Drawing upon his neuroscience background, Dr. Diamandis discovered that AI organizes patterns in a manner comparable to humans. He and his team started a research project to train AI to recognize tumour histology slides. They fed AI nearly one million images from over 1,000 brain tumours, each annotated by pathologists. Using deep neural networks, they developed a tool called “HAVOC” (Histomic Atlases of Variation of Cancers) to analyze cell patterns and map distinct regions of tumour with unique features.
HAVOC helps researchers understand tumour heterogeneity—a phenomenon where different regions of a tumour have different biology, leading to varying treatment responses. The team tested HAVOC’s accuracy with images from high-grade glioblastomas. HAVOC’s results were consistent with human experts’ interpretations and correlated to molecular variations in samples, supporting personalized combination therapies.
“HAVOC can assist pathologists to understand tumour heterogeneity directly from histology slides,” says Dr. Diamandis. “It’s a useful tool to complement other molecular approaches and contribute to personalized medicine.”
“The danger of AI is that we are not really sure how it behaves in new situations, “says Dr. Diamnadis. AI can make errors with rare tumours due to insufficient training data. To address this, Dr. Diamandis’ team developed PHARAOH (PHenotyping and Regional Analysis of Histology), an open platform allowing researchers to customize AI tools for specific needs and share them with the community.
The rise of AI poses uncertainty, but Dr. Diamandis embraced it to empower the field. “If you’re scared of something, it’s probably something that will make you stronger if you eventually embrace it,” he said.
Thank you for helping drive groundbreaking research discoveries that bring us closer to Conquering Cancer In Our Lifetime.
