![\"ct](\"https://i.ibb.co/V361Whr/ID00228637202259965313869.gif\")
As we know baseline CT scans are acquired at Week 0, but I'm still struggling to understand what exactly they are capturing.
\n\nIn the image below, the lungs are expanding slowly as the black area grows larger. Patient is probably inhaling during that time. After that, the black area is disappearing rapidly and I assumed this is when the patient was exhaling.
\n\n
My questions are:\n* Is it correct that one CT scan is a single inhale/exhale process for one patient?\n* If so, is it a good idea make predictions based on a single breathe?\n* Why organizers decided to share only one CT scan per patient? Do they change at all on different weeks?
", "messages": [ { "id": "930060", "postDate": "07/15/2020 07:00:39", "content": "As we know baseline CT scans are acquired at Week 0, but I'm still struggling to understand what exactly they are capturing.
\n\nIn the image below, the lungs are expanding slowly as the black area grows larger. Patient is probably inhaling during that time. After that, the black area is disappearing rapidly and I assumed this is when the patient was exhaling.
\n\n
My questions are:\n* Is it correct that one CT scan is a single inhale/exhale process for one patient?\n* If so, is it a good idea make predictions based on a single breathe?\n* Why organizers decided to share only one CT scan per patient? Do they change at all on different weeks?
", "rawMarkdown": "As we know baseline CT scans are acquired at Week 0, but I'm still struggling to understand what exactly they are capturing.\n\nIn the image below, the lungs are expanding slowly as the black area grows larger. Patient is probably inhaling during that time. After that, the black area is disappearing rapidly and I assumed this is when the patient was exhaling. \n\n\n\nMy questions are:\n* Is it correct that one CT scan is a single inhale/exhale process for one patient?\n* If so, is it a good idea make predictions based on a single breathe?\n* Why organizers decided to share only one CT scan per patient? Do they change at all on different weeks?", "votes": null }, { "id": "930089", "postDate": "07/15/2020 07:30:43", "content": "\n\n> Is it correct that one CT scan is a single inhale/exhale process for one patient?
\n\nthe CT's are acquired within a single breath-hold maneuver!
\n\nThe patient inhales deeply and the scan goes from bottom to top (usually) or from top to bottom (less good).\nPatients with impaired lung-function cannot hold their breath long enough, even though that the acquisition is nowadays with multi slice CT very rapid, only few seconds (however it was not always the case).\nThis is why the bottom-up approach is considered to be better because if you breathe, the lower part of the chest moves more than the upper part.
\n\nIf the patient cannot hold the breath anymore, there will be motion artifacts on the CT (due to the nature of acquisition). In the reconstruction of the image discontinued or doubled contours appear.
\n\nExample for this here \n the 9th figure, (https://www.researchgate.net/publication/278670308_Clinical_and_Forensic_Radiology_Are_Not_the_Same/figures)
\n\nHowever the scan above is one of the nicely acquired ones, without any motion artifact.
\n\n> If so, is it a good idea make predictions based on a single breathe?
\n\nExposing the patient to further ionizing radiation to obtain another CT scan in expiration imposes elevated hazard. The additional value is however known, in special cases also scans in expiration are acquired, but not regularly, not by every patient.
", "rawMarkdown": "gunesevitan,\n\n> Is it correct that one CT scan is a single inhale/exhale process for one patient?\n\n\nthe CT's are acquired within a single breath-hold maneuver! \n\nThe patient inhales deeply and the scan goes from bottom to top (usually) or from top to bottom (less good).\nPatients with impaired lung-function cannot hold their breath long enough, even though that the acquisition is nowadays with multi slice CT very rapid, only few seconds (however it was not always the case).\nThis is why the bottom-up approach is considered to be better because if you breathe, the lower part of the chest moves more than the upper part. \n\nIf the patient cannot hold the breath anymore, there will be motion artifacts on the CT (due to the nature of acquisition). In the reconstruction of the image discontinued or doubled contours appear. \n\nExample for this [here](https://www.researchgate.net/profile/Michael_Thali/publication/278670308/figure/fig5/AS:390171573932044@1470035560708/Motion-artifacts-in-radiology-a-Two-diaphragms-on-each-side-due-to-respiration-during_W640.jpg) \n the 9th figure, (https://www.researchgate.net/publication/278670308_Clinical_and_Forensic_Radiology_Are_Not_the_Same/figures)\n\nHowever the scan above is one of the nicely acquired ones, without any motion artifact.\n\n> If so, is it a good idea make predictions based on a single breathe?\n\nExposing the patient to further ionizing radiation to obtain another CT scan in expiration imposes elevated hazard. The additional value is however known, in special cases also scans in expiration are acquired, but not regularly, not by every patient.", "votes": null }, { "id": "930104", "postDate": "07/15/2020 07:45:41", "content": "Thanks for the quick answer @sandorkonya. This explains a lot. If I understood correctly, most of the information lies within the slices when the patients are fully inhaled, but that sudden burst of exhale is also giving us decent information about lung function.
", "rawMarkdown": "Thanks for the quick answer @sandorkonya. This explains a lot. If I understood correctly, most of the information lies within the slices when the patients are fully inhaled, but that sudden burst of exhale is also giving us decent information about lung function.", "votes": null }, { "id": "930172", "postDate": "07/15/2020 08:40:48", "content": "\n\nif such motion artifacts are present, it means, that the patient could not hold the breath for the duration of the scan, and indeed, that could signalise a functional change (or status).
\n\nTo meaningfully be able to correlate this, the scan time information about the acquisition itself is needed!
\n\nBy a single detector CT it is pretty easy to calculate the scan time if we know:\nTable Speed, Table Feed per Rotation, Spiral Pitch Factor, and the acquisition length, but with multi detector CT it gets somewhat trickier...
\n\nThere is however \"Performed Procedure Step Start Time Attribute\" for each slice (not present in the current dataset) that could give us exactly the duration of the scan.
", "rawMarkdown": "gunesevitan ,\n\nif such motion artifacts are present, it means, that the patient could not hold the breath for the duration of the scan, and indeed, that could signalise a functional change (or status).\n\nTo meaningfully be able to correlate this, the scan time information about the acquisition itself is needed! \n\nBy a single detector CT it is pretty easy to calculate the scan time if we know:\nTable Speed, Table Feed per Rotation, Spiral Pitch Factor, and the acquisition length, but with multi detector CT it gets somewhat trickier...\n\nThere is however \"Performed Procedure Step Start Time Attribute\" for each slice (not present in the current dataset) that could give us exactly the duration of the scan.", "votes": null }, { "id": "930337", "postDate": "07/15/2020 11:22:53", "content": "The change in lung volume are not related to inhale/exhale in most of these cases. As mentioned, these are slices from top to bottom (or bottom to top) of the chest (from the neck to the belly). The lungs are smaller at the top of the chest than they are at the bottom of the chest.
