Axial3D Insight

Axial3D Insight is a cloud-based service that transforms DICOM medical scanner images into digital output files and physical 3D printed replicas to support treatment planning and diagnostic processes in orthopedic, maxillofacial, and cardiovascular fields. The system's methodology relies on an image segmentation framework featuring machine learning for initial segmentation, which is then critically reviewed and finalized by trained personnel, with its output validated through clinical performance and usability studies.

Axial3D Insight is a cloud-based medical image processing software designed as an image segmentation framework. It converts DICOM imaging data from medical scanners, specifically CT and CTA modalities, into a digital output file. This output file serves a dual purpose: it can be used directly or for the fabrication of patient-specific physical replicas using additive manufacturing techniques. The resulting digital files or physical models are intended for use in treatment planning and as diagnostic aids in orthopedic trauma, general orthopedics, maxillofacial surgery, and cardiovascular applications. It is emphasized that Axial3D Insight should be utilized in conjunction with other standard diagnostic tools and the expert clinical judgment of medical professionals. The system operates as a secure, cloud-based service, providing 3D modeling capabilities. Its segmentation process combines automated tools, including AxialML machine learning models for initial segmentation, with smart editing tools operated by trained Axial3D staff who perform final validation of the model quality. The device underwent comprehensive validation, including software design verification. Clinical segmentation performance was assessed by three radiologists reviewing twelve cases across its indicated applications, with all cases meeting predefined acceptance criteria (RADPEER scores 1 or 2a). An intended use study involving nine physicians also successfully validated the utility of the 3D models for end-user needs and indications. The AxialML machine learning models were independently verified and validated using thousands of CT/CTA images from a diverse range of scanner manufacturers and models.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. What specific algorithms are employed by the AxialML machine learning models for image segmentation, and how are they optimized for different anatomical regions (orthopedic, maxillofacial, cardiovascular)?; The document mentions RADPEER scores of 1 or 2a for clinical segmentation performance; what was the inter-rater reliability among the radiologists, and were any specific case types or anatomical complexities more challenging for segmentation?; How does the system ensure the accuracy and dimensional fidelity of the 3D printed physical replicas in relation to the digital output file, particularly concerning material properties relevant for surgical planning?; What are the typical turnaround times from DICOM image upload to the availability of the final output file and/or physical replica, and how might this vary with case complexity or anatomical region?; Beyond CT and CTA modalities, are there plans to expand Axial3D Insight's compatibility to other imaging modalities like MRI or ultrasound, and what are the anticipated challenges in adapting the segmentation algorithms?