Abstract B040: Applying histopathology foundation model-enhanced neural networks to ovarian tumor whole slide images and tissue microarray for survival prediction and tumor microenvironment subtyping

Key Points

• Neural networks trained on tumor microarrays and whole slide images achieved strong survival prediction performance (c-index=0.76-0.96).
• The study identified distinct tumor microenvironment subtypes, particularly distinguishing Immune-Enriched + Fibrotic and Depleted clusters.
• Foundation model features enhanced neural networks for analyzing hematoxylin and eosin-stained ovarian tumor samples.
• Despite challenges, including sample size and spatial resolution, TMA-derived features were crucial for capturing important biological signals.