea is basically that you are asked to predict, for each week, the FVC prediction and model confidence. Model confidence is the standard deviation of the prediction that reflects how confident is the model about its prediction. The reason why we incorporate confidence in the cycle is that medical applications are quite critical and we need to know the model confidence about its prediction in such applications.\nI hope this helps.", "votes": 5 }, { "id": 921696, "postDate": "2020-07-09T13:38:32.010Z", "content": "Thanks for your support,i understood the concept.
", "rawMarkdown": "Thanks for your support,i understood the concept." } ] }, { "id": 1082566, "postDate": "2020-11-18T02:52:15.617Z", "content": "Hello Carmela,
\nI am just curious about this competition winners, I want to know about their Paper submission and code Evaluation.
\nIf you have access and authority to share their work that will be great.
Please let me know about top 3 submissions.
\nThank you.
\nRegards,
\nSumanth N.
I have the complete benefits of the competer in reserve. I joined the thread for joining a team, but did found one. What can I do, what am I good for.
", "rawMarkdown": "I have the complete benefits of the competer in reserve. I joined the thread for joining a team, but did found one. What can I do, what am I good for." }, { "id": 948687, "postDate": "2020-07-28T06:39:10.020Z", "content": "👌 yes
", "rawMarkdown": "👌 yes" }, { "id": 948447, "postDate": "2020-07-28T00:45:54.213Z", "content": "Hello, will all the test patient id include in train dataset in real submission test?
", "rawMarkdown": "Hello, will all the test patient id include in train dataset in real submission test? ", "replies": [ { "id": 948459, "postDate": "2020-07-28T01:13:21.237Z", "content": "The test is a subset of the train set
", "rawMarkdown": "The test is a subset of the train set", "votes": -2 } ] } ], "comments": [ { "id": 1014256, "author_name": "alemazav", "author_url": "", "post_date": "2020-09-17T09:49:50.153000", "content": "Hello, could you explain to me please how the amount of DCM images is related to the patients' weekly FVC scan?
\nI mean how many dcm images per time are taken? for example, patient ID00007637202177411956430 FVC scans start at week -4 up to week 57 (total weeks = 62) with 9 scans and apparent random intervals (week: -4, 5, 7, 9 11, 17….) and it has a total of 30 dcm images(.5 images per week, or 3.3 images per scan), whereas next patient ID00009637202177434476278 FVC scans start at week 8 up to week 60 (total weeks = 52)with 9 scans and random intervals as well(like all the patients) and has a total of 394 dcm images(7.6 images per week or 43.7 images per scan ), as far as I know, this random pattern continues and I don't know how to relate the images to the weekly FVC scan. If you could give me an insight of this would be of great help.
\nthanks
Multiple dcm images represent slices. CT imaging produces a 3D volume for each scan, this volume consists of 2D slices, each slice is a dcm image in our case. In other words, by stacking 2D images, you get the volume (case or patient). They're all taken at the same time. I hope this helps.
\nCT scan has been done once (at week 0), while FVC measurements are recorded on several weeks. We assume that the week when the CT scan has been taken is week 0. So, positive week value in the CSV file indicates that this FVC has been recorded after the CT scan. Negative values mean that this FVC has been recorded before the scan. For example: a week value of -2 means that the corresponding FVC value has been recorded two weeks before the CT scan. Hope that helps.
Not really, the algorithm needs to be robust against it as it is likely to happen at test time (having a variable no of slices).
", "votes": 0, "replies": [] } ] }, { "id": 994945, "author_name": "ybh2018", "author_url": "", "post_date": "2020-09-02T04:51:24.963000", "content": "What is the specific meaning of Percent in train.csv? Can you give me an example? Thank you.
\nMy understanding is the proportion of FVC of healthy people of the same age as the patient. For example, the FVC of the patient with ID00007637202177411956430 is 2315, Percent = the FVC of this patient divided by the FVC of healthy people with the same conditions = 58.2536487166583.
\nIs this understanding correct?
I can see that the minimum week number is -12 and the maximum is 133 in the sample_submission.csv. So, are the week numbers guaranteed to be in this range only in the private test set too ? In other words, do we need to predict for this range only ?
