Fatima explained this really well. The point about thresholding not capturing cortical surface detail properly is probably the key issue here, especially around the sulci and gyri.
A lot of Brain MRI segmentation and 3D reconstruction workflows run into this when trying to prepare models for printing. The compatibility issues with some extensions can also make the process more frustrating than most tutorials suggest.
Hopefully Kurt shares an update because it would be interesting to see how the final reconstruction turns out.