\n\nIt's like slicing a loaf of bread and looking at each slice. Stack them back together and you have a 3D model of the lung.
", "rawMarkdown": "The change in lung volume are not related to inhale/exhale in most of these cases. As mentioned, these are slices from top to bottom (or bottom to top) of the chest (from the neck to the belly). The lungs are smaller at the top of the chest than they are at the bottom of the chest.\n\nIt's like slicing a loaf of bread and looking at each slice. Stack them back together and you have a 3D model of the lung.", "votes": null }, { "id": "930701", "postDate": "07/15/2020 16:50:57", "content": "@sandorkonya @richardepstein Thanks for the answers. It may sound dumb but do those CT scans are taken from a fixed angle?
", "rawMarkdown": "sandorkonya @richardepstein Thanks for the answers. It may sound dumb but do those CT scans are taken from a fixed angle?", "votes": null }, { "id": "930732", "postDate": "07/15/2020 17:13:57", "content": "\n\nTake a look at this video, it shows the basic principle of the data collection and the backprojection, reconstruction of the \"volumen\" you see:\nhttps://www.youtube.com/watch?v=rKh_XIpsuc4
\n\nAs you see, the gantry rotates constantly around the patient and makes projections from every angle... these are then reconstructed (with filtered backprojection or other methods).
\n\nHere a short vid about the multislice CT (no sound though, but the slices are enough):\nhttps://youtu.be/Eny8Sev1E88?t=48
\n\nBy lung scans the gantry (the ring around the patient) is perpendicular to the axis Z, only by skull examination is it tilted to somewhat angle.
", "rawMarkdown": "gunesevitan ,\n\nTake a look at this video, it shows the basic principle of the data collection and the backprojection, reconstruction of the \"volumen\" you see:\nhttps://www.youtube.com/watch?v=rKh_XIpsuc4\n\nAs you see, the gantry rotates constantly around the patient and makes projections from every angle... these are then reconstructed (with filtered backprojection or other methods).\n\nHere a short vid about the multislice CT (no sound though, but the slices are enough):\nhttps://youtu.be/Eny8Sev1E88?t=48\n\nBy lung scans the gantry (the ring around the patient) is perpendicular to the axis Z, only by skull examination is it tilted to somewhat angle.", "votes": null }, { "id": "931433", "postDate": "07/16/2020 07:31:38", "content": "The volume of the lungs is increasing slowly in the first couple seconds then it shrinks very rapidly. It looks like top and bottom are very small, but the bottom has a more pointed edge. Do you guys think the CT scans are taken from top to bottom because of this reason?
\n\n![\"lungs\"](\"https://upload.wikimedia.org/wikipedia/commons/thumb/a/a1/Lungs_diagram_detailed.svg/1200px-Lungs_diagram_detailed.svg.png\")
I really don't understand why people are downvoting questions.
", "rawMarkdown": "The volume of the lungs is increasing slowly in the first couple seconds then it shrinks very rapidly. It looks like top and bottom are very small, but the bottom has a more pointed edge. Do you guys think the CT scans are taken from top to bottom because of this reason?\n\n\n\nI really don't understand why people are downvoting questions.", "votes": null }, { "id": "931493", "postDate": "07/16/2020 08:26:44", "content": "\n\nthe volume of the lung is constant during the scan, it does not change (except the patient cannot hold breath, see above).
\n\nThe entire scan is done within one breathold maneuvre, i found one video here, just look at the speed, how the patient is moving, and that was it.\n It takes really only a few seconds to capture the whole lung volume.
\n\nWhat you really mean is, when you scroll through the cross-sectional images, the lung area is getting larger from the top towards the bottom and it has a curved shape \"at the very bottom\", near the diaphragm, yes that is true. However this does give any info about the scan direction itself.
\n\nI would recommend you to watch this movie:\nhttps://www.youtube.com/watch?v=koHKAJWGNhU\nthere it is clearly visible how the anatomy looks like.
\n\nps: i made an upvote on it, so it became compensated =) =)
", "rawMarkdown": "gunesevitan ,\n\nthe volume of the lung is constant during the scan, it does not change (except the patient cannot hold breath, see above).\n\nThe entire scan is done within one breathold maneuvre, i found one [video here](https://youtu.be/u4bvisvSAu8), just look at the speed, how the patient is moving, and that was it.\n It takes really only a few seconds to capture the whole lung volume.\n\nWhat you really mean is, when you scroll through the cross-sectional images, the lung area is getting larger from the top towards the bottom and it has a curved shape \"at the very bottom\", near the diaphragm, yes that is true. However this does give any info about the scan direction itself.\n\nI would recommend you to watch this movie:\nhttps://www.youtube.com/watch?v=koHKAJWGNhU\nthere it is clearly visible how the anatomy looks like.\n\nps: i made an upvote on it, so it became compensated =) =)", "votes": null }, { "id": "933660", "postDate": "07/17/2020 21:49:50", "content": "Hello, CT captures dense tissue structure like bones with good contrast, and presents low resolution of soft tissues. During a CT scan, while the table moves the patient into the scanner, detectors and the X-ray tube rotate around him. Each rotation yields several images of thin slices of the body. Single inhale/exhale time can part of slicing process. Typically, making prediction on a single breath will be efficient since no changes will exhibit the capture process.
", "rawMarkdown": "Hello, CT captures dense tissue structure like bones with good contrast, and presents low resolution of soft tissues. During a CT scan, while the table moves the patient into the scanner, detectors and the X-ray tube rotate around him. Each rotation yields several images of thin slices of the body. Single inhale/exhale time can part of slicing process. Typically, making prediction on a single breath will be efficient since no changes will exhibit the capture process.", "votes": null }, { "id": "936042", "postDate": "07/19/2020 23:14:40", "content": "I agree with. @richardepstein. Like what @sandorkonya said, the CT scan is done in one respiratory phase (most probably inspire and hold). I believe the image you shared is showing a top to bottom scan, because as the lung tissue getting smaller, you can see abdominal organs showing up in the image.
", "rawMarkdown": "I agree with. @richardepstein. Like what @sandorkonya said, the CT scan is done in one respiratory phase (most probably inspire and hold). I believe the image you shared is showing a top to bottom scan, because as the lung tissue getting smaller, you can see abdominal organs showing up in the image.", "votes": null }, { "id": "940151", "postDate": "07/22/2020 17:41:53", "content": "@sandorkonya Thank you for all the insights! From my understanding of your explanation(please correct me if I'm wrong): the patient hold his/her breath while the CT scanner moves from top to bottom. \nCan you also explain why towards the end of the scan, there's two black circle area on top?
