Search Results

The HealthFLD device is an image processing software that provides quantitative and qualitative analysis of the liver from CT images to support clinicians in the evaluation and assessment of Fatty Liver. The HealthFLD software provides measurements of liver attenuation (mean HU in a region of interest). HealthFLD is indicated for use in non-contrast and contrast CT scans, with any clinical indication, for patients aged 18 up to 75. CTs must include a significant part of the liver. The HealthFLD device is not intended to provide a diagnosis or risk assessment of fatty liver disease.

Device Description

The HealthFLD software provides measurements of liver attenuation (mean HU in a region of interest) for any compatible CT scan that includes a significant part of the liver

The Liver measurement display threshold is <40 HU for non-contrast/non portal venous phase CTs. When portal venous contrast phase is identified by the algorithm, the HealthFLD device automatically adjusts the display threshold to <75 HU.

The following modules compose the HealthFLD software:

1. Data input and validation: DICOM validation receives imaging study from hosting application and the validation feature assessed the input data (i.e. age, modality, view, etc.) to ensure compatibility for processing by the algorithm.
1. HealthFLD algorithm: Once a study has been validated, the algorithm analyzes the CT for analysis and quantification.
1. IMA Integration feature: The results of a successful study analysis is provided to the hosting application.
1. 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.

AI/ML Overview

Here's a breakdown of the acceptance criteria and the study proving the HealthFLD device meets them, based on the provided text:

1. Table of Acceptance Criteria and Reported Device Performance

Acceptance Criteria Category	Acceptance Criteria	Reported Device Performance
Binary Classification Agreement	Agreement with ground truth liver score binary classification of < 40HU	95.98% (95% CI: [92.77%, 97.8%])
Binary Classification Agreement	Agreement with ground truth liver score binary classification of < 50HU versus ≥ 50HU	98.39% (95% CI: [95.94%, 99.37%])
Bland-Altman 95% Limits of Agreement (LOAs) for Bias	Within an acceptance interval of [-10HU, 10HU]	[-7.80HU, 7.00HU]
Percentage of Differences within LOA	Not explicitly stated as a separate acceptance criterion, but mentioned in relation to substantial equivalence to the predicate.	94.78% (95% CI:[91.24%-97.19%])
Algorithm Yield	Not explicitly stated as an acceptance criterion, but reported as an outcome.	100% (250 out of 250 cases)
Portal Venous Phase Identification Agreement	Not explicitly stated, but reported as an outcome.	95.98%

Note: The document states that the observed performances for binary classification "both exceed the stated performance goal," indicating these were indeed acceptance criteria. The Bland-Altman LOAs are stated to "lie within the acceptance interval of [-10HU,10HU]".

2. Sample Size Used for the Test Set and Data Provenance

Sample Size: 250 cases.
Data Provenance:
- Country of Origin: 62.25% (155) of CTs were from U.S. data. The remaining data's country of origin is not specified, but it's stated the data came from "4 healthcare institutions."
- Retrospective/Prospective: Retrospective study.

3. Number of Experts Used to Establish the Ground Truth for the Test Set and Their Qualifications

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

4. Adjudication Method for the Test Set

The document does not explicitly state the adjudication method (e.g., 2+1, 3+1). It only mentions that "Ground truth measurements were determined by three US board-certified radiologists." This often implies a consensus or majority vote among the radiologists, but the exact process for resolving discrepancies (if any) is not detailed.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study Was Done

No, an MRMC comparative effectiveness study was not performed to assess how much human readers improve with AI vs. without AI assistance. The study focuses on the standalone performance of the HealthFLD device against ground truth.

6. If a Standalone (Algorithm Only Without Human-in-the-Loop Performance) Was Done

Yes, a standalone retrospective study of the device's performance was done. The document states: "The HealthFLD device performance was evaluated in a stand-alone retrospective study of its performance compared to the established ground truth..."

7. The Type of Ground Truth Used

Expert Consensus: The ground truth measurements were "determined by three US board-certified radiologists."

8. The Sample Size for the Training Set

The document does not provide the sample size for the training set. It only mentions the validation data-set of 250 cases.

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

The document does not provide information on how the ground truth for the training set was established. It only describes the establishment of ground truth for the validation data-set.

Ask a Question

Ask a specific question about this device

Page 1 of 1