{ "id": 255540, "title": "Confusion about the genetic analysis aspect metioned in the Overview. ", "url": "/competitions/rsna-miccai-brain-tumor-radiogenomic-classification/discussion/255540", "author_name": "Ayush Thakur", "post_date": "2021-07-28T02:49:59.504000", "votes": 15, "comment_count": 4, "views": 0, "content": "

I am asking this question primarily to the competition hosts. I believe the hosts can help clear this confusion.

\n

To fellow Kagglers, if you have noticed there's a paragraph in the Overview section:

\n
\n

Currently, genetic analysis of cancer requires surgery to extract a tissue sample. Then it can take several weeks to determine the genetic characterization of the tumor. Depending upon the results and type of initial therapy chosen, a subsequent surgery may be necessary. If an accurate method to predict the genetics of the cancer through imaging (i.e., radiogenomics) alone could be developed, this would potentially minimize the number of surgeries and refine the type of therapy required.

\n
\n

When I hear the term genetic analysis, I think of DNA sequencing of the mutated cell (tumor) to understand the mutation type and come up with a cure accordingly.

\n

The statement,

\n
\n

If an accurate method to predict the genetics of the cancer through imaging

\n
\n

confuses me because I don't think we can get a molecular understanding (DNA) of the mutated cell via imaging. Maybe the above-quoted statement can be reframed in a different way.

\n

The confusion might be because of my limited understanding of genetic analysis (well my understanding comes from many Netflix documentaries I had watched).

", "messages": [ { "id": 1402217, "postDate": "2021-07-28T02:49:59.503Z", "content": "

I am asking this question primarily to the competition hosts. I believe the hosts can help clear this confusion.

\n

To fellow Kagglers, if you have noticed there's a paragraph in the Overview section:

\n
\n

Currently, genetic analysis of cancer requires surgery to extract a tissue sample. Then it can take several weeks to determine the genetic characterization of the tumor. Depending upon the results and type of initial therapy chosen, a subsequent surgery may be necessary. If an accurate method to predict the genetics of the cancer through imaging (i.e., radiogenomics) alone could be developed, this would potentially minimize the number of surgeries and refine the type of therapy required.

\n
\n

When I hear the term genetic analysis, I think of DNA sequencing of the mutated cell (tumor) to understand the mutation type and come up with a cure accordingly.

\n

The statement,

\n
\n

If an accurate method to predict the genetics of the cancer through imaging

\n
\n

confuses me because I don't think we can get a molecular understanding (DNA) of the mutated cell via imaging. Maybe the above-quoted statement can be reframed in a different way.

\n

The confusion might be because of my limited understanding of genetic analysis (well my understanding comes from many Netflix documentaries I had watched).

", "rawMarkdown": "I am asking this question primarily to the competition hosts. I believe the hosts can help clear this confusion. \n\nTo fellow Kagglers, if you have noticed there's a paragraph in the Overview section:\n\n> Currently, **genetic analysis of cancer requires surgery to extract a tissue sample**. Then it can take several weeks to determine the genetic characterization of the tumor. Depending upon the results and type of initial therapy chosen, a subsequent surgery may be necessary. **If an accurate method to predict the genetics of the cancer through imaging** (i.e., radiogenomics) alone could be developed, this would potentially minimize the number of surgeries and refine the type of therapy required.\n\nWhen I hear the term genetic analysis, I think of DNA sequencing of the mutated cell (tumor) to understand the mutation type and come up with a cure accordingly. \n\nThe statement, \n> If an accurate method to predict the **genetics** of the cancer through imaging\n\nconfuses me because I don't think we can get a molecular understanding (DNA) of the mutated cell via imaging. Maybe the above-quoted statement can be reframed in a different way. \n\nThe confusion might be because of my limited understanding of genetic analysis (well my understanding comes from many Netflix documentaries I had watched). ", "votes": 14 }, { "id": 1402413, "postDate": "2021-07-28T07:16:20.437Z", "content": "

Hi, not a competition host here :)

\n

Having related background, I could try to answer.

\n

The main point, we have here a specific task. It is not genetic analysis in general, I would say.

\n

For this specific task, host ask us to build a model which can reliably predict the MGMT status of the tumor given the MRI image.
\nMGMT status is only one of many (already defined) mutations, which is greatly correlated with a chemo-treatment outcome (specifically Temozolomide). So, given we can build a reliable and accurate enough model, it could help in decision-making and also facilitate the start of the treatment (sometimes even without preoperative biopsy).

\n

\"Predict the genetics of the cancer through imaging\" statement does not require the model to be as good as genetic analysis per se, nor to be genetic analysis, the model should be useful enough. As statistician George Box put it: \"All models are wrong, but some are useful\".