\n\nThanks in advance.
", "rawMarkdown": "sandorkonya Thank you for all the insights! From my understanding of your explanation(please correct me if I'm wrong): the patient hold his/her breath while the CT scanner moves from top to bottom. \nCan you also explain why towards the end of the scan, there's two black circle area on top?\n\nThanks in advance.", "votes": null }, { "id": "940181", "postDate": "07/22/2020 18:09:06", "content": "You might be referring to air in the bowel:
\n\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F22185%2F7c36bf5a41858d206afe57107ae04675%2FCapture.PNG?generation=1595441283957966&alt=media\")
Some of the air in this image is in the stomach and some in the colon.
\n\nIf you are referring to other air (this example is not from the Cine above), please paste a photo.
\n\nRich
", "rawMarkdown": "You might be referring to air in the bowel:\n\n\n\nSome of the air in this image is in the stomach and some in the colon.\n\nIf you are referring to other air (this example is not from the Cine above), please paste a photo.\n\nRich", "votes": null }, { "id": "940185", "postDate": "07/22/2020 18:11:28", "content": "Hi @unkownhihi let me try to answer your question (I am medical graduate, but of course @sandorkonya can also add). I am not sure what you are referring as \"two black circle area\", but let me give you a general principle. Its good to have some idea of the anatomy, the body is divided in to multiple compartments, chest and abdominal cavities being among them. In the chest cavity we have the lungs, the heart and some major blood vessels. In the abdominal cavity we have several solid organs (liver, spleen, kidneys etc), hollow organs and some empty space. In CT scan, bones and muscles can be seen as being white and hollow organs can be seen as black areas circles by white circle. So what you see towards the end is when the scan reaches the abdominal cavity, you are seeing air in the abdomen or some part of a hollow organ (which I think is not so relevant for this competition). I hope this answers your question. Let me know if you have any question. I am also looking to pair with someone, let me know if you are open to work with someone.
", "rawMarkdown": "Hi @unkownhihi let me try to answer your question (I am medical graduate, but of course @sandorkonya can also add). I am not sure what you are referring as \"two black circle area\", but let me give you a general principle. Its good to have some idea of the anatomy, the body is divided in to multiple compartments, chest and abdominal cavities being among them. In the chest cavity we have the lungs, the heart and some major blood vessels. In the abdominal cavity we have several solid organs (liver, spleen, kidneys etc), hollow organs and some empty space. In CT scan, bones and muscles can be seen as being white and hollow organs can be seen as black areas circles by white circle. So what you see towards the end is when the scan reaches the abdominal cavity, you are seeing air in the abdomen or some part of a hollow organ (which I think is not so relevant for this competition). I hope this answers your question. Let me know if you have any question. I am also looking to pair with someone, let me know if you are open to work with someone.", "votes": null }, { "id": "940204", "postDate": "07/22/2020 18:26:34", "content": "@drelias15 Thanks a lot for replying. I think I have a better understanding now.
\n\nI'd love to team up with someone that's an expert in this current field! *(Though, TBH, I don't have the best(not even close) DL/ML skills around so if you want to compete for the money, I would advise you to find a competition GM or Master. I'm sure plenty would agree) *
\n\nBut anyway(if you want to just have some fun in this competition), if you are interested, I can work around 3-4 hours per day(GMT-8 time zone), have some google cloud credits left.
", "rawMarkdown": "drelias15 Thanks a lot for replying. I think I have a better understanding now. \n\nI'd love to team up with someone that's an expert in this current field! *(Though, TBH, I don't have the best(not even close) DL/ML skills around so if you want to compete for the money, I would advise you to find a competition GM or Master. I'm sure plenty would agree) *\n\nBut anyway(if you want to just have some fun in this competition), if you are interested, I can work around 3-4 hours per day(GMT-8 time zone), have some google cloud credits left.", "votes": null }, { "id": "940212", "postDate": "07/22/2020 18:34:02", "content": "@unkownhihi I am totally not looking for the money. I am here to craft my DL/ML skils and I am just starting. Teaming up with you will work for me. You can create the team and DM me the request. Thank you. We can discuss on it.
", "rawMarkdown": "unkownhihi I am totally not looking for the money. I am here to craft my DL/ML skils and I am just starting. Teaming up with you will work for me. You can create the team and DM me the request. Thank you. We can discuss on it.", "votes": null }, { "id": "940218", "postDate": "07/22/2020 18:39:50", "content": "Sounds great! I'll send you a request!
", "rawMarkdown": "Sounds great! I'll send you a request!", "votes": null }, { "id": "940247", "postDate": "07/22/2020 19:10:49", "content": "Thanks I have joined the team and I have also sent you a message. My linkedin is https://www.linkedin.com/in/eliasadhanom/, you can DM me there as well.
", "rawMarkdown": "Thanks I have joined the team and I have also sent you a message. My linkedin is https://www.linkedin.com/in/eliasadhanom/, you can DM me there as well.", "votes": null }, { "id": "940278", "postDate": "07/22/2020 19:44:18", "content": "@unkownhihi and @drelias15 nice tobsee you found each other =)\nAnd yes, the gas filled cavity is the bowel, and has (probably) not much useful information content in this case.
", "rawMarkdown": "unkownhihi and @drelias15 nice tobsee you found each other =)\nAnd yes, the gas filled cavity is the bowel, and has (probably) not much useful information content in this case.", "votes": null }, { "id": "940281", "postDate": "07/22/2020 19:48:23", "content": "@drelias15 the gif is showing the image series in a top to bottom direction, however this does not give you a clue about how the scan itself was performed!
", "rawMarkdown": "drelias15 the gif is showing the image series in a top to bottom direction, however this does not give you a clue about how the scan itself was performed!", "votes": null }, { "id": "983673", "postDate": "08/24/2020 13:49:59", "content": "I'm sorry these questions might be dumb. But if my understanding about what you discussed is right, does this mean all the dcm documents of one patient are some continus scans from a single breath-hold maneuver? Also I noticed that some of the patients have much more dcm documents than others, does this mean they can hold the breathe much longer than others?
", "rawMarkdown": "I'm sorry these questions might be dumb. But if my understanding about what you discussed is right, does this mean all the dcm documents of one patient are some continus scans from a single breath-hold maneuver? Also I noticed that some of the patients have much more dcm documents than others, does this mean they can hold the breathe much longer than others?", "votes": null }, { "id": "983698", "postDate": "08/24/2020 14:20:15", "content": "@xiejialun That isn't correct. File count is dependent to spacing between slices (distance between slices in mm).
