Differences in MRI imaging methods - MRIの撮影方法の違いについて
\nThe principles of MRI require knowledge of physics, and it would make more sense if you could understand that much, but even few doctors in the field understand that much.
\nMRIの原理は物理学的な知識が必要であり、そこまで理解できるとより深く納得できると思われるが、現場の医師でもそこまで理解している人は少ない。
So, it would be good if you could roughly understand what the doctor's perspective feels he wants to see mainly in which imaging method. How to summarize lightly.
\nなので、ざっくりとどの撮影方法で何を主にみたいと医師の目線で感じているかを理解してもらえれば良いと思う。いかに軽くまとめる。
However, in the field, MRIs were not originally used to diagnose this issue, and one can think that it would be better to work on the competition without being bound by this concept.
\nただ、現場では本来MRIで今回の課題を診断することはなく、この考え方にとらわれずに競技に取り組む方が良いという考えもできる。
T1
\nFat is depicted in white and water in black.
\nThe shape of the brain can be clearly seen, and morphological abnormalities are easy to detect. (Atrophy, tumors, etc.)
\n脂肪が白く、水が黒く描かれる。
\n脳の形がよくわかり、形態的な異常が捉えやすい。（萎縮や腫瘍など）
T1Gd
\nT1Gd is T1 imaging with contrast medium and is the method that best reflects the location, size, and shape of the mass.
\nT1Gdは造影剤を用いたT1撮影で、腫瘤の位置、大きさ、形が最もよく反映される方法である。
T2
\nWater is painted white.
\nLesions appear white. Suitable for lesion evaluation.
\n水が白く描かれる。
\n病変が白く映る。病変の評価に適している。
FLAIR
\nIn T2, the spinal fluid (water) is white and the lesion is also white, so you have to look for the white in the white, which is difficult to understand.
\nFLAIR can be roughly thought of as T2, in which the water is also black, making it easier to find the lesion.
\nT2は髄液（水）も白、病変も白となり、白の中で白を探すことになり、わかりにくい
\nFLAIRは水も黒くすることで、より病変を探しやすくなったT2とざっくり捉えてもらえれば良い。
Since I am a medical student, I would appreciate any corrections from current doctors.
\n私は医学生なので、現役の医師から訂正箇所があれば訂正してほしいです。
Differences in MRI imaging methods - MRIの撮影方法の違いについて
\nThe principles of MRI require knowledge of physics, and it would make more sense if you could understand that much, but even few doctors in the field understand that much.
\nMRIの原理は物理学的な知識が必要であり、そこまで理解できるとより深く納得できると思われるが、現場の医師でもそこまで理解している人は少ない。
So, it would be good if you could roughly understand what the doctor's perspective feels he wants to see mainly in which imaging method. How to summarize lightly.
\nなので、ざっくりとどの撮影方法で何を主にみたいと医師の目線で感じているかを理解してもらえれば良いと思う。いかに軽くまとめる。
However, in the field, MRIs were not originally used to diagnose this issue, and one can think that it would be better to work on the competition without being bound by this concept.
\nただ、現場では本来MRIで今回の課題を診断することはなく、この考え方にとらわれずに競技に取り組む方が良いという考えもできる。
T1
\nFat is depicted in white and water in black.
\nThe shape of the brain can be clearly seen, and morphological abnormalities are easy to detect. (Atrophy, tumors, etc.)
\n脂肪が白く、水が黒く描かれる。
\n脳の形がよくわかり、形態的な異常が捉えやすい。（萎縮や腫瘍など）
T1Gd
\nT1Gd is T1 imaging with contrast medium and is the method that best reflects the location, size, and shape of the mass.
\nT1Gdは造影剤を用いたT1撮影で、腫瘤の位置、大きさ、形が最もよく反映される方法である。
T2
\nWater is painted white.
\nLesions appear white. Suitable for lesion evaluation.
\n水が白く描かれる。
\n病変が白く映る。病変の評価に適している。
FLAIR
\nIn T2, the spinal fluid (water) is white and the lesion is also white, so you have to look for the white in the white, which is difficult to understand.
\nFLAIR can be roughly thought of as T2, in which the water is also black, making it easier to find the lesion.
\nT2は髄液（水）も白、病変も白となり、白の中で白を探すことになり、わかりにくい
\nFLAIRは水も黒くすることで、より病変を探しやすくなったT2とざっくり捉えてもらえれば良い。
Since I am a medical student, I would appreciate any corrections from current doctors.
