We Built a Radiology Workflow Tool and Got It FDA Cleared in 14 Months

Built and cleared a DICOM image processing SaMD in 14 months. Full-cycle: engineering to 510(k) clearance.

Tim Szczykutowicz, PhD

Working with Innolitics has been a fantastic experience. Their technical and regulatory expertise is exceptional—it's not just programming; it's science combined with a deep understanding of FDA regulations. Interacting with their team feels like collaborating with fellow scientists who navigate the complexities of regulatory compliance effortlessly. Their mastery in detailing the ends of the slice ranges (they'll know what I mean) and guiding us through the FDA process has been impressive. I've recommended them multiple times and will continue to do so.

Tim Szczykutowicz, PhDCEO
ClientMedical areaRadiology FDA RadUnity K242810 · cleared 2024 Software DevelopmentWeb-AppImage ProcessingDICOM510k

The Problem

In radiology, the variability in imaging parameters across different CT scanners and clinical sites leads to inconsistent image presentations. This inconsistency forces radiologists to adapt to varying image qualities and presentations, potentially leading to inefficiencies, fatigue, and challenges in accurate diagnosis. The solution aims to standardize image presentation, streamline workflow, and enhance radiologists' efficiency.

What We Did

Together with the RadUnity team, we developed a software medical device designed to aid in the management and processing of CT images. We also prepared the FDA 510(k) submission, performed software validation, and implemented cybersecurity controls necessary for a successful 510(k).

Key Accomplishments

  • Built a data-intensive web application capable of:
    • Sending/receiving DICOM CT images
    • Automatically mapping DICOM metadata to pre-configured image-processing configurations
    • Performing image processing jobs on DICOM image data (to reorient, reslice, and resample images)
    • Presenting processing status and user configurations in a modern web UI dashboard
  • Completed the full software engineering lifecycle and achieved 510(k) clearance within 14 months.
Clinicians are shown a dashboard, displaying the status of received DICOM data, and allowing them to take any needed actions.
Administrative users configure “mappings” that are used to automatically link DICOM data received by the application with a “profile” of image processing configurations. E.g., a Chest CT might be configured to automatically produce axial and coronal views with certain desired slice thicknesses.
A user can select a subset of the extent of a CT scan to be processed using sliders in a coronal view (shown in the center). Previews in the axial plane for the endpoints of the selected extent and midpoint are displayed (shown on the right).

Key Technologies

  • A modern React-based web UI
  • A Python-based web server
  • A Python-based image-processing server
  • An Orthanc-based DICOM sender/receiver

Timeline

October 31, 2023: Scoping and strategy complete

November 30, 2023: Presubmission meeting

February 4, 2024: Backend implementation complete

May 1, 2024: Frontend implementation complete

September 18, 2024: 510(k) Submitted

November 26, 2024: 510(k) Cleared!

Related case studies

We went from idea to FDA clearance in 18 months and first Install for AI Tumor Tracking SoftwareRadiologyCancer Annotation SaMD ManufacturerWe Ran a Gap Analysis — They Passed Their FDA AuditRadiologyWe embedded with their engineering team and delivered work they could maintain long after we left.Radiology

Let's Talk

Every great partnership starts with a conversation. Fill out the form below for a discovery call, and an Innolitics team member will contact you soon.