3 ms·

AI platform to advance stereotactic radiosurgery (SRS) for brain tumor patients

1 points by coolwulf 5y ago

coolwulf 5y ago
4 years ago, I posted here to introduce some work I did using AI for breast tumor detection and classification: https://www.reddit.com/r/MachineLearning/comments/8rdpwy/pi_made_a_gpu_cluster_and_free_website_to_help/ https://www.reddit.com/r/MachineLearning/comments/8rdpwy/pi_... That post gain some traction on Reddit and I hope you would like the one I am gonna introduce here again. In the recent years, I have been shifting my focus from cancer detection to the actual treatment. One particular problem we really want to solve is to have more brain cancer patients to be accessible to stereotactic radiosurgery (SRS) which has a lot better treatment outcome and much better quality of life (QoL) for the patient than whole brain radiotherapy (WBRT) which is more common for patients with multiple brainiest (say more than 5 or 10). The reason behind that is it takes a really lot of efforts for physician or neurosurgeon for the SRS procedures for multiple brain mets patients since each lesion needs to be correctly contoured / labeled / tracked and also a lot of efforts for medical physicists for treatment planning for these kinds of multiple lesion cases. With patients of more than 10 or 20 brain mets, there are just not enough resources out there to help them usually. Therefore starting from 2019, we collaborated with UTSW and Stanford trying to tackle this problem. We have since developed models for auto contouring / labeling brain mets lesions, together with Organ-at-Risks (OAR), also we developed models for fast false positive reduction based on SVM-radiomics. In addition, we developed models for quickly splitting multiple mets into different treatment sessions based on optimized radiation dose maps. This work will be presented tomorrow at 2022 AAPM spring clinical meeting by our collaborator at Stanford https://w4.aapm.org/meetings/2022SCM/programInfo/programAbs.php?sid=9476&aid=60551 https://w4.aapm.org/meetings/2022SCM/programInfo/programAbs.... And we are accelerating to trying to have this technology and platform available to the whole SRS community very soon. We really hope this platform could bring more patients to SRS procedures instead of the brutal whole brain RT. And we have seem patients keep coming back every a few months for SRS to extend the quality of their life (comparing with WBRT which is a once-in-a-life treatment which has not other treatment methods after it when the patients have reoccurrence occurred after 3 months). Hope you like this work. Hao
coolwulf 5y ago
In the meantime, we will be offering a live demo to our revolutionary SRS platform if you are interested. You could book a 20-min meeting with us using the link: https://lnkd.in/eSNg9vMK https://lnkd.in/eSNg9vMK at your convenience.

1 points by coolwulf 5y ago

4 years ago, I posted here to introduce some work I did using AI for breast tumor detection and classification: https://www.reddit.com/r/MachineLearning/comments/8rdpwy/pi_made_a_gpu_cluster_and_free_website_to_help/ https://www.reddit.com/r/MachineLearning/comments/8rdpwy/pi_... That post gain some traction on Reddit and I hope you would like the one I am gonna introduce here again. In the recent years, I have been shifting my focus from cancer detection to the actual treatment. One particular problem we really want to solve is to have more brain cancer patients to be accessible to stereotactic radiosurgery (SRS) which has a lot better treatment outcome and much better quality of life (QoL) for the patient than whole brain radiotherapy (WBRT) which is more common for patients with multiple brainiest (say more than 5 or 10). The reason behind that is it takes a really lot of efforts for physician or neurosurgeon for the SRS procedures for multiple brain mets patients since each lesion needs to be correctly contoured / labeled / tracked and also a lot of efforts for medical physicists for treatment planning for these kinds of multiple lesion cases. With patients of more than 10 or 20 brain mets, there are just not enough resources out there to help them usually. Therefore starting from 2019, we collaborated with UTSW and Stanford trying to tackle this problem. We have since developed models for auto contouring / labeling brain mets lesions, together with Organ-at-Risks (OAR), also we developed models for fast false positive reduction based on SVM-radiomics. In addition, we developed models for quickly splitting multiple mets into different treatment sessions based on optimized radiation dose maps. This work will be presented tomorrow at 2022 AAPM spring clinical meeting by our collaborator at Stanford https://w4.aapm.org/meetings/2022SCM/programInfo/programAbs.php?sid=9476&aid=60551 https://w4.aapm.org/meetings/2022SCM/programInfo/programAbs.... And we are accelerating to trying to have this technology and platform available to the whole SRS community very soon. We really hope this platform could bring more patients to SRS procedures instead of the brutal whole brain RT. And we have seem patients keep coming back every a few months for SRS to extend the quality of their life (comparing with WBRT which is a once-in-a-life treatment which has not other treatment methods after it when the patients have reoccurrence occurred after 3 months). Hope you like this work. Hao

coolwulf 5y ago

In the meantime, we will be offering a live demo to our revolutionary SRS platform if you are interested. You could book a 20-min meeting with us using the link: https://lnkd.in/eSNg9vMK https://lnkd.in/eSNg9vMK at your convenience.