4 ms·
Exactly! MR ("my" modality) has it right -- raw data and reconned images are very, very different and although most raw data never ends up in a dicom the mere f
by azalemeth 5y ago
Exactly! MR ("my" modality) has it right -- raw data and reconned images are very, very different and although most raw data never ends up in a dicom the mere fact that you genuinely could reconstruct dramatically different bits of info (e.g. magnitude vs phase images) means that the vast expanse of the dicom spec is wide enough to encompass all possible metadata requirements.
US machines do a lot of fun physics on proprietary FPGAs. For inexplicable reasons, every one I've ever worked with or done echo with saves the images as some variation on a theme of screenshots, shoehorned badly into a dicom wrapper, with the metadata burned at 640x480 px (or similar) on top. Even for clever derived modes like doppler -- even for annotations showing things like cardiac E/E' or E/A. They are laptops with a custom pcmcia / pcie card and a 100k-UNIT_OF_CURRENCY price tag, inevitably running a shitty OS with a shittier custom UI...
- lostlogin 5y agoMRI is my modality of choice too. I’m currently loving most of what Siemens is up to (with some notable exceptions). The hell of US knows no bounds. Most modalities calibrate a display and then display images (with varying degrees of post processing). US calibrates the screen, sometimes with each boot or even each probe change. Their black levels are abysmal. > saves the images as some variation on a theme of screenshots GE has a habit of making DICOMs from screen grabs. I’ve seen it on their PET, CT and MR systems. It causes irritating problems - like reference lines won’t work so you can’t cross reference.