Artificial intelligence technology shows promise in facilitating the analysis and interpretation of pathology images. Despite the many benefits of integrating AI into pathology workflows, adoption of such novel tools has been slow. Roche seeks to enable rapid adoption of AI tools via the Roche Open Environment.
Open digital pathology environments can break down barriers to allow for rapid technology advances across industry and research partners, which is critical in developing next-generation diagnostic tests.
What is the Roche Digital Pathology Open Environment?*
The Roche Digital Pathology Open Environment allows software developers to easily integrate their image analysis tools for tumor tissue with Roche’s navify® Digital Pathology. The Open Environment allows for the secure exchange and flow of data so that pathologists can access advanced algorithms from third parties alongside Roche’s menu of artificial intelligence-based image analysis tools. This enables software developers globally to distribute their digital products through Roche’s navify Digital Pathology, offering a broader set of AI tools for pathologists and ultimately, the potential for better and faster answers.
Innovation in digital pathology
Pathology is undergoing a transformation towards digitization. A workflow solution that can enable pathologists to efficiently complete their daily workload with easy access to new digital tools is essential.
As artificial intelligence tools for pathology workflows continue to develop, Roche seeks to break barriers and expand access to AI algorithms with the Roche Digital Pathology Open Environment.
This expanded access will accelerate further image analysis algorithm development, building a future where integrated AI pathology workflows are more widely leveraged.
Featured products
Benefits of the Digital Pathology Open Environment from Roche*
Roche Digital Pathology Open Environment has several key advantages:
- Provides access to image analysis algorithms from third parties as well as Roche’s suite of AI-based algorithms
- Enables fully integrated third-party algorithms within navify Digital Pathology
- Multi-AI workflows allow any combination of third-party algorithms to be run side by side with Roche algorithms
- Platform-based architecture ensures secure exchange of data between navify Digital Pathology and third-party AI vendors
Roche has now integrated more than 20 AI algorithms from 12 new collaborators✝ into its digital pathology Open Environment:
✝Please note that the collaborator tools other than Ibex and PathAI are products in development, have not been commercialized yet in the US and are subject to change without notice. Timelines for launch of partner tools vary and will be communicated upon commercialization in the US. All collaborator algorithms are for Research Use Only (RUO)* and are not for use in diagnostic procedures.
Avoiding unfair bias in AI algorithm development is critical to ensure that the models are fair, ethical, and effective across different populations. Here are some strategies to minimize bias:
Diverse and Representative Data:
Data Collection: Ensure that the data used to train AI models is diverse and representative of all relevant groups. This includes considering factors such as race, gender, age, other variables such as histological subtype, tumor differentiation etc.
Balanced Datasets: Address imbalances in the data by either collecting more data for underrepresented groups or using techniques like oversampling, undersampling, or synthetic data generation to balance the dataset.
Transparent and Explainable Models:
Model Explainability: Use interpretable models or add explainability tools to understand how the model makes decisions. This helps identify potential biases in the decision-making process.
Transparency: Document the model development process, including data sources, feature selection, and any steps taken to address bias. This transparency helps stakeholders understand the model’s limitations and potential biases.
Inclusive Team and Stakeholder Involvement:
Inclusive Development Teams: Encourage thorough and inclusive representation within the AI development team, as inclusive perspectives can help identify and address potential biases that might be overlooked by a more homogeneous group.
Stakeholder Engagement: Involve stakeholders in the development process to gather feedback and identify biases that the development team may not have considered.
PathAI PD-L1 (28-8) Bundle RUO:
- PathAI PD-L1 (28-8) HNSCC
- PathAI PD-L1 (28-8) Melanoma
- PathAI PD-L1 (28-8) NSCLC
- PathAI PD-L1 (28-8) Urothelial Carcinoma
Ibex RUO:
- Ibex Breast
- Ibex Prostate
Working together to drive innovation: A look at our collaborators
Related health topics
VENTANA and NAVIFY are trademarks of Roche.
*For Research Use Only. Not for use in diagnostic procedures. Roche is dedicated to clinically deploying its Digital Pathology Solutions and is actively pursuing approval/clearance for our portfolio of products.
Roche Digital Pathology Dx, including navify Digital Pathology, is cleared for specific clinical applications and is intended for research and educational use for other applications. Please review product labeling prior to utilization.
Note that partner collaborators other than Ibex and PathAI are not yet launched and are subject to change.