Harnessing Data to Power the Future of Pathology
At Deciphex, we are leveraging an unprecedented array of digital pathology datasets to build cutting-edge AI models that are transforming diagnostics worldwide. By partnering with key organisations like Novartis, Charles River Laboratories (CRL), and BigPicture, and drawing on datasets from Diagnexia and other clinical platforms, we are creating AI-driven solutions that provide unparalleled diagnostic precision across multiple species, diseases, and research environments.
Diagnexia and Global Pathology Datasets
Our Diagnexia platform serves as a key data hub, connecting pathologists globally and generating a wealth of diverse clinical pathology datasets. Through Diagnexia, we gather real-world diagnostic data from a broad range of histologies, including cancer, gastrointestinal, and dermatopathology cases, which are integral to training our AI models. This vast collection of clinical data helps ensure that our AI can deliver accurate and timely diagnoses across various disease processes, improving patient outcomes globally.
BigPicture: A Pan-European Initiative
As part of the BigPicture project, Deciphex is contributing to and drawing from one of the world's largest digital pathology image repositories. This pan-European initiative aims at collecting three million of pathology slides to train AI models for both clinical and non-clinical applications. The data from BigPicture enhances our ability to build robust, multi-institutional AI models that are trained on a wide range of normal and diseased tissue, making our solutions more versatile and adaptable to different diagnostic environments.
Collaboration with Charles River Laboratories (CRL)
Our collaboration with Charles River Laboratories (CRL) plays a critical role in expanding the preclinical datasets available for AI development. By partnering with CRL, we have access to toxicological pathology data, particularly from rodent studies, which is crucial for building AI tools that assist in identifying lesions and abnormalities in **preclinical safety evaluations**. These datasets are used to improve Patholytix Foresight, our AI model designed for toxicology, which supports faster and more accurate preclinical assessments, reducing the time and cost associated with early drug development.
Novartis Collaboration for Multispecies AI Models
In collaboration with Novartis, we are developing next-generation unsupervised AI models that power lesion detection across multiple species. This partnership provides access to extensive non-clinical datasets from multi-species safety studies, which are critical for training AI models to recognize pathological changes in different animal species. By combining data from human pathology with these multi-species datasets, we are building AI solutions that accelerate drug development by ensuring comprehensive drug safety evaluations.
Comprehensive Data Collections Driving AI Innovation
In addition to these partnerships, Deciphex is continuously expanding its data repository through collaborations with academic institutions, CROs, and pharmaceutical companies. Each of these partnerships adds valuable, diverse datasets to our AI development programs, allowing us to refine our models for both clinical and research applications. From routine diagnostics to non-clinical safety studies, these datasets empower our AI to evolve continuously, making it more robust, reliable, and adaptable to different diagnostic contexts.
Revolutionising Global Diagnostics Through Data-Driven AI
By integrating datasets from Diagnexia, BigPicture, CRL, Novartis, and other data collection initiatives, Deciphex is revolutionizing the future of AI-driven pathology. Our AI models are built on the world’s largest, most diverse pathology image repositories, ensuring their accuracy and applicability across multiple diagnostic fields. These collaborations and datasets fuel continuous innovation, driving faster, more accurate diagnoses, and improving both clinical outcomes and non-clinical drug safety evaluations globally.
Resources
IMI BIGPICTURE Project Consortium
