I have adjusted the alignment settings as seen on the picture below and you can already see some improvement in the high detail reconstruction. What this means is:
The area is too dark and gets close to what the backdrop is
The area also has not the best features on it which are hard to detect
To answer beforehand to a question you would ask (why does normal detail work better then?) It may be due to this: The normal detail lowers the resolution of the images before reconstructing and the pixels defining the tiny dust detail could then come together creating something more likely to be recognised as a pattern.
Solution is to convert the RAW files better by increasing the shadows, decreasing highlights,sharpening, clarity or texture adjustments. Or simply adding some fake detail onto the surface of your scanned object. (coating spray, mud, flour) and doing two sets of images. One clean and one “dirty”.
I also recommend going for some other backdrop instead.