CuraMate AI Development

CuraMate AI Development is a highly customizable and extensible software platform for curating medical images, making it ideally suited for preparing data for the training and evaluation of AI models.

The platform integrates seamlessly with other AI development tools and streamlines the collaborative data curation process - starting with quality control, continuing through image segmentation, and annotating subjects, timepoints, or individual images, and even directly integrating an AI training loop.

The training “loop” refers to the iterative process of using the best currently available models to pre-segment a case, enabling experts to correct any AI-made errors, and then triggering the next training step to produce an improved model for the following round. Uncertainty estimation is also built in to guide users toward cases and regions that are likely to contain errors.

CuraMate (formerly known as SATORI) guides users through complex curation processes, tracking the state of each case to ensure every user takes the right step at the right time. Consistency checks help prevent errors without getting in the annotator’s way. Users do not have to explicitly save their work; instead, workflow management allows cases to be marked as “finished” (once all consistency checks pass), automatically passing them on to the reviewer’s work list.

Guided Data Curation

CuraMate AI Development guides readers through a specific data curation workflow. Study administrators can assign cases to readers, and CuraMate tracks the state of each individual case within worklists. It is possible to define several steps with specific hangings, form fields and consistency checks, such that CuraMate ensures that all data is brought into a consistently enriched state during the curation process.

This ensures reliable data quality for AI training and evaluation.

AI Preprocessing

AI is integrated in multiple ways. AI-based algorithms can be used for preprocessing cases, creating preliminary segmentations that should be checked and possibly corrected by users, for instance.

AI Training

CuraMate comes with an optional extension offering the necessary components for AI training: Model management, assignment of cases to training / validation / test sets, and managing training and inference jobs.

Smart Tools

Manual structure editing has been refined over many years for maximum efficiency, and we also integrate the latest AI based segmentation tools. Workflows are configured to offer the right tools for the job and not overburden users with long toolbars.

CuraMate AI Development streamlines the complex workflows of collaborative AI development projects and allows data scientists and physicians to focus on the important things: the quality of the curated data and the derived AI models.

Radiologists and other clinical users can use CuraMate to perform annotations, rate and comment on cases, mark or delineate structures of interest, review and correct AI results, or validate complex algorithms and application prototypes.

Data scientists (either in hospitals, in commercial research departments, or in CROs) can make use of CuraMate AI Development to set up convenient annotation workflows for their annotators. They can manage worklists, track progress, and benefit from consistently annotated data through the guided process that helps to increase overall data quality.

CuraMate AI Development is an integrated toolkit for collaborative research and development of imaging AI models.

We are always looking for research projects in which we use this toolkit together with our partners to develop AI models efficiently and conveniently.

Commercial research groups may also license our technology to use it without us. We would provide support for integration into the desired target environment.

Alternatively, we could also take over the development of AI models completely in a contract research setting.

CuraMate is continuously improved and extended. New features are typically developed as CuraMate extensions and thus can be made available for various projects and customers.

As the AI ecosystem is developing so rapidly, the integration of the latest AI tools (as for example for interactive segmentation) is one area of constant improvement.

Another area that's always relevant is the user experience; we constantly strive to make the UI more intuitive and to reduce the effort required from human expert readers.