", "rawMarkdown": "xiejialun That isn't correct. File count is dependent to spacing between slices (distance between slices in mm).", "votes": null }, { "id": "983706", "postDate": "08/24/2020 14:26:20", "content": "There is variability in how we scan patients with CT. We could do continous 1.25 mm scans. We could do continously 5 mm scans. We could do 1.25 mm slices every 10 mm. That accounts for the variation.
", "rawMarkdown": "There is variability in how we scan patients with CT. We could do continous 1.25 mm scans. We could do continously 5 mm scans. We could do 1.25 mm slices every 10 mm. That accounts for the variation.", "votes": null }, { "id": "983758", "postDate": "08/24/2020 15:15:53", "content": "Awesome discussion here! Gives much needed understanding of the CT Scans. Thanks!
", "rawMarkdown": "Awesome discussion here! Gives much needed understanding of the CT Scans. Thanks!", "votes": null }, { "id": "984172", "postDate": "08/24/2020 22:55:04", "content": "@gunesevitan @richardepstein Thank you so much for the answer!! So we can only sure that all the dcm documents from one patient are from a single breath-hold maneuver. But since the scan distance might be different for each patient, so the amount of dcm documents are also different.
", "rawMarkdown": "gunesevitan @richardepstein Thank you so much for the answer!! So we can only sure that all the dcm documents from one patient are from a single breath-hold maneuver. But since the scan distance might be different for each patient, so the amount of dcm documents are also different.", "votes": null }, { "id": "984187", "postDate": "08/24/2020 23:20:26", "content": "We only have scans from a single day per patient.
\nI believe there is at least one study where they scanned the patient twice. I'm sure someone can provide the patient ID. This could be because the patient moved/breathed during the scan. Or they might have scanned once in inspiration and once in expiration. Sometimes we scan a patient lying on their back (most common) and lying on their stomach.
\nI know at least one patient was scanned on their stomach (prone).
\nYou can use the Dicom Metadata for the patient position to try and reconstruct these special cases. Obviously easier to do with the Train Data where you can look at the images than with real Test Data where you cannot see the images.
\nThere are also cases that we are missing an image or two from the normal sequence. I don't know if this is \"accidental\" as the images were anonymized and moved around or if the organizers want us to be able handle real-world data which is not always so clean.
", "rawMarkdown": "We only have scans from a single day per patient.\n\nI believe there is at least one study where they scanned the patient twice. I'm sure someone can provide the patient ID. This could be because the patient moved/breathed during the scan. Or they might have scanned once in inspiration and once in expiration. Sometimes we scan a patient lying on their back (most common) and lying on their stomach. \n\nI know at least one patient was scanned on their stomach (prone).\n\nYou can use the Dicom Metadata for the patient position to try and reconstruct these special cases. Obviously easier to do with the Train Data where you can look at the images than with real Test Data where you cannot see the images.\n\nThere are also cases that we are missing an image or two from the normal sequence. I don't know if this is \"accidental\" as the images were anonymized and moved around or if the organizers want us to be able handle real-world data which is not always so clean.", "votes": null }, { "id": "984204", "postDate": "08/25/2020 00:03:47", "content": "I think ID00078637202199415319443 is an example of a double scan (there are one or a couple of others too).
", "rawMarkdown": "I think ID00078637202199415319443 is an example of a double scan (there are one or a couple of others too).", "votes": null }, { "id": "984220", "postDate": "08/25/2020 00:39:05", "content": "@richardepstein @christoffer Thank you so much for the information, it helps a lot!!
", "rawMarkdown": "richardepstein @christoffer Thank you so much for the information, it helps a lot!!", "votes": null }, { "id": "994641", "postDate": "09/01/2020 18:58:57", "content": "Hi! The CT scans are not supposed to reflect the patient's breathing activity. Rather, they are \"scanning\" through the patient's chest, from the neck until the stomach region.
\n\nOne way to visualize this is to think of the old magician trick of sawing a woman in half: imagine (if the trick was real!) that you could take a picture of the upper body section sawed by the blade. That would be one so-called \"slice\" of the CT scan. And then imagine the magician would slice a few millimeters upward, closer to the chest. That would be another CT \"slice\". And so on...
\n\nAfter collecting all the pictures, you can basically make a flipbook video. This is exactly what you are seeing in the animation. Now it's time to ask the magician to join all the slices back again!
", "rawMarkdown": "Hi! The CT scans are not supposed to reflect the patient's breathing activity. Rather, they are \"scanning\" through the patient's chest, from the neck until the stomach region.\n\nOne way to visualize this is to think of the [old magician trick of sawing a woman in half](https://en.wikipedia.org/wiki/Sawing_a_woman_in_half): imagine (if the trick was real!) that you could take a picture of the upper body section sawed by the blade. That would be one so-called \"slice\" of the CT scan. And then imagine the magician would slice a few millimeters upward, closer to the chest. That would be another CT \"slice\". And so on... \n\nAfter collecting all the pictures, you can basically make a [flipbook video](https://www.youtube.com/watch?v=Un-BdBSOGKY). This is exactly what you are seeing in the animation. Now it's time to ask the magician to join all the slices back again!", "votes": null }, { "id": "995097", "postDate": "09/02/2020 06:58:31", "content": "Thanks for insightful answers @gunesevitan @sandorkonya @richardepstein.
\nIf CT were captured during a single breath-hold maneuver then it can help to benchmark lung volume/ condition for that particular day. I was wondering how this can help to predict the FVC multiple weeks after this day? There are patients for whom FVC increases then decreases and then again increases in subsequent weeks. So finally FVC may improve (or deteriorate )depending on the medication, external factors, mild exercise, etc. etc.
@prasunmishra ,
\nyoure right, it is a \"snapshot\" for that particular day.
\n\nI was wondering how this can help to predict the FVC multiple weeks after this day?
\n
This is the question of the competition =) Maybe features can be found that mirror the fibroblastic activity in the lung that can predict long time changes.
", "rawMarkdown": "prasunmishra ,\nyoure right, it is a \"snapshot\" for that particular day. \n> I was wondering how this can help to predict the FVC multiple weeks after this day?\n\nThis is the question of the competition =) Maybe features can be found that mirror the fibroblastic activity in the lung that can predict long time changes.", "votes": null }, { "id": "1030572", "postDate": "09/28/2020 18:26:49", "content": "Learned a lot about CT scans.
", "rawMarkdown": "Learned a lot about CT scans.", "votes": null } ], "comments": [ { "id": 983758, "author_name": "pratikkumar", "author_url": "", "post_date": "08/24/2020 15:15:53", "content": "Awesome discussion here! Gives much needed understanding of the CT Scans. Thanks!
", "votes": null, "replies": [] }, { "id": 995097, "author_name": "prasunmishra", "author_url": "", "post_date": "09/02/2020 06:58:31", "content": "Thanks for insightful answers @gunesevitan @sandorkonya @richardepstein.
