EcSeg: Semantic Segmentation of Metaphase Images Containing Extrachromosomal DNA

Area of Science:

• Cancer Biology
• Genomics
• Computational Biology

Background:

• Oncogene amplification is a key driver of cancer development and progression.
• Amplified oncogenes frequently reside on extrachromosomal DNA (ecDNA), contributing to tumor heterogeneity.
• Current methods lack the spatial resolution to quantify oncogenes on ecDNA within single cells.

Purpose of the Study:

• To develop a novel method for accurately quantifying oncogene amplification on ecDNA at the single-cell level.
• To address the limitations of existing sequencing technologies in resolving ecDNA-associated oncogene copy number.
• To investigate the role of ecDNA-located oncogenes in cancer pathogenesis.

Main Methods:

• Integration of conventional microscopy with deep neural networks.
• Development of an image analysis tool named ecSeg.
• Application of ecSeg for single-cell resolution of ecDNA and oncogene amplification.

Main Results:

• ecSeg accurately resolves ecDNA and oncogene amplification at the single-cell level.
• The tool provides spatial resolution for amplified oncogenes, overcoming limitations of next-generation sequencing.
• Enables precise quantification of oncogenes located on ecDNA.

Conclusions:

• ecSeg is a powerful new tool for studying ecDNA-driven oncogene amplification in cancer.
• This approach will advance our understanding of how subnuclear oncogene location impacts tumor pathogenesis.
• Facilitates research into intratumoral genetic heterogeneity driven by ecDNA.