[btomba_77]

That would be an Amit/Chip thing

This is literally the first I have heard of it.

Im still reading MRI knee analog-style 😉

[kaldridgewv2211]

Its a specific GE portable we have that can send a secondary capture along with this the actual image for chest portables. Ive not paid much attention but I saw a couple images. We have it setup so we can look at the DICOM tags for AI findings. Itll dump that into the PACS DB and then you can use a worklist criteria to look for that tag. Overall seems like a good idea as the machine might flag something as being priority for review which I suppose you otherwise wouldnt know until a rad looks at it.

[nzalewski]

We have it at our facility.  Had someone do an internal QA study of the pneumothorax AI.  Of the cases where either the AI or radiologist called a pneumothorax, there was 20% agreement.  Of the remaining 80% discrepancies, the AI overcalled 54% of the time and undercalled 25% of the time.

[donna_912]

So… slightly better than a PA doing plain films then?

[kaldridgewv2211]

The images I saw looked more like measurements for tube placements. Heres the blurb from GE. It looks like they have a picture of our PACS worklist on their website.

[link=https://www.gehealthcare.com/products/radiography/mobile-xray-systems/critical-care-suite-on-optima-xr240amx]https://www.gehealthcare….ite-on-optima-xr240amx[/link]

[joshua.glaze_811]

Is the sensitivity adjustable?  Sometime ago I was told that a related system was basically adjustable to 100 sensitivity and acted as a text notification to the neurologists. Not sure if it was true or commonplace.
 

Quote from calltaker

We have it at our facility.  Had someone do an internal QA study of the pneumothorax AI.  Of the cases where either the AI or radiologist called a pneumothorax, there was 20% agreement.  Of the remaining 80% discrepancies, the AI overcalled 54% of the time and undercalled 25% of the time.

[amyelizabethbarrett28_711]

Quote from calltaker

We have it at our facility.  Had someone do an internal QA study of the pneumothorax AI.  Of the cases where either the AI or radiologist called a pneumothorax, there was 20% agreement.  Of the remaining 80% discrepancies, the AI overcalled 54% of the time and undercalled 25% of the time.

Sounds about like mammo CAD