Dear AuntMinnie Member,
Large language models (LLM) continue to grow in power and demonstrate potential in a variety of radiology applications. But they also consume a lot of energy, a shortcoming that doesn't jibe well with the specialty’s growing focus on sustainability.
In good news, our featured article reports on a new study that found using small, finely tuned LLMs instead of larger models could yield significant energy savings – without negatively impacting accuracy.
In other AI news, another study this week concluded that an AI model can help detect interval cancers missed by radiologists on screening mammograms. The algorithm was also helpful in localizing these cancers.
Are commercial AI medical devices receiving adequate clinical validation prior to receiving U.S. Food and Drug Administration (FDA) clearance? Maybe not, according to a recent analysis that revealed a lack of published clinical validation data in close to half of FDA marketing authorizations.
Meanwhile, a deep-learning algorithm was shown to be highly accurate for detecting and segmenting cerebral aneurysms.
Other articles we're highlighting in this issue include the following:
- Can AI help rule out ‘unremarkable' chest x-rays?
- Can an open-source large language model make the grade in radiology?
- The PACSMan Pontificates: How can AI save itself?
Be sure to visit our AI content area often to keep up with all of the latest developments in radiology AI technology.
