You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Auto-detect coverage bounding boxes for brain MRI images
Good day everyone,
My name is Nuren Zhaksylyk and I am currently pursuing MSc in Computer Vision at MBZUAI, Abu Dhabi, UAE. Currently, I am the Graduate Student Researcher at BioMedIA Lab here at MBZUAI under Dr. Mohammad Yaqub. Currently my ongoing research is application of model soups in medical image tasks and now we are pushing our findings to MICCAI 2024.
In my understanding, you already have some pipeline that is giving you bounding boxes according to pre-defined protocols and you want to build something upon it to asses how good your detector is actually doing.
The easiest way is to do brain parts segmentation using brain atlas and by comparing intersection of each regions in protocol with the bounding box. You can use Intersection over Union as a metric for evaluation. I am able to do that in a short period of time as I have all the needed skills. I am familiar with MONAI pipeline as it will fit to your project as it is only 90 hours project. I am proficient with writing custom functions and how to incorporate them under MONAI library. I have been working with MONAI when our lab was participating in SegAorta challenge in MICCAI 2023 and I know how to do 3D segmentation using voxels or slice by slice. I am also proficient working with DICOM format and how to do resampling before feeding the data to the model.
I am ready to spend as much time as needed and I am flexible with working schedule. I can train models on our machines once we set up connection between our labs.
I am very proficient with PyTorch and building pipelines for training and testing the models. Also, I am mainly working in medical domain and have deep understanding of many concepts and have publication in Nature Scientific Data. We have collected ARCADE dataset for benchmarking deep learning models focusing in tackling CAD.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
Auto-detect coverage bounding boxes for brain MRI images
Good day everyone,
My name is Nuren Zhaksylyk and I am currently pursuing MSc in Computer Vision at MBZUAI, Abu Dhabi, UAE. Currently, I am the Graduate Student Researcher at BioMedIA Lab here at MBZUAI under Dr. Mohammad Yaqub. Currently my ongoing research is application of model soups in medical image tasks and now we are pushing our findings to MICCAI 2024.
In my understanding, you already have some pipeline that is giving you bounding boxes according to pre-defined protocols and you want to build something upon it to asses how good your detector is actually doing.
The easiest way is to do brain parts segmentation using brain atlas and by comparing intersection of each regions in protocol with the bounding box. You can use Intersection over Union as a metric for evaluation. I am able to do that in a short period of time as I have all the needed skills. I am familiar with MONAI pipeline as it will fit to your project as it is only 90 hours project. I am proficient with writing custom functions and how to incorporate them under MONAI library. I have been working with MONAI when our lab was participating in SegAorta challenge in MICCAI 2023 and I know how to do 3D segmentation using voxels or slice by slice. I am also proficient working with DICOM format and how to do resampling before feeding the data to the model.
I am ready to spend as much time as needed and I am flexible with working schedule. I can train models on our machines once we set up connection between our labs.
I am very proficient with PyTorch and building pipelines for training and testing the models. Also, I am mainly working in medical domain and have deep understanding of many concepts and have publication in Nature Scientific Data. We have collected ARCADE dataset for benchmarking deep learning models focusing in tackling CAD.
Contributor:
Nuren Zhaksylyk
Potential Mentors
Kind regards,
Nuren
Beta Was this translation helpful? Give feedback.
All reactions