Evidence · Reader feedback

We put ViewFinder in front of twenty breast imagers and asked if they would use it

In June 2021, ViewFinder was installed on the workstations of a Hologic tomosynthesis reading course at King's College Hospital. The delegates were practising UK breast imagers. They used it on real cases and filled in a feedback form. This is what they said.

The question we asked

“How likely would you be to use it in your practice?”

Delegates answered on a scale from 0, absolutely not, to 10, definitely. We are showing the full distribution rather than only the average, because the spread is the honest picture: a clear cluster at the top, and a tail of readers who were not convinced.

Likelihood to use ViewFinder in practice

All 20 delegate responses, King's College Hospital, June 2021

0 · Absolutely not10 · Definitely

In their words

What the readers said it was for

The most useful part of the exercise was not the score. It was that working breast imagers, unprompted, described the problem ViewFinder solves in their own vocabulary, and independently landed on the same use cases.

“This feature is similar to cross-reference lines in multi-planar CT and MR reconstruction machines. It'll help localise exact quadrant of a lesion.”

Consultant radiologist · scored 10

“Confirmation that the area of interest correlates between the two views.”

Consultant radiographer · scored 8

“For dense breast tissue and a single view presentation of an abnormality. For more than one abnormality, when uncertain which correlates to position on other view.”

Delegate · scored 5

“Orientation of lesions, especially useful for small lesions in screening assessment.”

Trainee consultant radiographer · scored 10

“I felt it was a good guideline with uncertain lesions.”

Advanced radiographer · scored 7

And what they wanted fixed

The criticism was as useful as the praise

Delegates were asked how ViewFinder could be improved. Their answers were specific and practical, and they shaped the product that went on to FDA clearance:

Speed. Several readers found the response and loading time too slow, one noting it was “occasionally slow to register when the hand moves”.
Standard image controls. Requests for pan, zoom and scrolling within the tool.
A depth indicator, so the reader can see where they are in the tomosynthesis stack as they scroll.
Interoperability beyond one vendor, asked as “does it work with all imaging software?”
Selective use. One reader captured the sensible position exactly: “not for every case, on individual basis when required”.

What happened next

From a training room to a peer-reviewed study

This was early, informal feedback from a single course, not a clinical trial, and we treat it as such. It told us the problem was real and worth pursuing. The formal evidence came later: a multi-reader study across UK and US centres, published in the European Journal of Radiology, which found localization errors were 86.5% higher without ViewFinder in key cases. ViewFinder is now FDA 510(k) cleared.

Try ViewFinder on live cases at demo.elaitra.com →