HealthOST

HealthOST is an image processing software providing qualitative and quantitative analysis of the spine from CT images to support clinicians in assessing musculoskeletal diseases in patients aged 50 and older. It employs an algorithm on FBP-reconstructed CTs to label vertebrae (T1-L4), measure vertebral height loss, and quantify mean Hounsfield Units (T11-L4) for bone attenuation assessment.

HealthOST is an image processing software designed for the qualitative and quantitative analysis of the spine from Computed Tomography (CT) images, aimed at supporting clinicians in evaluating and assessing musculoskeletal diseases of the spine. Indicated for patients aged 50 and over undergoing CT scans for any clinical reason, the software specifically analyzes FBP-reconstructed images and provides functionalities such as labeling vertebrae from T1 to L4, measuring height loss in each vertebra within this range, and determining the mean Hounsfield Units (HU) as a measure of bone attenuation within a volume of interest in vertebrae T11 to L4, requiring analysis of at least four vertebrae in the specified spinal portions for these respective measurements. The software operates by first validating input data (e.g., age, modality, view) from retrieved CT studies. Its core algorithm then analyzes the validated CT scans for quantification and analysis, with results integrated into IMA systems; error codes are generated for study failures, and axial scans can be used as a backup if sagittal scans are unavailable. The device's performance was validated in a retrospective study using 150 anonymized CT scans (including 1425 vertebrae) from multiple sites, compared against ground truth established by three U.S. board-certified radiologists and a predicate device. Key results demonstrated a 91.49% overall agreement for vertebral naming, limits of agreement (LOA) for height loss of [-7.98, 7.36], and LOA for bone attenuation of [-20.83, 18.29], with performance also validated for axial scans and shown to be generalizable to U.S. populations.

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 HealthOST's performance in measuring vertebral height loss and bone attenuation correlate with specific stages or severities of common musculoskeletal diseases of the spine, such as osteoporosis or vertebral compression fractures?; What is the variability in Hounsfield Unit measurements by HealthOST across different CT scanner manufacturers, models, or acquisition protocols, beyond the specifically indicated FBP-reconstruction?; Beyond the configurable display thresholds for height loss and bone attenuation, can the HealthOST software be adapted to flag other complex patterns or combinations of findings indicative of early-stage spinal pathologies?; How does the software's feature for identifying 'significant height deviation from neighboring vertebrae' differentiate between pathological changes and normal anatomical variations, and what is its diagnostic accuracy for subtle abnormalities?; What are the typical computational requirements and processing times for HealthOST analysis on clinical CT spine datasets, and how does its IMA integration impact existing radiological workflows?