readcell vs xlsread: substitute "1x1 missing" with "NaN"

I used to import my mixed (char/num) content from xls-files as cell array with:

[~, ~, raw1] = xlsread(myfile, mysheet, myrange, 'basic')

Because xlsread is "not recommended", I'm trying to move to readcell. So according to "compatibility consideration" in documentation, I'm now using:

raw2 = readcell(myfile, 'Sheet', mysheet, 'Range', myrange);

It works, but where I had "NaN" in raw1, now I see "1x1 missing" in raw2. I'd like to replace those "1x1 missing" with "NaN" to get the very same output as with xlsread, so I tried:

raw2(cellfun(@(x) any(ismissing(x)), raw2)) = {NaN};

The problem is, not only "1x1 missing", but also some (but not all) valid text values are replaced. What am I doing wrong?

[Jan]

It would be much easier to find the problem, if we have the input data or know, which data are unexpectedly replaces. "Some bit not all" does not reveal the details.

[JarryG]

Wow, you actually helped me! I checked what data were replaced, and it was not so random as I thought. Only strings with empty characters like "test string" were replaced, while others like "test-string" were not. It seems "any" statement in my cell function is incorrect, as empty character is evaluated as "missing", and it is enough if single one is found. So I changed my function to:

raw2(cellfun(@(x) all(ismissing(x)), raw2)) = {NaN};

Now it works as expected: all "1x1 missing" are replaced by "NaN", and nothing else...