Artificial intelligence in ultrasonography: what already exists today

Automatic structure recognition, assisted measurements, and image acquisition support — without replacing medical judgment.

Unlike other areas of radiology, artificial intelligence applied to ultrasonography faces an additional challenge: the image depends directly on the examiner's hand. This makes acquisition automation — not just interpretation — as important a frontier as analysis of the final image.

What's already in use

Pattern-recognition algorithms already assist with automatic identification of anatomical structures, assisted measurement of fetal biometry, automatic estimation of ejection fraction on basic cardiac windows, and flagging suspicious nodules on thyroid and breast screenings.

What AI does not do

No current tool replaces clinical correlation, the decision about what to investigate next, or the responsibility for the final report. The real role of artificial intelligence today is to reduce acquisition time, standardize measurements, and flag findings for review — never to replace medical judgment.

This distinction isn't just ethical — it's also where the current scientific evidence actually supports the use of the technology.