\nIf CT were captured during a single breath-hold maneuver then it can help to benchmark lung volume/ condition for that particular day. I was wondering how this can help to predict the FVC multiple weeks after this day? There are patients for whom FVC increases then decreases and then again increases in subsequent weeks. So finally FVC may improve (or deteriorate )depending on the medication, external factors, mild exercise, etc. etc.
@prasunmishra ,
\nyoure right, it is a \"snapshot\" for that particular day.
\n\nI was wondering how this can help to predict the FVC multiple weeks after this day?
\n
This is the question of the competition =) Maybe features can be found that mirror the fibroblastic activity in the lung that can predict long time changes.
", "votes": null, "replies": [] } ] }, { "id": 1030572, "author_name": "knight1101", "author_url": "", "post_date": "09/28/2020 18:26:49", "content": "Learned a lot about CT scans.
", "votes": null, "replies": [] }, { "id": 930089, "author_name": "sandorkonya", "author_url": "", "post_date": "07/15/2020 07:30:43", "content": "\n\n> Is it correct that one CT scan is a single inhale/exhale process for one patient?
\n\nthe CT's are acquired within a single breath-hold maneuver!
\n\nThe patient inhales deeply and the scan goes from bottom to top (usually) or from top to bottom (less good).\nPatients with impaired lung-function cannot hold their breath long enough, even though that the acquisition is nowadays with multi slice CT very rapid, only few seconds (however it was not always the case).\nThis is why the bottom-up approach is considered to be better because if you breathe, the lower part of the chest moves more than the upper part.
\n\nIf the patient cannot hold the breath anymore, there will be motion artifacts on the CT (due to the nature of acquisition). In the reconstruction of the image discontinued or doubled contours appear.
\n\nExample for this here \n the 9th figure, (https://www.researchgate.net/publication/278670308_Clinical_and_Forensic_Radiology_Are_Not_the_Same/figures)
\n\nHowever the scan above is one of the nicely acquired ones, without any motion artifact.
\n\n> If so, is it a good idea make predictions based on a single breathe?
\n\nExposing the patient to further ionizing radiation to obtain another CT scan in expiration imposes elevated hazard. The additional value is however known, in special cases also scans in expiration are acquired, but not regularly, not by every patient.
", "votes": null, "replies": [ { "id": 930104, "author_name": "gunesevitan", "author_url": "", "post_date": "07/15/2020 07:45:41", "content": "Thanks for the quick answer @sandorkonya. This explains a lot. If I understood correctly, most of the information lies within the slices when the patients are fully inhaled, but that sudden burst of exhale is also giving us decent information about lung function.
", "votes": null, "replies": [] }, { "id": 930172, "author_name": "sandorkonya", "author_url": "", "post_date": "07/15/2020 08:40:48", "content": "\n\nif such motion artifacts are present, it means, that the patient could not hold the breath for the duration of the scan, and indeed, that could signalise a functional change (or status).
\n\nTo meaningfully be able to correlate this, the scan time information about the acquisition itself is needed!
\n\nBy a single detector CT it is pretty easy to calculate the scan time if we know:\nTable Speed, Table Feed per Rotation, Spiral Pitch Factor, and the acquisition length, but with multi detector CT it gets somewhat trickier...
\n\nThere is however \"Performed Procedure Step Start Time Attribute\" for each slice (not present in the current dataset) that could give us exactly the duration of the scan.
", "votes": null, "replies": [] }, { "id": 930337, "author_name": "richardepstein", "author_url": "", "post_date": "07/15/2020 11:22:53", "content": "The change in lung volume are not related to inhale/exhale in most of these cases. As mentioned, these are slices from top to bottom (or bottom to top) of the chest (from the neck to the belly). The lungs are smaller at the top of the chest than they are at the bottom of the chest.
\n\nIt's like slicing a loaf of bread and looking at each slice. Stack them back together and you have a 3D model of the lung.
", "votes": null, "replies": [] }, { "id": 930701, "author_name": "gunesevitan", "author_url": "", "post_date": "07/15/2020 16:50:57", "content": "@sandorkonya @richardepstein Thanks for the answers. It may sound dumb but do those CT scans are taken from a fixed angle?
", "votes": null, "replies": [] }, { "id": 930732, "author_name": "sandorkonya", "author_url": "", "post_date": "07/15/2020 17:13:57", "content": "\n\nTake a look at this video, it shows the basic principle of the data collection and the backprojection, reconstruction of the \"volumen\" you see:\nhttps://www.youtube.com/watch?v=rKh_XIpsuc4
\n\nAs you see, the gantry rotates constantly around the patient and makes projections from every angle... these are then reconstructed (with filtered backprojection or other methods).
\n\nHere a short vid about the multislice CT (no sound though, but the slices are enough):\nhttps://youtu.be/Eny8Sev1E88?t=48
\n\nBy lung scans the gantry (the ring around the patient) is perpendicular to the axis Z, only by skull examination is it tilted to somewhat angle.
", "votes": null, "replies": [] }, { "id": 931433, "author_name": "gunesevitan", "author_url": "", "post_date": "07/16/2020 07:31:38", "content": "The volume of the lungs is increasing slowly in the first couple seconds then it shrinks very rapidly. It looks like top and bottom are very small, but the bottom has a more pointed edge. Do you guys think the CT scans are taken from top to bottom because of this reason?
\n\n
I really don't understand why people are downvoting questions.
", "votes": null, "replies": [] }, { "id": 931493, "author_name": "sandorkonya", "author_url": "", "post_date": "07/16/2020 08:26:44", "content": "\n\nthe volume of the lung is constant during the scan, it does not change (except the patient cannot hold breath, see above).
\n\nThe entire scan is done within one breathold maneuvre, i found one video here, just look at the speed, how the patient is moving, and that was it.\n It takes really only a few seconds to capture the whole lung volume.
\n\nWhat you really mean is, when you scroll through the cross-sectional images, the lung area is getting larger from the top towards the bottom and it has a curved shape \"at the very bottom\", near the diaphragm, yes that is true. However this does give any info about the scan direction itself.
\n\nI would recommend you to watch this movie:\nhttps://www.youtube.com/watch?v=koHKAJWGNhU\nthere it is clearly visible how the anatomy looks like.
\n\nps: i made an upvote on it, so it became compensated =) =)
", "votes": null, "replies": [] }, { "id": 940151, "author_name": "unkownhihi", "author_url": "", "post_date": "07/22/2020 17:41:53", "content": "@sandorkonya Thank you for all the insights! From my understanding of your explanation(please correct me if I'm wrong): the patient hold his/her breath while the CT scanner moves from top to bottom. \nCan you also explain why towards the end of the scan, there's two black circle area on top?
