Mmh. Using Excel, you need to take care to, when you open the file, always select all cells in the import dialog (which, I assume looks similar to that of OpenOffice Calc; if not, I'll have to check at my day job on friday) and choose the format "Text" explicitely (auto detect will corrupt some values for arbitrary leaders). For cell separators, only semicolon should be selected.
... At this point, I fear the only thing you can do is to start over with the original file and reapply your changes; if Excel corrupted values, it might have happened anywhere in the file.
Of course, diffing with a text-only tool (TortoiseDiff, Notepad++ Compare, DiffUtils, ...) is still an alternative, but you might end up with dozens of unwanted changes you'd have to revert manually. If too many things were changed, the tool might not even be able to correctly align the file versions and then your work would silently introduce further errors as you fix them. So, I cannot recommend this approach at this point.
As general advice for the future, I'd only use Excel or OpenOffice Calc or similar spreadsheet programs to view a csv file, and to never save with them. Changes can then safely be applied from a raw text editor or coding tool, such as Notepad++, Visual Studio Code, nano, vim, emacs, etc.
