ATP Barcelona 2017

QA methods • The algorithm works technically • Use phantom or simulated data

• The program works in general • Best: use patients with implanted markers (data scarce) • Second: compare with human observers

• The program works for this patient • Visual verification • Consistency, plausibility