(158 days)
The e-Lung software provides reproducible CT values for pulmonary tissue, which is essential for providing quantitative support in the examination of radiological findings. These radiological findings can then be evaluated by the physician in conjunction with a range of ancillary information to form a potential diagnosis or list of likely diagnoses. The e-Lung software package is intended to be a workflow enhancement and for the assessment of CT thoracic datasets. e-Lung can be used to support the physician when examining the pulmonary and thoracic tissue (i.e. lung parenchyma) in CT thoracic datasets. 3D segmentation, volumetric measurements, density evaluations, and reporting tools are combined with a dedicated workflow.
Brainomix 360 e-Lung is a software package compliant with the DICOM standard and running on an off-the-shelf physical or virtual server. e-Lung is a CT processing module which operates within the integrated Brainomix 360 platform.
Brainomix 360 e-Lung is a stand-alone software device which uses a set of image processing algorithms to perform evaluation (3D segmentation and isolation of sub-compartments, volumetric measurements, and density evaluations), editing, and reporting tools which are combined with a dedicated workflow.
e-Lung can be used to support the physician in the examination of radiological findings that may be indicative of chest diseases e.g. when examining the pulmonary and thoracic tissue (i.e. lung parenchyma) in CT thoracic datasets. These radiological findings can then be evaluated by the physician in conjunction with a range of ancillary information to form a potential diagnosis or list of likely diagnoses.
e-Lung is designed to analyze pulmonary CT slice data and display analysis results. Each voxel of the scan is measured by Hounsfield units (HU), a measurement of x-ray attenuation that is applied to each volume element in three dimensional space. The HU are utilized to distinguish between air, water, tissue and bone, such distinction is common in the industry.
e-Lung provides computed tomography (CT) viewing, and parenchymal density analysis in one application. e-Lung provides quantitative measurements and tabulates quantitative properties.
e-Lung focuses on what is visible to the eye and applies volumetric methods that might otherwise be too time consuming to use.
The software does not perform any function which cannot be accomplished by a trained user utilizing manual tracing methods; the software does not reconstruct a 3D rendering image of the lung; the intent of the software is to enhance the workflow by saving time and automating potential error prone manual tasks.
e-Lung has functions for loading, and saving datasets, and will generate screen displays, computations and aggregate statistics. e-Lung data output may be exported to a CSV, Excel or PDF file.
Here is an analysis of the acceptance criteria and study for the Brainomix 360 e-Lung device, based on the provided text:
1. Table of Acceptance Criteria and Reported Device Performance
| Feature | Acceptance Criteria | Reported Device Performance |
|---|---|---|
| Lung Segmentation | Average Dice Similarity Coefficient (DSC) across all cases of over 0.95 (defined by the lower bound of the confidence interval) | Median DSC: 0.978 (IQR: 0.974-0.980); all cases showed a DSC above 0.95. |
| Density Evaluation | Good Dice score (min 0.80) between the e-Lung structural densities and histogram densities and those pre-defined parameters generated in the digital phantom dataset. | "The density evaluations are validated by ensuring a good Dice score (min 0.80) between the e-Lung structural densities and histogram densities and those pre-defined parameters generated in the digital phantom dataset." (Implicitly, this means the criterion was met.) |
2. Sample Sizes Used for the Test Set and Data Provenance
-
Lung Segmentation Study:
- Test Set Sample Size: 100 cases (N=38 from Boston Medical Center, N=62 from a commercial database).
- Data Provenance: Retrospective study. Cases were selected from a research registry at Boston Medical Center and a commercial database of clinical imaging data. The demographic and clinical variables were enriched to allow generalizability. The hospital locations mentioned for the cases are Massachusetts, New York, Ohio, New Jersey, Wisconsin, Florida, Maryland, and South Dakota (all in the USA), indicating the data origin is the USA.
-
Density Evaluation Study:
- Test Set Sample Size: Not explicitly stated, but involved "synthetic digital phantom data" and a "real-world data bridging study."
3. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications
- Lung Segmentation Study:
- Number of Experts: Three (3)
- Qualifications: Experienced US board certified radiologists.
4. Adjudication Method for the Test Set
- Lung Segmentation Study: The ground truth mask was generated from the consensus of the three experienced US board certified radiologists. This implies an adjudication method where agreement among the experts was used to define the ground truth. The specific "2+1" or "3+1" approach is not explicitly detailed, but "consensus" indicates multiple readers informing the final ground truth.
5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study was Done
- No, a multi-reader multi-case (MRMC) comparative effectiveness study was not done. The study evaluated the device's accuracy against a ground truth created by human experts, not the improvement of human readers with AI assistance versus without AI assistance.
6. If a Standalone Study (Algorithm Only Without Human-in-the-Loop Performance) was Done
- Yes, the lung segmentation study described is a standalone study. The device's lung mask generation was compared directly to the expert-derived ground truth, without human interaction with the device's output during the evaluation phase.
7. The Type of Ground Truth Used
- Lung Segmentation Study: Expert consensus, specifically from three experienced US board certified radiologists who segmented the lungs.
- Density Evaluation Study: Synthetic digital phantom data and a real-world data bridging study to establish pre-defined parameters.
8. The Sample Size for the Training Set
- The document does not explicitly state the sample size for the training set. It only describes the validation set (test set).
9. How the Ground Truth for the Training Set was Established
- The document does not explicitly describe how the ground truth for the training set was established. It focuses on the validation (test) set. Given the algorithm is described as "non-adaptive deterministic," it's possible that a formal "training set" with ground truth in the machine learning sense wasn't used for an adaptive algorithm, but rather a set of cases for algorithm development and refinement, which would typically involve expert annotations for ground truth. However, the document does not provide these details.
§ 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.