\n私は医学生なので、現役の医師から訂正箇所があれば訂正してほしいです。
Nice work upwote.
\nCould you look my notebook. https://www.kaggle.com/saruhangngr/price-prediction-with-gb-backward-elemination.
How does a doctor decide if a brain contains a tumor or not? What types of geometrical structures or artifacts does she look for? (small round spots, big gray regions…? idk). Is color an important factor or is the shape of a lesion more important?
\nAnother question: in practice, do doctors take T2 into account, or do they ignore T2 and use FLAIR instead?
\nThese considerations will allow us to tailor our models to the problem.
", "rawMarkdown": "How does a doctor decide if a brain contains a tumor or not? What types of geometrical structures or artifacts does she look for? (small round spots, big gray regions...? idk). Is color an important factor or is the shape of a lesion more important?\n\nAnother question: in practice, do doctors take T2 into account, or do they ignore T2 and use FLAIR instead?\n\nThese considerations will allow us to tailor our models to the problem.", "votes": null }, { "id": "1391262", "postDate": "07/17/2021 12:46:30", "content": "Both can be said to be important.
\nThe coronal or axial section should be viewed while comparing the color and shape of the left and right sides.
\nThe following are some of the characteristic imaging findings seen in glioblastoma.
Thanks for the amazing explanation, i was looking for something like this. One question, are all these scans taken from the same angle? (Like Top view For all FLAIR, T1w, T1wCE, T2w) as i can see, in examples, the shapes look quite different. Or are different Scans are taken from different angles for the same patient.
", "rawMarkdown": "Thanks for the amazing explanation, i was looking for something like this. One question, are all these scans taken from the same angle? (Like Top view For all FLAIR, T1w, T1wCE, T2w) as i can see, in examples, the shapes look quite different. Or are different Scans are taken from different angles for the same patient.", "votes": null }, { "id": "1441395", "postDate": "08/03/2021 20:16:52", "content": "I am a Radiolgy Resident of the first year so I might be wrong. I will put here some bullets from what I see from the competition data and what I have looked up into Radiomics feature of gliomas:
\nI guess that for having a great prediction we might have to use every sequence.
", "rawMarkdown": "I am a Radiolgy Resident of the first year so I might be wrong. I will put here some bullets from what I see from the competition data and what I have looked up into Radiomics feature of gliomas:\n\n- All the dataset exams contains tumors\n- Some tumors have MGMT mutation -- a important feature on the disease outcome and treatment\n- A neuroradiologist looks all the four sequences (T1 , T1CE, T2 and FLAIR) to guess whether a tumor has or not the MGMT mutation and look up for shape and texture of the lesion and additional features such as edema circulating the lesion.\n- A lot of public notebooks are using only FLAIR sequence to predict whether has the mutation or not and that is because FLAIR is a sequence of high sensitivity for alterations.\n\n\nI guess that for having a great prediction we might have to use every sequence.", "votes": null }, { "id": "1441413", "postDate": "08/03/2021 20:18:51", "content": "No they are not. Some of the scans or \"sequences\" are in coronal, sagittal or axial planes of view. You need to be aware when trying to transpose it for other plane (reconstruct). If the slice scanned has a thickness higher than 3 mm it might loose features when you reconstruct it for another plane of view.
", "rawMarkdown": "No they are not. Some of the scans or \"sequences\" are in coronal, sagittal or axial planes of view. You need to be aware when trying to transpose it for other plane (reconstruct). If the slice scanned has a thickness higher than 3 mm it might loose features when you reconstruct it for another plane of view.", "votes": null }, { "id": "1446069", "postDate": "08/04/2021 10:57:33", "content": "Thanks. This helps a lot. All the best for the competition.
", "rawMarkdown": "Thanks. This helps a lot. All the best for the competition.", "votes": null }, { "id": "1470649", "postDate": "08/13/2021 16:31:07", "content": "Hey, some interesting points!!
\nI am curious how can you say that all the images have tumours?
\nand can you tell me, suppose if someone doesn't have a tumour at all, can then his MGMT value be positive then?
@eduardofarina This is really useful information. Can you please elaborate on the appearance of MGMT promoter methylation? Is it something non-experts in the field can recognize by looking at the scans?
