I think you can start with a review - there had been many techniques try to improve the segmentation of brain tumors, like different sequence imagining technique using MRI, fMRI mapping, real time intra-operational functional mapping, real time fluoroscent dye (locally or pre-surgical hyper-selective using interventional technique) technique, real time fast frozen histology analysis etc. (These are all old methods, btw, I'm sure if you look, there'll be a lot new ones).

However, I think the highlight of the paper should be the real clinical impact on these segmentation technique: There has been many clinical studies as well (many of them were ignored by scientists, as they usually discourage previous findings), most of them suggest the conventional segmentation is not as helpful to clinical practice. The reason be, even the segmentation were provided via imagining technique, it is not histologically matching the actual tumor border. Take glioma for example, there has been debates on the extent of recession over many years, we all want to complete remove the tumor while preserve as much healthy brain tissue as we could, but this had been proven to be challenging even as the imagining technique advancing. In the old days, they advice us to remove the tumor via the color mismatch on the tissue, then they suggest we could remove the edema region surrounding the tumor, and some study even point out we should perform extensive excision up to a 2 mm radius outside the edema region. All they want is to remove as much more as possible to suppress recurrent and relapse.

Base on my experience, I don't think there is any good technique currently that can be applied as gold standard to determine the ideal border of removal for glioma, but the future trend should be to identify the actual histology border using the imagining technology we have. If anyone could do that, it should be enough to revolutionize the treatment guideline and strategies for glioma.