Accuracy of Medical Image-Based Deep Learning for Detecting Microvascular Invasion in Hepatocellular Carcinoma: Systematic Review and Meta-Analysis - PubMed
7 hours ago
- #deep learning
- #hepatocellular carcinoma
- #medical imaging
- Hepatocellular carcinoma (HCC) is a leading cause of cancer-related mortality worldwide.
- Microvascular invasion (MVI) is a critical pathological indicator of postoperative recurrence and poor prognosis in HCC patients.
- This meta-analysis evaluates the diagnostic accuracy of deep learning (DL) models using medical images for MVI detection in HCC.
- 52 studies with 19,531 HCC patients were included in the analysis.
- DL models showed an overall sensitivity of 0.80, specificity of 0.82, and an SROC of 0.88 for MVI prediction.
- Contrast-enhanced computed tomography (CECT) based models performed excellently with sensitivity of 0.84, specificity of 0.83, and SROC of 0.90.
- DL models using pathological sections achieved the highest diagnostic performance (sensitivity: 0.91, specificity: 0.90, SROC: 0.92).
- Model performance was less consistent in independent external validation (SROC: 0.85) compared to internal validation (SROC: 0.90).
- The study highlights the need for prospective, multicenter studies and integrated algorithms for robust clinical tools.