", "rawMarkdown": "eduardofarina This is really useful information. Can you please elaborate on the appearance of MGMT promoter methylation? Is it something non-experts in the field can recognize by looking at the scans?", "votes": null }, { "id": "1514560", "postDate": "09/16/2021 07:38:58", "content": "@novice03 I suppose methylation of the MGMT promoter is not obvious from the image alone, even to experts.
", "rawMarkdown": "novice03 I suppose methylation of the MGMT promoter is not obvious from the image alone, even to experts.", "votes": null } ], "comments": [ { "id": 1388033, "author_name": "mikeyclark", "author_url": "", "post_date": "07/14/2021 15:54:49", "content": "Nice work upwote.
\nCould you look my notebook. https://www.kaggle.com/saruhangngr/price-prediction-with-gb-backward-elemination.
How does a doctor decide if a brain contains a tumor or not? What types of geometrical structures or artifacts does she look for? (small round spots, big gray regions…? idk). Is color an important factor or is the shape of a lesion more important?
\nAnother question: in practice, do doctors take T2 into account, or do they ignore T2 and use FLAIR instead?
\nThese considerations will allow us to tailor our models to the problem.
", "votes": null, "replies": [ { "id": 1391262, "author_name": "dlbsabu", "author_url": "", "post_date": "07/17/2021 12:46:30", "content": "Both can be said to be important.
\nThe coronal or axial section should be viewed while comparing the color and shape of the left and right sides.
\nThe following are some of the characteristic imaging findings seen in glioblastoma.
I am a Radiolgy Resident of the first year so I might be wrong. I will put here some bullets from what I see from the competition data and what I have looked up into Radiomics feature of gliomas:
\nI guess that for having a great prediction we might have to use every sequence.
", "votes": null, "replies": [] }, { "id": 1470649, "author_name": "mrinath", "author_url": "", "post_date": "08/13/2021 16:31:07", "content": "Hey, some interesting points!!
\nI am curious how can you say that all the images have tumours?
\nand can you tell me, suppose if someone doesn't have a tumour at all, can then his MGMT value be positive then?
@eduardofarina This is really useful information. Can you please elaborate on the appearance of MGMT promoter methylation? Is it something non-experts in the field can recognize by looking at the scans?
", "votes": null, "replies": [] }, { "id": 1514560, "author_name": "dlbsabu", "author_url": "", "post_date": "09/16/2021 07:38:58", "content": "@novice03 I suppose methylation of the MGMT promoter is not obvious from the image alone, even to experts.
", "votes": null, "replies": [] } ] }, { "id": 1430185, "author_name": "abtexp", "author_url": "", "post_date": "08/03/2021 11:23:08", "content": "Thanks for the amazing explanation, i was looking for something like this. One question, are all these scans taken from the same angle? (Like Top view For all FLAIR, T1w, T1wCE, T2w) as i can see, in examples, the shapes look quite different. Or are different Scans are taken from different angles for the same patient.
", "votes": null, "replies": [ { "id": 1441413, "author_name": "eduardofarina", "author_url": "", "post_date": "08/03/2021 20:18:51", "content": "No they are not. Some of the scans or \"sequences\" are in coronal, sagittal or axial planes of view. You need to be aware when trying to transpose it for other plane (reconstruct). If the slice scanned has a thickness higher than 3 mm it might loose features when you reconstruct it for another plane of view.