Yes, you provide predictions for this range only. This range covers all possible weeks in the hidden test set.
", "votes": 2, "replies": [] }, { "id": 995415, "author_name": "Kishore Badyakar", "author_url": "", "post_date": "2020-09-02T12:30:16.660000", "content": "Okay Thanks!
\nJust to make sure, I'd additionally like to ask that the Sex column will contain Male or Female values only right ?
Yes. You are welcome.
", "votes": 1, "replies": [] }, { "id": 1019218, "author_name": "Jaideep", "author_url": "", "post_date": "2020-09-20T09:28:56.260000", "content": "@ahmedhshahin
\ncan it be fare assumption that final three weeks of patients visit may be either after initial visit at which FVC measured called Base Week or including the Base Week ?
\nSay suppose Base week in Test is 6 ,its unlikely that it could be the finial visit for a patient and he would no more have need to visit..
Yes, all patients had more than 4 visits.
", "votes": 0, "replies": [] } ] }, { "id": 994280, "author_name": "Jsato", "author_url": "", "post_date": "2020-09-01T13:54:12.330000", "content": "Are all patients diagnosed IPF by biopsy??
", "votes": 1, "replies": [] }, { "id": 951936, "author_name": "Mark", "author_url": "", "post_date": "2020-07-30T13:59:59.790000", "content": "Can we train the model on Notebooks in kaggle? The dataset is very large!
", "votes": 1, "replies": [ { "id": 951971, "author_name": "Ahmed Shahin", "author_url": "", "post_date": "2020-07-30T14:26:04.037000", "content": "yes
", "votes": 1, "replies": [] } ] }, { "id": 944108, "author_name": "Astghik Baghdasaryan", "author_url": "", "post_date": "2020-07-24T20:48:47.500000", "content": "Very interesting initiative!\nIndeed, the importance of Fibrotic Lung Diseases is even more obvious in these challenging times when the heterogeneous group of diseases has been supplemented with COVID 19. Good luck to all competitors!
", "votes": 2, "replies": [] }, { "id": 921642, "author_name": "Yasr", "author_url": "", "post_date": "2020-07-09T12:51:22.127000", "content": "Can you elaborate more about confidence in submission file
", "votes": 2, "replies": [ { "id": 921664, "author_name": "Ahmed Shahin", "author_url": "", "post_date": "2020-07-09T13:10:59.597000", "content": "Hi Yasr,\nI would advise you to check the evaluation page (if you haven't seen it already):\nhttps://www.kaggle.com/c/osic-pulmonary-fibrosis-progression/overview/evaluation
\n\nThe idea is basically that you are asked to predict, for each week, the FVC prediction and model confidence. Model confidence is the standard deviation of the prediction that reflects how confident is the model about its prediction. The reason why we incorporate confidence in the cycle is that medical applications are quite critical and we need to know the model confidence about its prediction in such applications.\nI hope this helps.
", "votes": 5, "replies": [] }, { "id": 921696, "author_name": "Yasr", "author_url": "", "post_date": "2020-07-09T13:38:32.010000", "content": "Thanks for your support,i understood the concept.
", "votes": 0, "replies": [] } ] }, { "id": 1082566, "author_name": "nacharam sumanth", "author_url": "", "post_date": "2020-11-18T02:52:15.617000", "content": "Hello Carmela,
\nI am just curious about this competition winners, I want to know about their Paper submission and code Evaluation.
\nIf you have access and authority to share their work that will be great.
Please let me know about top 3 submissions.
\nThank you.
\nRegards,
\nSumanth N.
I have the complete benefits of the competer in reserve. I joined the thread for joining a team, but did found one. What can I do, what am I good for.
", "votes": 0, "replies": [] }, { "id": 948687, "author_name": "import tensorflow as plt", "author_url": "", "post_date": "2020-07-28T06:39:10.020000", "content": "👌 yes
", "votes": 0, "replies": [] }, { "id": 948447, "author_name": "", "author_url": "", "post_date": "2020-07-28T00:45:54.213000", "content": "Hello, will all the test patient id include in train dataset in real submission test?
