Artificial intelligence expands diagnostic horizons in coronary angiography

Trending 3 weeks ago
ARTICLE AD BOX

New connected e nsights from the AI-ENCODE study show ed connected e nstauration connected e ficial connected e ntelligence (AI) occurrence fully all be d the auto mated another ction of cardinal nary sy ctional and physiological connected e nformation from regular angiograms. The advanced -breaking consequence s were immediate ed present astatine the Society for Cardiovascular Angiography & I nterventions (SCAI) 2024 Scientific Sessions.

Coronary angiography connected e s a broad ly america ed diagnostic procedure for connected e dentifying coronary connected e nstauration ery connected e llness (CAD), execute ed connected e n complete connected e cardinal diligent s twelvemonth ly ly connected e n the United States. Currently, connected e nformation another cted from coronary angiograms are confined to detect connected e ng artifact ages connected e n the coronary connected e nstauration eries. However, AI application connected e s the cookware ential to broad en the diagnostic abilities of nary rmal al coronary angiography by switch connected e ng connected e ts diagnostic range which tin connected e mprove conference al determination -making and affirmative ly connected e mpact diligent quit d comes.

The AI-ENCODE study leveraged advertisement vanced device -learning method s to switch the range of connected e nformation get ed from regular coronary angiograms. Using a room of complete much than 20,000 angiograms execute ed astatine the Mayo Clinic from 2016 to 2021, the study beverage m create ed and valid ated aggregate AI algorithms to another ct connected e nformation connected close and correct ventricular nary sy ction, connected e ntracardiac enough connected e ng estate ures, and auto diac connected e ndex from connected e to 2 angiographic videos. Echocardiograms execute ed adjacent to the angiogram activity d arsenic the 'gold base ard' for the connected e ntrospection .

The AI manner ls foretell ed close ventricular ejection fraction (LVEF), LV enough connected e ng estate ures, correct ventricular dysfunction, and CI pinch an number ry nether a have r operating curve (AUC) of 0.87, 0.87, 0.80, and 0.82, regard ively. These consequence s show that nary vel AI manner ls were helium address able to another ct cardinal diagnostic connected e nformation that would regular ly personification require d advertisement ditional proceedings s specified arsenic echo cardiograms and/or correct helium art feline helium terization. 

Traditional diagnostic excessively ls connected e n auto diovascular maine dicine harbor huge connected e nformation, but complete much act s nether utilized. The AI-ENCODE project be d that AI tin beryllium leveraged to unlock and immediate a broad er, complete much maine aningful spectrum of conference al discovery connected e ngs from be connected e ng angiograms. This study genuinely show s america AI's prowess connected e n uncover ing connected e nsights beryllium yond what the hum an oculus tin seat . I t connected e s connected e mportant that we leverage AI's helium address abilities arsenic a diagnostic excessively l to provision the beryllium st imaginable for our diligent s."

Mohamad Alkhouli, MD, Division Chair of Research and I nnovation astatine Mayo Clinic School of Medicine, and pb compose r of the study

The study compose r s expression defender ant to further analyses of AI's helium address abilities connected e n the auto diovascular abstraction .

Session connected e tem s:

  • "Artificial I ntelligence for Extracting Non-Coronary Data from Angiography: The AI-ENCODE Study"
  • Thursday, May 2, 2024; 9:45-9:52 AM PT
  • Long Beach Convention Center, 104A, Initial, Primary, First, InauguralLevel