Acorn 3D Software (AC-SEG-4009); Acorn 3DP Model (AC-101-XX)

The Acorn 3D Software and Acorn 3DP Model system provides a comprehensive solution for converting CT/CTA medical images into accurate digital and physical 3D anatomical models for musculoskeletal and craniomaxillofacial applications. Utilizing a combination of manual, semi-automatic, and machine learning-driven segmentation techniques, the system aims to enhance diagnostic capabilities and treatment planning through precise patient-specific visualizations and tangible replicas.

Acorn 3D Software (Acorn Segmentation) is a medical image processing system designed to assist clinicians in musculoskeletal and craniomaxillofacial applications. It functions by importing CT or CTA medical images in DICOM format, allowing users to visualize, segment, and create digital 3D models of anatomical structures. The software incorporates various segmentation tools: machine learning-based automatic segmentation specifically for thoracic (T1-T12) and lumbar (L1-L5) vertebrae, semi-automatic algorithmic tools, and manual tools for broader musculoskeletal and craniomaxillofacial bone. These digital models can then be used for measurements, treatment planning, and generating output files. The output files from Acorn Segmentation can be utilized for additive manufacturing to create Acorn 3DP Models, which are physical replicas of the patient's anatomy. These models are fabricated using a biocompatible PA-12 polymer powder with HP multi-jet fusion systems and are intended for diagnostic purposes. Performance data demonstrated the software's accuracy, with segmentation methods achieving an average Dice-Sorenson coefficient exceeding 0.93 when compared to predicate devices. Furthermore, the resulting physical 3DP models showed high geometric fidelity, with a mean deviation of less than 1mm from their corresponding digital 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. How does the performance of the machine learning-based auto-segmentation for thoracic and lumbar spine compare to experienced human segmentation in terms of time efficiency and accuracy for complex pathological cases?; What is the range of anatomical variability (e.g., severe deformities, presence of implants) within the musculoskeletal and craniomaxillofacial regions for which the semi-automatic and manual segmentation tools maintain the reported accuracy (Dice-Sorenson > 0.93)?; Beyond geometric accuracy (Dice-Sorenson, <1mm deviation), what specific diagnostic benefits or changes in treatment planning have been observed when using the Acorn 3DP Models compared to relying solely on 2D/3D digital images?; What are the specific computational requirements (e.g., processing time, hardware specifications) for the machine learning auto-segmentation feature, and how do these scale with increasing image dataset complexity?; Given the use of PA-12 polymer for the 3DP models, what is the expected durability and sterilizability of these models if they were to be considered for intraoperative reference or patient-specific surgical guide fabrication, considering the current indication is diagnostic?