\n\nThanks in advance.
", "votes": null, "replies": [ { "id": 940181, "author_name": "richardepstein", "author_url": "", "post_date": "07/22/2020 18:09:06", "content": "You might be referring to air in the bowel:
\n\n
Some of the air in this image is in the stomach and some in the colon.
\n\nIf you are referring to other air (this example is not from the Cine above), please paste a photo.
\n\nRich
", "votes": null, "replies": [] } ] }, { "id": 940185, "author_name": "drelias15", "author_url": "", "post_date": "07/22/2020 18:11:28", "content": "Hi @unkownhihi let me try to answer your question (I am medical graduate, but of course @sandorkonya can also add). I am not sure what you are referring as \"two black circle area\", but let me give you a general principle. Its good to have some idea of the anatomy, the body is divided in to multiple compartments, chest and abdominal cavities being among them. In the chest cavity we have the lungs, the heart and some major blood vessels. In the abdominal cavity we have several solid organs (liver, spleen, kidneys etc), hollow organs and some empty space. In CT scan, bones and muscles can be seen as being white and hollow organs can be seen as black areas circles by white circle. So what you see towards the end is when the scan reaches the abdominal cavity, you are seeing air in the abdomen or some part of a hollow organ (which I think is not so relevant for this competition). I hope this answers your question. Let me know if you have any question. I am also looking to pair with someone, let me know if you are open to work with someone.
", "votes": null, "replies": [] }, { "id": 940204, "author_name": "unkownhihi", "author_url": "", "post_date": "07/22/2020 18:26:34", "content": "@drelias15 Thanks a lot for replying. I think I have a better understanding now.
\n\nI'd love to team up with someone that's an expert in this current field! *(Though, TBH, I don't have the best(not even close) DL/ML skills around so if you want to compete for the money, I would advise you to find a competition GM or Master. I'm sure plenty would agree) *
\n\nBut anyway(if you want to just have some fun in this competition), if you are interested, I can work around 3-4 hours per day(GMT-8 time zone), have some google cloud credits left.
", "votes": null, "replies": [] }, { "id": 940212, "author_name": "drelias15", "author_url": "", "post_date": "07/22/2020 18:34:02", "content": "@unkownhihi I am totally not looking for the money. I am here to craft my DL/ML skils and I am just starting. Teaming up with you will work for me. You can create the team and DM me the request. Thank you. We can discuss on it.
", "votes": null, "replies": [] }, { "id": 940218, "author_name": "unkownhihi", "author_url": "", "post_date": "07/22/2020 18:39:50", "content": "Sounds great! I'll send you a request!
", "votes": null, "replies": [] }, { "id": 940247, "author_name": "drelias15", "author_url": "", "post_date": "07/22/2020 19:10:49", "content": "Thanks I have joined the team and I have also sent you a message. My linkedin is https://www.linkedin.com/in/eliasadhanom/, you can DM me there as well.
", "votes": null, "replies": [] }, { "id": 940278, "author_name": "sandorkonya", "author_url": "", "post_date": "07/22/2020 19:44:18", "content": "@unkownhihi and @drelias15 nice tobsee you found each other =)\nAnd yes, the gas filled cavity is the bowel, and has (probably) not much useful information content in this case.
", "votes": null, "replies": [] }, { "id": 983673, "author_name": "xiejialun", "author_url": "", "post_date": "08/24/2020 13:49:59", "content": "I'm sorry these questions might be dumb. But if my understanding about what you discussed is right, does this mean all the dcm documents of one patient are some continus scans from a single breath-hold maneuver? Also I noticed that some of the patients have much more dcm documents than others, does this mean they can hold the breathe much longer than others?
", "votes": null, "replies": [] }, { "id": 983698, "author_name": "gunesevitan", "author_url": "", "post_date": "08/24/2020 14:20:15", "content": "@xiejialun That isn't correct. File count is dependent to spacing between slices (distance between slices in mm).
", "votes": null, "replies": [ { "id": 983706, "author_name": "richardepstein", "author_url": "", "post_date": "08/24/2020 14:26:20", "content": "There is variability in how we scan patients with CT. We could do continous 1.25 mm scans. We could do continously 5 mm scans. We could do 1.25 mm slices every 10 mm. That accounts for the variation.
", "votes": null, "replies": [] }, { "id": 984172, "author_name": "xiejialun", "author_url": "", "post_date": "08/24/2020 22:55:04", "content": "@gunesevitan @richardepstein Thank you so much for the answer!! So we can only sure that all the dcm documents from one patient are from a single breath-hold maneuver. But since the scan distance might be different for each patient, so the amount of dcm documents are also different.
", "votes": null, "replies": [ { "id": 984187, "author_name": "richardepstein", "author_url": "", "post_date": "08/24/2020 23:20:26", "content": "We only have scans from a single day per patient.
\nI believe there is at least one study where they scanned the patient twice. I'm sure someone can provide the patient ID. This could be because the patient moved/breathed during the scan. Or they might have scanned once in inspiration and once in expiration. Sometimes we scan a patient lying on their back (most common) and lying on their stomach.
\nI know at least one patient was scanned on their stomach (prone).
\nYou can use the Dicom Metadata for the patient position to try and reconstruct these special cases. Obviously easier to do with the Train Data where you can look at the images than with real Test Data where you cannot see the images.
\nThere are also cases that we are missing an image or two from the normal sequence. I don't know if this is \"accidental\" as the images were anonymized and moved around or if the organizers want us to be able handle real-world data which is not always so clean.
", "votes": null, "replies": [] }, { "id": 984204, "author_name": "christoffer", "author_url": "", "post_date": "08/25/2020 00:03:47", "content": "I think ID00078637202199415319443 is an example of a double scan (there are one or a couple of others too).
", "votes": null, "replies": [] }, { "id": 984220, "author_name": "xiejialun", "author_url": "", "post_date": "08/25/2020 00:39:05", "content": "@richardepstein @christoffer Thank you so much for the information, it helps a lot!!
", "votes": null, "replies": [] } ] } ] } ] }, { "id": 933660, "author_name": "haithemhermessi", "author_url": "", "post_date": "07/17/2020 21:49:50", "content": "Hello, CT captures dense tissue structure like bones with good contrast, and presents low resolution of soft tissues. During a CT scan, while the table moves the patient into the scanner, detectors and the X-ray tube rotate around him. Each rotation yields several images of thin slices of the body. Single inhale/exhale time can part of slicing process. Typically, making prediction on a single breath will be efficient since no changes will exhibit the capture process.
