AI Application Development

Our experience developing regulated Software as Medical Devices (SaMDs) and non-FDA regulated applications for leading healthcare delivery organizations and medical technology manufacturers around the world is brought to every project we undertake. The medical R&D team is at home with DICOM data from modalities such as Ultrasound, CT, MRI, X-Ray, Whole Slide Imaging (WSI) data, and non-DICOM data. Example projects have included:

  • Computer-assisted diagnosis (CADx)
  • Computer-assisted detection (CADe)
  • Computer-assisted diagnostic quantification (CADq)
  • Computer-assisted triage (CADt)
  • Biomarker discovery (radiomics)

The intellectual property assets created by custom AI R&D projects are typically assigned to the client, thus providing an enduring asset to the organization.

Image Annotation
Our imaging analytics center provides cost-effective annotation and reporting services to clients. We send DICOM image files to the imaging analytics center where they are processed by our trained technical staff under the supervision of radiologists. We offer customized solutions for research-related image annotations and reports, catering to the needs of our clients.
PHI De-Identification

National regulations in most countries require organizations that receive Protected Health Information (PHI) to apply both technical and administrative controls to protect the security and privacy of data. We have implemented controls such that a dedicated, separated team can de-identify PHI in a highly controlled environment. Data is then exported to a larger AI team in a PHI-cleansed state, limiting risks and allowing it to be used in research and other compliant applications. This satisfies such regulations as HIPAA in the USA and GDPR in the EU.

Data Curation and Preparation
Machine learning applications are especially sensitive to the quality of the data and associated annotations used to train models. Clinical data that are used for training or clinical validation also require the ethical use of data so that regulatory requirements for security, privacy, and informed consent are honored. For supervised learning requiring annotations, we can help with techniques including the use of specialized annotation tools custom-designed for efficiency, natural language processing tools, and the use of trained annotators to create high-quality data resources for training machine learning models cost-effectively. During the algorithm development process, we can also augment data for machine learning purposes to improve model performance and scalability.
Algorithm Development

Our research and development team develops deep learning models that interpret medical data, including imaging data and electronic clinical reports. Our research and development team continuously monitors new developments in the field of machine learning. We work with our clients to build computer vision algorithms for detection, classification, segmentation, and measurement from medical images. We also apply machine learning to natural language processing so that we can, for example, turn unstructured text reports into structured data. These algorithms can be embedded in new hardware devices, deployed as software medical devices, and used for decision support applications.

Scalable Software as a Service (SaaS) Engineering
Our software engineering team brings experience building applications using advanced user interfaces, scalable web server technologies, and cybersecurity for deployment across cloud, mobile, and other devices to every project. Our software engineering organization has implemented a quality management system that is ISO 13485 certified.
Proofs of Concept, Prototypes, and Demos
Our clients sometimes are seeking funding or organizational approval to build innovative applications. It can be very helpful to test hypotheses by developing models to prove- or disprove- that a given science problem is solvable before investing more resources. We develop prototypes to elicit clinician feedback about usability and workflow relevance, and product demos that can be used to get investments.
Clinical Integration
Bringing AI into the clinical workflow often involves integrating with PACS, VNAs, EMRs, DICOM routers, enterprise worklists, and other IT solutions. Accomplishing seamless integration requires expertise in technologies including RESTful APIs, HL7, FHIR, and proprietary interfaces. Our team works alongside you to bring your application into the clinical workflow, accomplishing seamless integration every time.
UX Research and Design

Our in-house UX researchers and designers work with clients to discover opportunities, support development, and design pixel-perfect user interfaces.

  • Research: user research, user modeling, task analysis, requirements definition
  • Design: ideation, prototyping, detailed design, Complete Design Specifications
  • Development: preliminary or conceptual design, design development, user interface specification, development consultation

Let's start something new together.