Deciphering tumour tissue organization by 3D electron microscopy and machine learning

Area of Science:

• Oncology
• Bioengineering
• Microscopy

Background:

• Tumor 3D organization and architectural parameters remain poorly understood.
• Hepatoblastoma is the most common childhood liver cancer.

Purpose of the Study:

• To analyze the spatial and 3D organization of hepatoblastoma patient-derived xenograft tissues.
• To identify bioarchitectural parameters influencing tumor internal architecture.

Main Methods:

• Serial block-face scanning electron microscopy for 3D imaging.
• Integrated workflow: 3D imaging, segmentation (manual & machine learning), mathematics, infographics.
• Digital reconstitution of entire hepatoblastoma samples, including cells, organelles, and vasculature.

Main Results:

• Hepatoblastoma cell size correlates with nucleus, cytoplasm, and mitochondrial mass.
• Anatomical connections observed between blood capillaries and tumor cell planar alignment/size.
• Tumor cells showed polarization towards a bile canaliculus-like structure.

Conclusions:

• Identified novel bioarchitectural parameters shaping tumor internal and spatial organization.
• This pilot study provides a foundation for the emerging field of onconanotomy.
• Highlights the importance of 3D ultrastructural analysis in cancer research.