Artificial intelligence for diagnostics of ALT-positive cancer (AI4CAN)

In a WWTF-funded project we are currently addressing the need to understand molecular drivers of rare pediatric cancers with very poor outcome, using neuroblastoma as a model for metastatic, aggressive disease, and how this knowledge can be leveraged for better diagnostics and therapy. One of our project results is the development of an assay to detect alternative lengthening of telomeres (ALT), an alteration of chromosome ends associated with ultra-high-risk neuroblastoma and other cancers. The assay combines simple and cost-effective fluorescence in situ hybridization (FISH) with microscopy and artificial intelligence (AI) algorithms for ALT classification. However, application of AI models are not ready-to-use for end-users but need experienced data scientists to be executed, which currently prevents clinical diagnostic labs to test and validate this urgently needed assay. We propose to develop a web-based AI software prototype that will allow diagnostic labs as a new target group to assess ALT in an automated, AI-supported way. The tool will be web-based to enable easy access for collaborating diagnostic labs in Vienna and Germany and future rollout in clinical networks across Europe. A user interface coupled to an image processing pipeline will allow microscopy image upload, execution of AI algorithms and presentation of results as reports. Ultimately, the aim is to test and evaluate the AI prototype and to collect performance results. If successful, we will seek further funding together with our interdisciplinary team of collaborating experts for product development.