MSKai

MSKai is software that processes T2-weighted lumbar spine MRI images to assist radiologists and spine surgeons in identifying, segmenting, measuring, and reporting on anatomical structures. It provides qualitative viewing and quantitative measurements of the lumbar spine but does not make diagnoses; users must review and confirm all software-generated measurements and reports based on their medical training.

MSKai uses a Mask Region-based Convolutional Neural Network algorithm, whereas the predicate CoLumbo uses a Deep Convolutional Image-to-Image Neural Network. Both are software-as-a-medical-device (SaMD) that perform segmentation, measurement, and reporting on MR images. MSKai supports T2-weighted MRI from lumbar spine exams; the predicate also analyzes lumbar spine but uses different neural network architecture.

IEC 62304:2006/AMD 1:2015 (software life cycle), ISO 14971:2019 (risk management), IEC 62366-1:2015+AMD1:2020 (usability engineering), ISO 15223-1:2016 (medical device symbols), and NEMA PS 3.1–3.20 (DICOM digital imaging standard).

MSKai shares the same intended users (radiologists, neurosurgeons, ortho/spine surgeons), same body part (lumbar spine), and same functional outputs (segmentation, measurement, reporting) as the predicate CoLumbo. Although the neural network architecture differs slightly, both employ convolutional neural networks for similar purposes. The standalone performance study of 238 independent patient MRI exams demonstrated segmentation accuracy (Mean Dice Coefficients 0.91–0.99) and measurement accuracy within prospectively defined error limits across all scanner manufacturers and demographic subgroups, proving comparable safety and effectiveness to the predicate.