Search Results

CareMiBrain is dedicated brain PET scanner, and intended to obtain Position Emission Tomography (PET) inages of human brain to detect abnormal pattern of distribution of radioactivity after injection of a position emitting radiopharmaceutical. This device is to be used by trained healthcare professionals. This information can assist in research, diagnosis, therapeutic planning and therapeutic outcome assessment.

CareMiBrain is a small aperture Positron Emission Tomography (PET) scanner to image the distribution of injected positron emitting radiopharmaceuticals in the head of live humans in seating/reclined position. CareMiBrain is a PET (Positron Emission Tomography) dedicated to brain imaging. All elements of the system are integrated into a compact volume, containing the detection system, acquisition and control electronics and software. All elements of the system are integrated into a compact volume, containing the detection system, acquisition and control electronics and software. The scanner consists of 48 monolithic Lutetium OrthoSilicate (LYSO) crystals arranged in 3 rings of 16 modules each. Physical ring diameter is 260mm, with an effective 220 mm transaxial and 152 mm axial FOV. Crystal dimensions are 50x50x15mm (width x height x thickness). Crystals are coupled to a photosensor array of 12x12 silicon photo-multiplier (SiPM), 3x3 mm each. The detectors of the equipment are integrated in a circular housing with the appropriate dimensions so that the patient can insert the head. The software that integrates the equipment allows the acquisition, reconstruction and export of tomographic images of the brain, as well as to make a diagnosis of the state of the detectors.

Here's a breakdown of the acceptance criteria and the studies proving the CareMiBrain device meets them, based on the provided FDA 510(k) summary:

1. Table of Acceptance Criteria and Reported Device Performance

Note: For Image Quality - % contrast/background variability, the reported values are higher than the acceptance criteria, which suggests better performance (lower variability is generally desired for image quality, but the metric here is contrast/background variability, implying a higher value reflects better contrast relative to background). The document claims these results "comply with its predetermined specification," indicating these values met the intended performance.

2. Sample size used for the test set and the data provenance

Test Set Sample Size: For the clinical effectiveness study, "Sample images from several clinical cases with different PET tracers using the CareMiBrain PET Scanner were provided." The exact number of clinical cases is not specified from this statement, but a previous mention notes that "More than 40 clinical images are provided from CareMiBrain to demonstrate the image capability". This suggests a test set of at least 40 clinical images.

Data Provenance:

• Bench Performance Data: Independently tested by the "Institute for Instrumentation of Molecular Imaging (i3m)" according to NEMA NU 2-2012 and NEMA NU 4-2008 standards. The results were published in Scientific Reports Journal (DOI 10.1038/s41598-019-51898-z). This implies the testing was conducted in a laboratory setting.
• Clinical Effectiveness Data: Tested by "independent hospitals" and the results published in the Spanish Journal of Nuclear Medicine and Molecular Imaging (DOI 10.1016/j.remn.2021.04.002). This indicates prospective data collection from real-world clinical use.

3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts

This information is not explicitly provided in the document. While clinical effectiveness was assessed and images were provided, there's no detail on how ground truth for these clinical cases was established (e.g., through a panel of qualified radiologists, comparing with other diagnostic methods, or follow-up).

4. Adjudication method for the test set

This information is not explicitly provided in the document.

5. If a multi-reader multi-case (MRMC) comparative effectiveness study was done, If so, what was the effect size of how much human readers improve with AI vs without AI assistance

No, a multi-reader multi-case (MRMC) comparative effectiveness study was not done, or at least not described in this document. The purpose of the study was to demonstrate the device's inherent performance and clinical effectiveness for obtaining PET images, not its impact on human reader performance or AI assistance. The device is purely an imaging system, not an AI-powered diagnostic aid that assists human readers.

6. If a standalone (i.e. algorithm only without human-in-the loop performance) was done

Yes, a standalone performance assessment was done. The entire bench performance testing, evaluated against NEMA standards, represents the standalone performance of the CareMiBrain device. The clinical images provided also demonstrate the device's output without human interpretation as part of the core performance metrics. The device itself is the "algorithm" in this context, producing images for human interpretation.

7. The type of ground truth used (expert consensus, pathology, outcomes data, etc.)

• Bench Performance: Ground truth was established by adherence to recognized phantom-based testing protocols (NEMA NU 4-2008 and NEMA NU 2-2012) using phantoms with known properties.
• Clinical Effectiveness: The document states "Sample images from several clinical cases with different PET tracers...were provided." While it mentions "clinical effectiveness," the specific type of ground truth for these clinical cases (e.g., confirmed diagnosis by pathology, follow-up outcomes, expert consensus on other imaging modalities) is not explicitly detailed.

8. The sample size for the training set

The document describes the device as a PET scanner (hardware and associated software for acquisition, reconstruction, and export of images), not an AI/Machine Learning model that would typically have a separate training set. Therefore, the concept of a "training set" in the context of an AI algorithm is not applicable to this device as described. The software's development would likely involve standard software engineering verification and validation processes, not machine learning training.

9. How the ground truth for the training set was established

As the concept of a "training set" for an AI algorithm is not applicable to this device as described, this question is not relevant.

Ask a specific question about this device