HistoGx: Histomic integration with multi-omic data for massively scalable precision oncology

Creating an artificial intelligence tool to enable the development of more accurate and clinically meaningful biomarkers to guide patient care

Project summary

As part of the diagnostic workup for all patients with cancer, a tissue sample is a standard of care. These tissues can be mounted on glass slides and examined by a pathologist to help with diagnosis and in many cases to determine how aggressive this cancer is. Computational pathology utilizes advanced computing, such as artificial intelligence, to extract useful information from pathology images. Using computational algorithms, these images can help improve diagnostic accuracy, provide accurate outcome prediction, and even to predict molecular features, such as the type of mutation a patient may have. Unfortunately, this extremely valuable information is not currently used in the clinic due to barriers including lack of standardization and lack of generalization.

We propose to develop a streamlined, reproducible workflow, called HistoGx, to process and analyze pathology images. We will then use this information to develop ways to identify which patients are more likely to respond to their cancer treatments. Given the complexity of this task, we will also incorporate other molecular information from cancer patients including their genetic mutations, expression patterns of certain genes as well as their detailed clinical information. Combining different data types will enable us to develop more accurate and clinically meaningful biomarkers to help clinicians make decisions about patient care.

As pathology images are cheap and widely available, this technology already exists at every cancer centre in Canada. Developing this tool and making this available to all researchers will significantly improve our ability to match the right treatment to the right patient at the right time.

Quotes

“I am deeply humbled to be receiving the MOHCCN HI&DS Award. As an oncologist and cancer researcher, there is a significant gap in how scientific discoveries are translated to benefit patients. Perhaps one of the biggest hurdles to precision oncology is the lack of access to comprehensive sequencing and profiling technologies. This pivotal support will allow us to develop a tool to harness a cheap and widely available histology slide as part of the clinical workflow. This novel approach will accelerate our mission to empower clinicians with precise, personalized approaches.”

• Dr. Kevin Wang, HI&DS Awardee
  

“Kevin’s transformative project, HistoGx, will leverage computational pathology and AI to revolutionize cancer care. This initiative promises to significantly enhance diagnostic accuracy, predict treatment responses, and promote precision oncology across our network in a tumour-agnostic way. Together, we pave the way for a future where every patient receives tailored care, maximizing outcomes and quality of life.”

• Dr. Benjamin Haibe-Kains and Dr. Robert Grant, mentors