I was recently submitted yet another broken DICOM file, it contains
a JP2 compressed image (header info), instead of a J2K code stream:


I anonymized it manually. For some reason the CD it was on,
contained a RAW image and a compressed one right next to eah other.
For some reason the Media Storage SOP Instance UID would contain the
Root UID of Algotech (Implementation Class confirms that too), but all
UID in the dataset contained the Root UID of TOSHIBA (image was
acquired on TOSHIBA machine).