Together with Dr. Andries Zijlstra, Associate Professor at Vanderbilt University, we submitted a project for the Strata Data Awards.
The Strata Data Awards recognize the most innovative startups, leaders, and data science projects from Strata sponsors and exhibitors around the world. This project was submitted under the “Data Impact” category, which recognizes projects or initiatives that use data science and analytics to have a widespread, positive impact on society. The project was aimed at improving identification of, and medical intervention for men with aggressive prostate cancer. Below is more information and an overview of the project.
KNIME Image Processing Extension for Biomedical Image Analysis/Analysis of Human Tissue Cells to Aid Diagnosis of Aggressive Cancer.
In spite of many advances in diagnosing cancer patients, few methods can distinguish between aggressive and benign disease. Machine Learning (ML) on cellular features captured through digital pathology holds enormous promise for providing a means to identify cancers with lethal potential before curative intervention is no longer possible. To achieve such transformative implementation of AI in medicine we pursued true integration of single-cell biology, imaging, statistics, computer sciences and informatics using the open computational environment KNIME.
This project’s goal is to achieve improved outcome for men with prostate cancer with personalized/precision medicine; reduce resource allocation through improved diagnosis, prognosis & treatment. To achieve this, a pipeline encompassing all expertise achieved accurate mapping of subcellular alterations associated with aggressive prostate cancer which was subsequently used to assess the risk of future disease progression.
The first milestone of the project was a proof-of-principle that demonstrated the success of integrating multiple non-computational expertise into an open & advanced computational platform without requiring re-training of clinical personnel or translation of mission objectives in computational terms. In excess of 5 million cells from 3000 images taken from 500 patients were used for the discovery and subsequent quantitative analysis of a novel subcellular alteration present in the nucleus of aggressive prostate cancer. These quantiations are subsequently used to predict which patients are at risk of future cancer progression and/or recurrence.
This initiative demonstrates the critical need for open platforms that enable the integration of multiple fields. In particular, the project has demonstrated that KNIME Analytics Platform lowers the threshold for participants across fields to participate in the building of complex AI environments that can be deployed to ask clinical questions. The specific product of this strategy is a computer-guided prediction of patient outcome. It takes into consideration the complexity of the disease at a single-cell level. Beyond the immediate technological gains and academic understanding, this work matches an ongoing evolution in digital pathology that will undoubtedly reshape how patients are diagnosed and how treatment decisions will be made in the clinic.