", "votes": null, "replies": [] }, { "id": 936042, "author_name": "drelias15", "author_url": "", "post_date": "07/19/2020 23:14:40", "content": "I agree with. @richardepstein. Like what @sandorkonya said, the CT scan is done in one respiratory phase (most probably inspire and hold). I believe the image you shared is showing a top to bottom scan, because as the lung tissue getting smaller, you can see abdominal organs showing up in the image.
", "votes": null, "replies": [ { "id": 940281, "author_name": "sandorkonya", "author_url": "", "post_date": "07/22/2020 19:48:23", "content": "@drelias15 the gif is showing the image series in a top to bottom direction, however this does not give you a clue about how the scan itself was performed!
", "votes": null, "replies": [] } ] }, { "id": 994641, "author_name": "douglaskgaraujo", "author_url": "", "post_date": "09/01/2020 18:58:57", "content": "Hi! The CT scans are not supposed to reflect the patient's breathing activity. Rather, they are \"scanning\" through the patient's chest, from the neck until the stomach region.
\n\nOne way to visualize this is to think of the old magician trick of sawing a woman in half: imagine (if the trick was real!) that you could take a picture of the upper body section sawed by the blade. That would be one so-called \"slice\" of the CT scan. And then imagine the magician would slice a few millimeters upward, closer to the chest. That would be another CT \"slice\". And so on...
\n\nAfter collecting all the pictures, you can basically make a flipbook video. This is exactly what you are seeing in the animation. Now it's time to ask the magician to join all the slices back again!
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "930060": "As we know baseline CT scans are acquired at Week 0, but I'm still struggling to understand what exactly they are capturing.\n\nIn the image below, the lungs are expanding slowly as the black area grows larger. Patient is probably inhaling during that time. After that, the black area is disappearing rapidly and I assumed this is when the patient was exhaling. \n\n\n\nMy questions are:\n* Is it correct that one CT scan is a single inhale/exhale process for one patient?\n* If so, is it a good idea make predictions based on a single breathe?\n* Why organizers decided to share only one CT scan per patient? Do they change at all on different weeks?", "930089": "gunesevitan,\n\n> Is it correct that one CT scan is a single inhale/exhale process for one patient?\n\n\nthe CT's are acquired within a single breath-hold maneuver! \n\nThe patient inhales deeply and the scan goes from bottom to top (usually) or from top to bottom (less good).\nPatients with impaired lung-function cannot hold their breath long enough, even though that the acquisition is nowadays with multi slice CT very rapid, only few seconds (however it was not always the case).\nThis is why the bottom-up approach is considered to be better because if you breathe, the lower part of the chest moves more than the upper part. \n\nIf the patient cannot hold the breath anymore, there will be motion artifacts on the CT (due to the nature of acquisition). In the reconstruction of the image discontinued or doubled contours appear. \n\nExample for this [here](https://www.researchgate.net/profile/Michael_Thali/publication/278670308/figure/fig5/AS:390171573932044@1470035560708/Motion-artifacts-in-radiology-a-Two-diaphragms-on-each-side-due-to-respiration-during_W640.jpg) \n the 9th figure, (https://www.researchgate.net/publication/278670308_Clinical_and_Forensic_Radiology_Are_Not_the_Same/figures)\n\nHowever the scan above is one of the nicely acquired ones, without any motion artifact.\n\n> If so, is it a good idea make predictions based on a single breathe?\n\nExposing the patient to further ionizing radiation to obtain another CT scan in expiration imposes elevated hazard. The additional value is however known, in special cases also scans in expiration are acquired, but not regularly, not by every patient.", "930104": "Thanks for the quick answer @sandorkonya. This explains a lot. If I understood correctly, most of the information lies within the slices when the patients are fully inhaled, but that sudden burst of exhale is also giving us decent information about lung function.", "930172": "gunesevitan ,\n\nif such motion artifacts are present, it means, that the patient could not hold the breath for the duration of the scan, and indeed, that could signalise a functional change (or status).\n\nTo meaningfully be able to correlate this, the scan time information about the acquisition itself is needed! \n\nBy a single detector CT it is pretty easy to calculate the scan time if we know:\nTable Speed, Table Feed per Rotation, Spiral Pitch Factor, and the acquisition length, but with multi detector CT it gets somewhat trickier...\n\nThere is however \"Performed Procedure Step Start Time Attribute\" for each slice (not present in the current dataset) that could give us exactly the duration of the scan.", "930337": "The change in lung volume are not related to inhale/exhale in most of these cases. As mentioned, these are slices from top to bottom (or bottom to top) of the chest (from the neck to the belly). The lungs are smaller at the top of the chest than they are at the bottom of the chest.\n\nIt's like slicing a loaf of bread and looking at each slice. Stack them back together and you have a 3D model of the lung.", "930701": "sandorkonya @richardepstein Thanks for the answers. It may sound dumb but do those CT scans are taken from a fixed angle?", "930732": "gunesevitan ,\n\nTake a look at this video, it shows the basic principle of the data collection and the backprojection, reconstruction of the \"volumen\" you see:\nhttps://www.youtube.com/watch?v=rKh_XIpsuc4\n\nAs you see, the gantry rotates constantly around the patient and makes projections from every angle... these are then reconstructed (with filtered backprojection or other methods).\n\nHere a short vid about the multislice CT (no sound though, but the slices are enough):\nhttps://youtu.be/Eny8Sev1E88?t=48\n\nBy lung scans the gantry (the ring around the patient) is perpendicular to the axis Z, only by skull examination is it tilted to somewhat angle.", "931433": "The volume of the lungs is increasing slowly in the first couple seconds then it shrinks very rapidly. It looks like top and bottom are very small, but the bottom has a more pointed edge. Do you guys think the CT scans are taken from top to bottom because of this reason?\n\n\n\nI really don't understand why people are downvoting questions.", "931493": "gunesevitan ,\n\nthe volume of the lung is constant during the scan, it does not change (except the patient cannot hold breath, see above).\n\nThe entire scan is done within one breathold maneuvre, i found one [video here](https://youtu.be/u4bvisvSAu8), just look at the speed, how the patient is moving, and that was it.\n It takes really only a few seconds to capture the whole lung volume.\n\nWhat you really mean is, when you scroll through the cross-sectional images, the lung area is getting larger from the top towards the bottom and it has a curved shape \"at the very bottom\", near the diaphragm, yes that is true. However this does give any info about the scan direction itself.\n\nI would recommend you to watch this movie:\nhttps://www.youtube.com/watch?v=koHKAJWGNhU\nthere it is clearly visible how the anatomy looks like.