Vent Creativity Knee v1.0 (Hermes)

Hermes is a software system that processes CT scan images to segment and identify knee bones (femur, tibia, fibula, patella) and calculate surgical cut planes for preoperative planning in orthopedic applications. The software generates 3D bone models and landmarks from DICOM files, which can be used for diagnosis, surgical planning, and fabrication of physical replicas via traditional or additive manufacturing methods.

Hermes uses state-of-the-art algorithmic techniques including a Convolutional Neural Network trained on 120 de-identified arthritic CT scans for automated segmentation, whereas the predicate ScanlIP relies on foundational image processing with bone outlines from automatic, semi-automatic, or manual sources. Hermes generates landmarks from point clouds at discrete locations rather than bone outlines, claimed to enable more accurate predictions. Both systems process DICOM data and generate output files for downstream applications.

Not stated in this summary.

Hermes and ScanlIP share identical indications for use as software interfaces and image segmentation systems for preoperative planning and physical replica fabrication in orthopedic applications. Substantial equivalence is demonstrated through DICE score validation studies comparing Hermes segmentation accuracy against two predicate tools (Synopsys and Mimics), showing Hermes accurately segments relevant knee bones from CT scans without unintended segmentations. The Segmentation Analysis and DICE Score Study confirmed Hermes met stringent design and software requirements equivalent to predicate functionality, demonstrating it performs the same intended function with similar safety and effectiveness despite using more advanced algorithmic techniques.