To confirm, you're saying that result of importing the file as UINT8 is double the array size as importing as UINT16? That's to be expected. uint8 will read in blocks of 8 bits from your file (range 0 to 255; e.g. one byte) sequentially, while uint16 will read in blocks of 16 bits (range 0 to 65536, two bytes). If your file is 96 bytes long, then you'll see 96 8-bit values, but only 48 16-bit values.
If you look at their sizes you'll likely see that the array bytes are the same. Each 16-bit value storage uses twice as many bytes.
a = uint8([1,2,3,4]);
b= uint16([1,2]);
whos;
Edit:
Important detail I left off: Since you have 10 bit data, assuming it's in block form, neither method is going to import your data correctly, because it does not know that directly. See the documentation for fread. You'll want to specify the bitn version. Use '*ubit10' instead of to place the 10-bit data into uint16 format for MATLAB to use.
data = fread(fid,'*ubit10');
