Recommendations for PM&R case study

I would like to ask for your recommendations. Although I am a data scientist, I have never worked professionally in the medical domain or the field of medical imaging. However, someone close to me — a specialist in Physical Medicine and Rehabilitation (PM&R) and neurorehabilitation — asked me to assist in putting together a case study on an individual patient, which can be presented to colleagues and students. My role mainly involves using CT data to add supporting images, figures, graphs, and statistics to the case study.

What I have: Extensive imaging data on a patient who underwent complex cervical spine surgery due to osteochondroma and was treated with postoperative cerebritis after. The cranial imaging is all from CT scans, with the patient being scanned continuously during the acute state and regularly thereafter.

What I have tried so far: I have delved into some tools in the domain, such as 3D Slicer, and have tried to grasp main terminology and techniques like registration and segmentation. I have also explored tools like FreeSurfer, FastSurfer, and Synthseg, successfully performing segmentation on the scans with Synthseg.

What I am looking for: I want to add visual figures and statistical analysis to the case study related to PM&R work, especially focusing on the lesions in his brain and the brain-related damage due to the cerebritis. I need ideas on how to extract useful statistical information and produce good visuals from these cranial CT images to demonstrate the case and the patient’s status, as well as potential rehabilitation efforts.

I would also be happy to learn about any research and state-of-the-art techniques on how to utilize medical imaging and deep learning/segmentation within the PM&R field, especially for planning and coordinating the rehabilitation of TBI (traumatic brain injury) patients.

Generally it sounds like you are on a good path of learning and exploring the tools. Slicer may provide useful tools for visualizing and presenting the case data and can be used to make renderings and animations. It should also be straightforward to register multiple timepoints so they can be compared side-by-side.