引導式測驗 · 中等 等級
人工智慧在血液抹片和血液學分析測驗的應用
測試數位形態、瘧疾影像分析、幻燈片驗證以及細胞標誌和診斷之間的差異。
問題 1 的 8
人工智慧數位形態學工具可以對血液抹片影像做什麼?
預覽每個問題8 問題
當您參加測驗時,就會出現答案和解釋。
- 人工智慧數位形態學工具可以對血液抹片影像做什麼?
- Estimate kidney risk from serial creatinine results
- Classify or flag cells for trained review
- Score fibrosis on a liver-biopsy slide
- Predict cancer origin from plasma DNA
- AIDMAN 前瞻性塗片驗證納入了多少患者?
- 64 image fields from one test slide
- 640 patients across multiple hospitals
- 64 hematology laboratories
- 64 patients with thin blood smears
- 哪些資訊有助於解釋異常的細胞標誌?
- A ranked list of model labels without checking the slide
- The model confidence value without other laboratory results
- The original smear, blood count and clinical context
- A training-set accuracy figure without patient-level context
- 哪些來源會導致影像模型效能漂移?
- Different stains, scanners or magnifications
- The use of an electronic lab order
- A clinician correcting a flag
- A cell being labeled on screen
- 哪一種後續措施可以進一步表徵細胞形態以外的疑似血液惡性腫瘤?
- A second image classifier on the same smear only
- Flow cytometry or molecular testing
- A symptom questionnaire by itself
- A routine cell count without morphology review
- Why is a high cell-level score not proof that a system can rule out disease?
- A high cell score proves the model works in every lab
- An image-level measure alone shows the test rules out disease
- Model confidence replaces a clinical reference standard
- Cell classification does not automatically establish patient-level performance
- What should a useful laboratory workflow preserve?
- Automatic acceptance of the first classification
- Using AI output as the final treatment decision
- Access to original images and professional correction of labels
- Removing manual review from the workflow
- What does a cell-level AI flag establish by itself?
- Only that the software marked an image for attention
- A confirmed disease diagnosis
- A required medication
- That additional testing is unnecessary