First, pleasure to be part of this discussion group. I am a newbie on this topic and struggling how to address on of the research someone has requested me to have a look. Objective is to analyze the likely progression of a traumatic brain injury based on some clinical attributes and CT scans with a expected outcome of Good or Bad progress training a predefined or custom CNN . So far, my understanding having a technical background in IT might be,
- Get all CT scans associated to each patient and linked them with clinical attributes (age…) in a CSV file
- Normalize and standarize data.
Several questions that come to my mind.
- How can I get features from CT scans with no masks using PyRadiomics. I dont really exactly know where to get them.
- Whats the preferred CNN to use? Any previous use case to address it?
- Dicom vs Nifty. Whats the best format? Should I use ITK for this purpose?
Any helpful hint how to start or any valuable guidance will be welcome