Search Results

NS-HGlio is intended for the semi-automatic labeling, visualization, and volumetric quantification of high-grade brain glioma (WHO grade 3 astrocytoma, WHO grade 4 astrocytoma and WHO grade 4 glioblastoma) from a set of standard MRI images of male or female patients 18 years of age or older who are known to have pathologically proven high-grade glioma. Volumetric measurements may be compared to past measurements if available. NS-HGlio is not to be used for primary diagnosis, and is intended to be used by qualified clinical personnel as an additional source of information and is not intended to be the sole diagnostic metric.

NS-HGlio is a non-invasive software as a medical device (SaMD) tool intended for labeling, visualization, and volumetric quantification of high-grade brain gliomas for a population that has been pathologically diagnosed to have brain tumors. The device is used as a tool by clinicians in determining the patient's disease conditions on pre- and post-operative MRI images. The device is not used for primary diagnosis. NS-HGlio device takes as an input imported Digital Imaging and Communications in Medicine (DICOM) images of high-grade brain glioma acquired with standard brain tumor MRI protocols and uses a deep learning methodology to semi-automatically label the different subcomponents of the high-grade glioma. Results are displayed on a Neosoma viewing software. Optionally, the software connects to clinicians' applications (e.g., PACS).

Here's a breakdown of the acceptance criteria and study details for NS-HGlio, based on the provided text:

1. Table of Acceptance Criteria & Reported Device Performance

Rationale for Implied Acceptance Criteria: The document states that the device's DSC "is higher than the mean DSC of the average of the three experts for the same task." This implicitly sets the expert performance as the acceptance threshold for DSC. For ICC, while no specific numerical threshold is given, the very high reported value (0.98) indicates that the device met an implicit high-correlation standard, crucial for volumetric quantification.

2. Sample Size and Data Provenance

• Test Set Sample Size: 33 subjects, comprising 132 MRIs.
• Data Provenance: The test dataset was acquired from medical sites that were not included in the training dataset, to ensure device generalizability. The data consisted of both males and females, aged 18 to 79, covering diverse ethnic backgrounds, reflecting real-world prevalence of high-grade glioma. The country of origin is not explicitly stated, but the mention of "US board certified neuroradiologists" suggests the study and data collection were primarily U.S.-based.
• Retrospective/Prospective: Not explicitly stated, but given the use of "acquired from medical sites" and "real world prevalence," it's highly likely to be retrospective data collection from existing medical records.

3. Experts for Ground Truth

• Number of Experts: Three (3)
• Qualifications: US board certified neuroradiologists with expertise in measuring high-grade gliomas. The text does not specify their years of experience.

4. Adjudication Method for Test Set

The text states that the "reference standard (ground truth) was established using three US board certified neuroradiologists." It then refers to the "mean DSC of the average of the three experts." This suggests that the ground truth was established by consensus among the three experts, likely through a process where their individual annotations were combined or averaged to form the final ground truth. An explicit "adjudication method" like 2+1 or 3+1 is not detailed, but consensus implies adjudication.

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

• Was it done?: No, an MRMC comparative effectiveness study explicitly showing how much human readers improve with AI vs. without AI assistance was not conducted or reported in this document. The study focuses on evaluating the standalone performance of the AI relative to expert consensus for ground truth.

6. Standalone (Algorithm Only) Performance

• Was it done?: Yes, the study evaluates the standalone performance of the NS-HGlio algorithm. The reported Dice Similarity Coefficient (DSC) and Intraclass Correlation Coefficient (ICC) are measures of the algorithm's performance in segmenting and quantifying gliomas without direct human intervention in the segmentation process itself. The comparison to the "mean DSC of the average of the three experts" further emphasizes this standalone evaluation.

7. Type of Ground Truth Used

• Type: Expert Consensus. The "reference standard (ground truth) was established using three US board certified neuroradiologists."

8. Sample Size for the Training Set

• Training Set Sample Size: Not explicitly provided in the furnished text. The text only states that the "test dataset was acquired from medical sites that were not included in the training dataset."

9. How Ground Truth for Training Set was Established

• How Established: This information is not provided in the furnished text.

Ask a specific question about this device