ScanDiags Ortho L-Spine MR-Q

ScanDiags Ortho L-Spine MR-Q is a software tool that processes previously-acquired DICOM lumbar spine MRI images to automatically segment and measure anatomical structures (vertebral bodies, intervertebral discs, neuroforamina, thecal sac) using deep learning. It generates quantitative measurements and PDF reports for radiologist review and approval, intended for use in hospitals and medical institutions on patients aged 22 and above.

Both subject and predicate devices are Class II medical image management and processing systems using MRI modality for lumbar spine analysis. The subject device uses supervised deep convolutional neural networks (DCNN) for both classification and segmentation, while the predicate uses deep convolutional image-to-image neural networks. Both provide semi-automatic segmentation, area/distance measurements, and PDF reports requiring human radiologist interpretation; the subject device measures slightly more parameters (spinal canal area, angle, biconcave height losses) but follows the same operational principles.

Not stated in this summary.

The subject device is substantially equivalent because it shares the same intended use (quantitative measurement of lumbar spine structures from MRI), same regulatory classification and product code (21 CFR 892.2050, QIH, Class II), and operates on the same principles as the predicate. Performance validation across 100 patient studies from multiple MRI manufacturers demonstrated acceptable measurement accuracy (ICC 0.74–0.95, Dice scores 0.86–0.95, MAE <1.3 mm) comparable to inter-radiologist agreement, with subgroup analyses confirming consistent performance across demographics and imaging parameters. The technological differences (refined neural network architecture, expanded measurement set) do not raise new safety or efficacy concerns since both require radiologist review and neither produces diagnostic recommendations.