\n\nps: i made an upvote on it, so it became compensated =) =)", "933660": "Hello, CT captures dense tissue structure like bones with good contrast, and presents low resolution of soft tissues. During a CT scan, while the table moves the patient into the scanner, detectors and the X-ray tube rotate around him. Each rotation yields several images of thin slices of the body. Single inhale/exhale time can part of slicing process. Typically, making prediction on a single breath will be efficient since no changes will exhibit the capture process.", "936042": "I agree with. @richardepstein. Like what @sandorkonya said, the CT scan is done in one respiratory phase (most probably inspire and hold). I believe the image you shared is showing a top to bottom scan, because as the lung tissue getting smaller, you can see abdominal organs showing up in the image.", "940151": "sandorkonya Thank you for all the insights! From my understanding of your explanation(please correct me if I'm wrong): the patient hold his/her breath while the CT scanner moves from top to bottom. \nCan you also explain why towards the end of the scan, there's two black circle area on top?\n\nThanks in advance.", "940181": "You might be referring to air in the bowel:\n\n\n\nSome of the air in this image is in the stomach and some in the colon.\n\nIf you are referring to other air (this example is not from the Cine above), please paste a photo.\n\nRich", "940185": "Hi @unkownhihi let me try to answer your question (I am medical graduate, but of course @sandorkonya can also add). I am not sure what you are referring as \"two black circle area\", but let me give you a general principle. Its good to have some idea of the anatomy, the body is divided in to multiple compartments, chest and abdominal cavities being among them. In the chest cavity we have the lungs, the heart and some major blood vessels. In the abdominal cavity we have several solid organs (liver, spleen, kidneys etc), hollow organs and some empty space. In CT scan, bones and muscles can be seen as being white and hollow organs can be seen as black areas circles by white circle. So what you see towards the end is when the scan reaches the abdominal cavity, you are seeing air in the abdomen or some part of a hollow organ (which I think is not so relevant for this competition). I hope this answers your question. Let me know if you have any question. I am also looking to pair with someone, let me know if you are open to work with someone.", "940204": "drelias15 Thanks a lot for replying. I think I have a better understanding now. \n\nI'd love to team up with someone that's an expert in this current field! *(Though, TBH, I don't have the best(not even close) DL/ML skills around so if you want to compete for the money, I would advise you to find a competition GM or Master. I'm sure plenty would agree) *\n\nBut anyway(if you want to just have some fun in this competition), if you are interested, I can work around 3-4 hours per day(GMT-8 time zone), have some google cloud credits left.", "940212": "unkownhihi I am totally not looking for the money. I am here to craft my DL/ML skils and I am just starting. Teaming up with you will work for me. You can create the team and DM me the request. Thank you. We can discuss on it.", "940218": "Sounds great! I'll send you a request!", "940247": "Thanks I have joined the team and I have also sent you a message. My linkedin is https://www.linkedin.com/in/eliasadhanom/, you can DM me there as well.", "940278": "unkownhihi and @drelias15 nice tobsee you found each other =)\nAnd yes, the gas filled cavity is the bowel, and has (probably) not much useful information content in this case.", "940281": "drelias15 the gif is showing the image series in a top to bottom direction, however this does not give you a clue about how the scan itself was performed!", "983673": "I'm sorry these questions might be dumb. But if my understanding about what you discussed is right, does this mean all the dcm documents of one patient are some continus scans from a single breath-hold maneuver? Also I noticed that some of the patients have much more dcm documents than others, does this mean they can hold the breathe much longer than others?", "983698": "xiejialun That isn't correct. File count is dependent to spacing between slices (distance between slices in mm).", "983706": "There is variability in how we scan patients with CT. We could do continous 1.25 mm scans. We could do continously 5 mm scans. We could do 1.25 mm slices every 10 mm. That accounts for the variation.", "983758": "Awesome discussion here! Gives much needed understanding of the CT Scans. Thanks!", "984172": "gunesevitan @richardepstein Thank you so much for the answer!! So we can only sure that all the dcm documents from one patient are from a single breath-hold maneuver. But since the scan distance might be different for each patient, so the amount of dcm documents are also different.", "984187": "We only have scans from a single day per patient.\n\nI believe there is at least one study where they scanned the patient twice. I'm sure someone can provide the patient ID. This could be because the patient moved/breathed during the scan. Or they might have scanned once in inspiration and once in expiration. Sometimes we scan a patient lying on their back (most common) and lying on their stomach. \n\nI know at least one patient was scanned on their stomach (prone).\n\nYou can use the Dicom Metadata for the patient position to try and reconstruct these special cases. Obviously easier to do with the Train Data where you can look at the images than with real Test Data where you cannot see the images.\n\nThere are also cases that we are missing an image or two from the normal sequence. I don't know if this is \"accidental\" as the images were anonymized and moved around or if the organizers want us to be able handle real-world data which is not always so clean.", "984204": "I think ID00078637202199415319443 is an example of a double scan (there are one or a couple of others too).", "984220": "richardepstein @christoffer Thank you so much for the information, it helps a lot!!", "994641": "Hi! The CT scans are not supposed to reflect the patient's breathing activity. Rather, they are \"scanning\" through the patient's chest, from the neck until the stomach region.\n\nOne way to visualize this is to think of the [old magician trick of sawing a woman in half](https://en.wikipedia.org/wiki/Sawing_a_woman_in_half): imagine (if the trick was real!) that you could take a picture of the upper body section sawed by the blade. That would be one so-called \"slice\" of the CT scan. And then imagine the magician would slice a few millimeters upward, closer to the chest. That would be another CT \"slice\". And so on... \n\nAfter collecting all the pictures, you can basically make a [flipbook video](https://www.youtube.com/watch?v=Un-BdBSOGKY). This is exactly what you are seeing in the animation. Now it's time to ask the magician to join all the slices back again!", "995097": "Thanks for insightful answers @gunesevitan @sandorkonya @richardepstein. \nIf CT were captured during a single breath-hold maneuver then it can help to benchmark lung volume/ condition for that particular day. I was wondering how this can help to predict the FVC multiple weeks after this day? There are patients for whom FVC increases then decreases and then again increases in subsequent weeks. So finally FVC may improve (or deteriorate )depending on the medication, external factors, mild exercise, etc. etc.", "999675": "prasunmishra ,\nyoure right, it is a \"snapshot\" for that particular day. \n> I was wondering how this can help to predict the FVC multiple weeks after this day?\n\nThis is the question of the competition =) Maybe features can be found that mirror the fibroblastic activity in the lung that can predict long time changes.", "1030572": "Learned a lot about CT scans." }, "source": "meta" }![here](\"https://www.researchgate.net/profile/Michael_Thali/publication/278670308/figure/fig5/AS:390171573932044@1470035560708/Motion-artifacts-in-radiology-a-Two-diaphragms-on-each-side-due-to-respiration-during_W640.jpg\"){kind=link}