I have been using a very efficient column volumetric magnetic resonance protocol for the past 5 years. I’m looking for a group of people interested in a segmentation protocol that can extract and segment bone tissue from a volumetric MRI file. I can see a very important use for this technology, specifically in the infant population, in order to generate volumetric reconstructions of the spine without exposure to ionizing radiation.
Hi Ricardo, this sounds like a fascinating project. MRI is a wonderful modality that is underutilized for procedure planning (at least in my opinion). Can you elaborate on the technical aims of what you are trying to accomplish?
It sounds like you’ve already accumulated a pretty good dataset, do you also have masks or models for some of them? Would a reference set need to be generated and validated as a starting point?
What degree of automation is needed? Is the goal to have a model that can create accurate masks without supervision or would operators be able to refine the masks before they are used downstream?
Is the goal purely research, or do you have specific applications (procedure planning, surgical navigation, 3D printing) which should be taken into account as part of the protocol design?
Dear @Ricardoneuro,
Great idea. I know of a company doing this task as a commercial software. That might be interesting for your question.
My PhD supervisor (Dr. Peter Seevinck) started working on the idea of generating synthetic CT images from MR images 5 years ago and the deep learning-based softwate product recently has extended CE marks for spine as well.
Here’s one study evaluating usage of this sCT spine for surgical planning:
The company is MRIGuidance:
Best,
Saeed
This is an interesting task. Have you considered utilising MONAI Label with the DeepEdit App (MONAILabel/sample-apps/deepedit at main · Project-MONAI/MONAILabel · GitHub)?
Hi. I´ve been doing segmentation and 3D printing of medical exams for some time. I would surely like to help in anyway I can.
Best regards,
Lucas