", "votes": 0, "replies": [ { "id": 948459, "author_name": "rpsantosa_kaggle", "author_url": "", "post_date": "2020-07-28T01:13:21.237000", "content": "The test is a subset of the train set
", "votes": -2, "replies": [] } ] } ], "raw_markdown_by_id": { "919234": "On behalf of the Open Source Imaging Consortium and the far too many patients all over the world, we are grateful you have chosen to participate! Too often, Fibrotic Lung Diseases are overlooked and underfunded. It’s not right or fair to those battling the disease or to the many who have lost their fight. This disease kills as many people as breast cancer every year, but most have never heard of it. We truly believe your efforts will help us change that, by not only raising awareness, but more importantly, by bringing your expertise and innovative ideas to the problem at hand. \n\nWhen you talk to Pulmonary Fibrosis experts, the one thing they all agree on is that this disease is heterogenous, meaning no patient experience is the same in disease progression. Currently, there is no way to help patients understand the path of their disease or what is to come. It’s beyond difficult for the patients, caregivers AND clinicians. You have a true opportunity with this challenge to create a deep learning algorithm based on data and image analysis.\n\nThroughout the competition we will have Pulmonology Clinicians, Radiologists and Computer Science/Machine Learning experts monitoring the forums and standing by to answer any questions you may have. We want to help in any way we can.\n\nWe have been told this may be perceived as a difficult challenge. For some, that may be true, for some, maybe not. However, we can guarantee your efforts will be truly meaningful to those battling this disease. Whether you win or you don’t, you will make a difference. We wish you the best of luck. \n\nElizabeth Estes\nOSIC Executive Director (on behalf of the OSIC Team)\n\n\n> **Remember**: Kaggle is for everyone. Whether you're teaming up or sharing tips in the competition forum, we expect everyone to follow our [Kaggle community guidelines](https://www.kaggle.com/community-guidelines).\n", "1014256": "Hello, could you explain to me please how the amount of DCM images is related to the patients' weekly FVC scan?\nI mean how many dcm images per time are taken? for example, patient ID00007637202177411956430 FVC scans start at week -4 up to week 57 (total weeks = 62) with 9 scans and apparent random intervals (week: -4, 5, 7, 9 11, 17....) and it has a total of 30 dcm images(.5 images per week, or 3.3 images per scan), whereas next patient ID00009637202177434476278 FVC scans start at week 8 up to week 60 (total weeks = 52)with 9 scans and random intervals as well(like all the patients) and has a total of 394 dcm images(7.6 images per week or 43.7 images per scan ), as far as I know, this random pattern continues and I don't know how to relate the images to the weekly FVC scan. If you could give me an insight of this would be of great help.\nthanks ", "994945": "What is the specific meaning of Percent in train.csv? Can you give me an example? Thank you.\nMy understanding is the proportion of FVC of healthy people of the same age as the patient. For example, the FVC of the patient with ID00007637202177411956430 is 2315, Percent = the FVC of this patient divided by the FVC of healthy people with the same conditions = 58.2536487166583.\nIs this understanding correct?", "995347": "I can see that the minimum week number is `-12` and the maximum is `133` in the `sample_submission.csv`. So, are the week numbers guaranteed to be in this range only in the private test set too ? In other words, do we need to predict for this range only ?", "994280": "Are all patients diagnosed IPF by biopsy??", "951936": "Can we train the model on Notebooks in kaggle? The dataset is very large!", "944108": "Very interesting initiative!\nIndeed, the importance of Fibrotic Lung Diseases is even more obvious in these challenging times when the heterogeneous group of diseases has been supplemented with COVID 19. Good luck to all competitors!", "921642": "Can you elaborate more about confidence in submission file", "1082566": "Hello Carmela,\n\nI am just curious about this competition winners, I want to know about their Paper submission and code Evaluation.\nIf you have access and authority to share their work that will be great.\n\nPlease let me know about top 3 submissions.\n\nThank you.\n\nRegards,\nSumanth N. ", "1031858": "I have the complete benefits of the competer in reserve. I joined the thread for joining a team, but did found one. What can I do, what am I good for.", "948687": "👌 yes", "948447": "Hello, will all the test patient id include in train dataset in real submission test? " } }