", "votes": null, "replies": [] }, { "id": 1446069, "author_name": "abtexp", "author_url": "", "post_date": "08/04/2021 10:57:33", "content": "Thanks. This helps a lot. All the best for the competition.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "1387094": "Differences in MRI imaging methods - MRIの撮影方法の違いについて\n\nThe principles of MRI require knowledge of physics, and it would make more sense if you could understand that much, but even few doctors in the field understand that much.\nMRIの原理は物理学的な知識が必要であり、そこまで理解できるとより深く納得できると思われるが、現場の医師でもそこまで理解している人は少ない。\n\nSo, it would be good if you could roughly understand what the doctor's perspective feels he wants to see mainly in which imaging method. How to summarize lightly.\nなので、ざっくりとどの撮影方法で何を主にみたいと医師の目線で感じているかを理解してもらえれば良いと思う。いかに軽くまとめる。\n\nHowever, in the field, MRIs were not originally used to diagnose this issue, and one can think that it would be better to work on the competition without being bound by this concept.\nただ、現場では本来MRIで今回の課題を診断することはなく、この考え方にとらわれずに競技に取り組む方が良いという考えもできる。\n\nT1\nFat is depicted in white and water in black.\nThe shape of the brain can be clearly seen, and morphological abnormalities are easy to detect. (Atrophy, tumors, etc.)\n脂肪が白く、水が黒く描かれる。\n脳の形がよくわかり、形態的な異常が捉えやすい。（萎縮や腫瘍など）\n\nT1Gd\nT1Gd is T1 imaging with contrast medium and is the method that best reflects the location, size, and shape of the mass.\nT1Gdは造影剤を用いたT1撮影で、腫瘤の位置、大きさ、形が最もよく反映される方法である。\n\nT2\nWater is painted white.\nLesions appear white. Suitable for lesion evaluation.\n水が白く描かれる。\n病変が白く映る。病変の評価に適している。\n\nFLAIR\nIn T2, the spinal fluid (water) is white and the lesion is also white, so you have to look for the white in the white, which is difficult to understand.\nFLAIR can be roughly thought of as T2, in which the water is also black, making it easier to find the lesion.\nT2は髄液（水）も白、病変も白となり、白の中で白を探すことになり、わかりにくい\nFLAIRは水も黒くすることで、より病変を探しやすくなったT2とざっくり捉えてもらえれば良い。\n\nSince I am a medical student, I would appreciate any corrections from current doctors.\n私は医学生なので、現役の医師から訂正箇所があれば訂正してほしいです。", "1388033": "Nice work upwote.\nCould you look my notebook. https://www.kaggle.com/saruhangngr/price-prediction-with-gb-backward-elemination.", "1390457": "How does a doctor decide if a brain contains a tumor or not? What types of geometrical structures or artifacts does she look for? (small round spots, big gray regions...? idk). Is color an important factor or is the shape of a lesion more important?\n\nAnother question: in practice, do doctors take T2 into account, or do they ignore T2 and use FLAIR instead?\n\nThese considerations will allow us to tailor our models to the problem.", "1391262": "Both can be said to be important.\nThe coronal or axial section should be viewed while comparing the color and shape of the left and right sides.\nThe following are some of the characteristic imaging findings seen in glioblastoma.\n- it is drawn in black on T1\n- White contrast around the tumor can be seen on T1Gd. It is characterized by an irregular shape.\n- White blotchy areas on T2-weighted or FLAIR", "1430185": "Thanks for the amazing explanation, i was looking for something like this. One question, are all these scans taken from the same angle? (Like Top view For all FLAIR, T1w, T1wCE, T2w) as i can see, in examples, the shapes look quite different. Or are different Scans are taken from different angles for the same patient.", "1441395": "I am a Radiolgy Resident of the first year so I might be wrong. I will put here some bullets from what I see from the competition data and what I have looked up into Radiomics feature of gliomas:\n\n- All the dataset exams contains tumors\n- Some tumors have MGMT mutation -- a important feature on the disease outcome and treatment\n- A neuroradiologist looks all the four sequences (T1 , T1CE, T2 and FLAIR) to guess whether a tumor has or not the MGMT mutation and look up for shape and texture of the lesion and additional features such as edema circulating the lesion.\n- A lot of public notebooks are using only FLAIR sequence to predict whether has the mutation or not and that is because FLAIR is a sequence of high sensitivity for alterations.\n\n\nI guess that for having a great prediction we might have to use every sequence.", "1441413": "No they are not. Some of the scans or \"sequences\" are in coronal, sagittal or axial planes of view. You need to be aware when trying to transpose it for other plane (reconstruct). If the slice scanned has a thickness higher than 3 mm it might loose features when you reconstruct it for another plane of view.", "1446069": "Thanks. This helps a lot. All the best for the competition.", "1470649": "Hey, some interesting points!!\nI am curious how can you say that all the images have tumours?\nand can you tell me, suppose if someone doesn't have a tumour at all, can then his MGMT value be positive then?", "1470656": "eduardofarina This is really useful information. Can you please elaborate on the appearance of MGMT promoter methylation? Is it something non-experts in the field can recognize by looking at the scans?", "1514560": "novice03 I suppose methylation of the MGMT promoter is not obvious from the image alone, even to experts." }, "source": "meta" }