These slices are not combined because the DICOM headers report they are from different studies (0020,000d) and from different series (0020,000e). I would check the providence of these images, and see if you can get earlier copies of the images before these tags were manipulated.
Based on 0002,0013, I suspect these images were mangled by a Matlab script (potentially attempting to anonymize data). The DICOM format is extremely complex, and this makes DICOM images fragile to manipulation. Personally, I am a fan of gdcmanon.
(0002,0013) SH [MATLAB IPT 7.1] # 14, 1 ImplementationVersionName