I appreciate the structural uncertainty in your paper. When running your code, I noticed that in the result of DMT, some small disconnected points in the initial segmentation result are incorrectly connected. Could you please advise me on how to avoid this issue?