\n

I hope, it makes some sense :)

\n

All the best,
\nAlexey

", "rawMarkdown": "Hi, not a competition host here :)\n \nHaving related background, I could try to answer.\n\nThe main point, we have here a **specific** task. It is not genetic analysis in general, I would say.\n\nFor this specific task, host ask us to build a **model** which can reliably predict the MGMT status of the tumor given the **MRI image**.\nMGMT status is only one of many (already defined) mutations, which is greatly correlated with a chemo-treatment outcome (specifically Temozolomide). So, given we can build a reliable and accurate enough model, it could help in decision-making and also facilitate the start of the treatment (sometimes even without preoperative biopsy).\n\n\"Predict the genetics of the cancer through imaging\" statement does not require the model to be as good as genetic analysis per se, nor to be genetic analysis, the model should be useful enough. As statistician George Box put it: \"All models are wrong, but some are useful\".\n\nI hope, it makes some sense :)\n\nAll the best,\nAlexey\n\n", "votes": 9, "replies": [ { "id": 1402443, "postDate": "2021-07-28T07:47:13.273Z", "content": "

Thanks, @ademyanchuk. It is insightful. I was doing my own research to learn more about GBM and found this article useful: https://radiopaedia.org/articles/glioblastoma

", "rawMarkdown": "Thanks, @ademyanchuk. It is insightful. I was doing my own research to learn more about GBM and found this article useful: https://radiopaedia.org/articles/glioblastoma" } ] }, { "id": 1403151, "postDate": "2021-07-28T20:52:58Z", "rawMarkdown": "", "votes": 1, "isDeleted": true, "replies": [ { "id": 1403511, "postDate": "2021-07-29T07:22:46.280Z", "content": "

Thanks, @weka511 :)

\n

It's the same as machine learning is not just neural network, it's just the past couple of years it's working really well. Thanks for sharing the course too.

", "rawMarkdown": "Thanks, @weka511 :)\n\nIt's the same as machine learning is not just neural network, it's just the past couple of years it's working really well. Thanks for sharing the course too. " } ] } ], "comments": [ { "id": 1402413, "author_name": "A.Demyanchuk", "author_url": "", "post_date": "2021-07-28T07:16:20.437000", "content": "

Hi, not a competition host here :)

\n

Having related background, I could try to answer.

\n

The main point, we have here a specific task. It is not genetic analysis in general, I would say.

\n

For this specific task, host ask us to build a model which can reliably predict the MGMT status of the tumor given the MRI image.
\nMGMT status is only one of many (already defined) mutations, which is greatly correlated with a chemo-treatment outcome (specifically Temozolomide). So, given we can build a reliable and accurate enough model, it could help in decision-making and also facilitate the start of the treatment (sometimes even without preoperative biopsy).

\n

\"Predict the genetics of the cancer through imaging\" statement does not require the model to be as good as genetic analysis per se, nor to be genetic analysis, the model should be useful enough. As statistician George Box put it: \"All models are wrong, but some are useful\".

\n

I hope, it makes some sense :)

\n

All the best,
\nAlexey

", "votes": 9, "replies": [ { "id": 1402443, "author_name": "Ayush Thakur", "author_url": "", "post_date": "2021-07-28T07:47:13.273000", "content": "

Thanks, @ademyanchuk. It is insightful. I was doing my own research to learn more about GBM and found this article useful: https://radiopaedia.org/articles/glioblastoma

", "votes": 0, "replies": [] } ] }, { "id": 1403151, "author_name": "", "author_url": "", "post_date": "2021-07-28T20:52:58", "content": "", "votes": 1, "replies": [ { "id": 1403511, "author_name": "Ayush Thakur", "author_url": "", "post_date": "2021-07-29T07:22:46.280000", "content": "

Thanks, @weka511 :)

\n

It's the same as machine learning is not just neural network, it's just the past couple of years it's working really well. Thanks for sharing the course too.

", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "1402217": "I am asking this question primarily to the competition hosts. I believe the hosts can help clear this confusion. \n\nTo fellow Kagglers, if you have noticed there's a paragraph in the Overview section:\n\n> Currently, **genetic analysis of cancer requires surgery to extract a tissue sample**. Then it can take several weeks to determine the genetic characterization of the tumor. Depending upon the results and type of initial therapy chosen, a subsequent surgery may be necessary. **If an accurate method to predict the genetics of the cancer through imaging** (i.e., radiogenomics) alone could be developed, this would potentially minimize the number of surgeries and refine the type of therapy required.\n\nWhen I hear the term genetic analysis, I think of DNA sequencing of the mutated cell (tumor) to understand the mutation type and come up with a cure accordingly. \n\nThe statement, \n> If an accurate method to predict the **genetics** of the cancer through imaging\n\nconfuses me because I don't think we can get a molecular understanding (DNA) of the mutated cell via imaging. Maybe the above-quoted statement can be reframed in a different way. \n\nThe confusion might be because of my limited understanding of genetic analysis (well my understanding comes from many Netflix documentaries I had watched). ", "1402413": "Hi, not a competition host here :)\n \nHaving related background, I could try to answer.\n\nThe main point, we have here a **specific** task. It is not genetic analysis in general, I would say.\n\nFor this specific task, host ask us to build a **model** which can reliably predict the MGMT status of the tumor given the **MRI image**.\nMGMT status is only one of many (already defined) mutations, which is greatly correlated with a chemo-treatment outcome (specifically Temozolomide). So, given we can build a reliable and accurate enough model, it could help in decision-making and also facilitate the start of the treatment (sometimes even without preoperative biopsy).\n\n\"Predict the genetics of the cancer through imaging\" statement does not require the model to be as good as genetic analysis per se, nor to be genetic analysis, the model should be useful enough. As statistician George Box put it: \"All models are wrong, but some are useful\".\n\nI hope, it makes some sense :)\n\nAll the best,\nAlexey\n\n", "1403151": "" } }