HealthOST by NanoxAI Ltd. | FDA 510(k) Summary

HealthOST is an image processing software that provides qualitative and quantitative analysis of the spine from CT images to support clinicians in the evaluation and assessment of musculoskeletal disease of the spine. The HealthOST software provides the following functionality:

-Labelling of T1-L4 vertebrae

-Measurement of height loss in each vertebra (T1-L4)

-Measurement of the mean Hounsfield Units (HU) in volume of interest within vertebra (T11-L4)

HealthOST is indicated for use in patients aged 50 and over undergoing CT scan for any clinical indication, that includes at least four vertebrae in the T1-L4 portion of the spine (for vertebral height loss) and T11-L4 (for bone attenuation) portions of the spine.

The device is indicated for FBP-reconstructed images only.

Device Description

HealthOST does this by analyzing CT scans of patients aged 50 and above being performed for any clinical indication and providing the following outputs for each analyzed vertebra

1. The vertebral name, e.g. "L3"
1. The percentage of vertebral height loss, calculated as:
  Height loss = 1 - (minimum(anterior or middle) vertebral height measurement / posterior vertebral height measurement)
  from three vertebral height measurements placed at the anterior, middle and posterior aspects of the vertebral body, at the point nearest to the center of the vertebral body.

The height loss measurement is provided for every complete vertebra in the range of T1-L4. Height loss above the height loss display threshold will be indicated to the user. The height loss display threshold is configurable per installation from a predefined list, with the default set at >20% height loss. In addition, vertebrae without height loss but with significant height deviation from neighboring vertebrae will be displayed to the user.

1. The mean hounsfield unit vertebral bone attenuation, calculated by taking the mean HU from a volume of interest of the trabecular bone. This measurement is provided for every complete vertebrae in the range of T11-L4. HU bone attenuation below a set bone attenuation display threshold will be indicated to the user. The bone attenuation display threshold is configurable per installation from a predefined list.
  The following modules compose the HealthOST software:

Data input and validation: Following retrieval of a study, the validation feature assessed the input data (i.e. age, modality, view, etc.) to ensure compatibility for processing by the algorithm.

HealthOST algorithm: Once a study has been validated, the algorithm analyzes the CT for analysis and quantification.

IMA Integration feature: The study analysis and the results of a successful study analysis is provided to IMA.

Error codes feature: In the case of a study failure during data validation or the analysis by the algorithm, an error is provided to the system.

Use of axial scans for analysis is only intended as a "back up" in cases that a sagittal scan is not available in the study.

AI/ML Overview

Here's a breakdown of the acceptance criteria and study details for the HealthOST device, based on the provided text:

1. Acceptance Criteria and Reported Device Performance

Note: The document explicitly states "stated performance goal" for each metric, implying these are the acceptance criteria. The reported device performance then shows that these goals were exceeded or met.

Acceptance Criterion (Performance Goal)	Reported Device Performance
Vertebral Naming Agreement	Overall agreement of 91.49% (95% CI: [89.91%, 92.89%]), exceeding the stated performance goal.
Height Loss (95% Limits of Agreement - LOA) comparing HealthOST to GT within GT's LOA	95% LOA for HealthOST height loss of [-7.98, 7.36], within the stated performance goal. Additionally, 91.86% of the differences between HealthOST and the GT lie within the ground truthers' LOAs.
Differentiating Height Loss (>20%) Agreement	91.28% overall agreement between HealthOST and the ground truth, exceeding the stated performance goal.
Bone Attenuation (95% Limits of Agreement - LOA)	95% LOA for the bone attenuation of [-20.83, 18.29], which exceeded the stated performance goal. Additionally, 96.06% of the differences between HealthOST and the GT lie within the ground truthers' LOAs. (Note: The GT's LOA was [-21.17, 17.19])

Performance on Axial Scans (Subset Analysis):

Performance Metric (Axial Scans)	Reported Device Performance
Vertebral Naming Agreement	92.19% (95% CI [89.31% - 95.07%]), exceeding the stated performance goal.
Height Loss (differences within GT's LOA)	94.81% of the differences between HealthOST and the GT lie within the ground truthers' LOAs.
Differentiating Height Loss (>20%) Agreement	Overall agreement of 91.89% between HealthOST and ground truth.
Bone Attenuation (95% Limits of Agreement - LOA)	[-15.78, 12.21].
Bone Attenuation (differences within GT's LOA)	96.06% of the differences between HealthOST and the GT lie within the ground truthers' LOAs.

2. Sample Size and Data Provenance for the Test Set

Sample Size for Test Set: 150 anonymized CT scans, including a minimum of four vertebrae between T1 to L4. This dataset included 1425 vertebrae in total.
Data Provenance: Retrospective study. The data was sourced from two healthcare institutions, composed of multiple clinical sites, in the US and OUS (Outside US). A secondary testing dataset from an additional, representative US data source was also used to demonstrate generalizability to US populations.

3. Number of Experts and Qualifications for Ground Truth

Number of Experts: Three.
Qualifications of Experts: US board-certified radiologists.

4. Adjudication Method for the Test Set

The document does not explicitly state an adjudication method like "2+1" or "3+1". It states that "Ground truth measurements were determined by the three US board-certified radiologists." This implies either consensus was reached between them, or a majority vote, or potentially a lead radiologist made the final determination after discussion. Without further detail, it's not possible to specify an exact adjudication method.

5. Multi Reader Multi Case (MRMC) Comparative Effectiveness Study

No, a multi-reader multi-case (MRMC) comparative effectiveness study was not explicitly mentioned in the provided text as a study done to assess human reader improvement with AI assistance. The study described is a standalone performance evaluation of the HealthOST device compared to ground truth and its predicate device.

6. Standalone Performance Study

Yes, a standalone (algorithm only, without human-in-the-loop performance) study was done. The document states: "The HealthOST device performance was evaluated in a stand-alone retrospective study of its performance compared to the established ground truth and respective to the predicate device."

7. Type of Ground Truth Used

The ground truth used was expert consensus / expert measurement. It states, "Ground truth measurements were determined by the three US board-certified radiologists."

8. Sample Size for the Training Set

The document does not specify the sample size used for the training set. It only details the test set.

9. How the Ground Truth for the Training Set Was Established

The document does not describe how the ground truth for the training set was established. It only focuses on the ground truth establishment for the validation (test) dataset.

Summary

Regulation Number and Section

§ 892.1750 Computed tomography x-ray system.

(a)
Identification. A computed tomography x-ray system is a diagnostic x-ray system intended to produce cross-sectional images of the body by computer reconstruction of x-ray transmission data from the same axial plane taken at different angles. This generic type of device may include signal analysis and display equipment, patient and equipment supports, component parts, and accessories.(b)
Classification. Class II.