Objective:

To develop AI platforms that infer molecular information from pathology slides, making insights more accessible and faster for clinicians.

Approach:

• Path2Omics: Predicts bulk molecular information from pathology images, focusing on inferred bulk transcriptomics and methylation across 30 tumor types.
• Path2Space: Aims to infer spatial transcriptomics from digital pathology images by linking image morphology to spatial molecular data.

Key Findings:

• AI can potentially reduce the cost and turnaround time for obtaining molecular insights from weeks to days.
• Path2Omics provides inferred bulk omics data from standard pathology slides without the need for extensive molecular profiling.
• Path2Space aims to infer spatial transcriptomics from digital pathology images, linking image morphology to spatial molecular data.

Interpretation:

Limitations:

• The effectiveness of inferred spatial information in improving decision-making remains to be fully validated.

Conclusion: