Gids-gekoppelde quiz · Moeilijk Niveau

AI in de oncologiequiz

Tests waarbij het bewijs van oncologische AI het sterkst is, lessen van Watson voor oncologie, en hoe slide tiling en Dice-scores werken in pathologie en radiotherapie.

Gerelateerde gidspadenAI in de oncologie
Vraag 1 van 8

What did the Swedish MASAI trial report about AI-supported mammography screening?

Bekijk elke vraag8 vragen

Antwoorden en uitleg verschijnen terwijl je de quiz doet.

  1. What did the Swedish MASAI trial report about AI-supported mammography screening?
    • It eliminated the need for radiologists
    • It detected more cancers while cutting screen-reading workload by roughly 44 percent
    • It reduced cancer detection but saved money
    • It only worked for lung cancer scans
  2. What was the key lesson from IBM's Watson for Oncology?
    • Training on hypothetical cases and one institution's preferences does not produce reliable general treatment advice
    • AI cannot read any medical text
    • Oncologists refused to use computers
    • It was too accurate to be trusted
  3. Which product did the FDA authorize in 2021 as the first AI to assist pathologists with prostate biopsies?
    • Sybil
    • MASAI
    • Paige Prostaat
    • Navify
  4. What does Sybil, from MIT and Massachusetts General Hospital, predict?
    • Breast density from mammograms
    • Radiotherapy doses
    • Prostate cancer grade
    • Future lung cancer risk from a single low-dose CT scan
  5. How do most AI systems handle gigapixel whole-slide pathology images?
    • They shrink the slide to a single small thumbnail
    • They tile the slide into patches and aggregate them with multiple-instance learning
    • They convert the slide to text first
    • They only read the slide's label
  6. Why is a high Dice score not enough to judge a radiotherapy contour?
    • Dice only measures color accuracy
    • Dice is measured in seconds
    • A small error near a critical structure like the spinal cord can matter more than overall overlap
    • Dice scores are always 100 percent
  7. Why does the guide say finding more small cancers is not automatically a benefit?
    • Small cancers cannot be treated
    • It only helps if it improves survival rather than increasing overdiagnosis
    • Radiologists dislike extra findings
    • Screening programs have fixed quotas
  8. What risk does the guide highlight when using large language models to summarize records for tumor boards?
    • They are too slow for weekly meetings
    • They cannot read pathology reports
    • They require patients to attend
    • Hallucinated or outdated facts