Search Results

Green X (Model : PHT-75CHS) is intended to produce panoramic, cephalometric or 3D digital x-ray images. It provides diagnostic details of the dento-maxillofacial, ENT, sinus and TMJ for adult and pediatric patients. The system also utilizes carpal images for orthodontic treatment The device is to be operated by healthcare professionals.

Green X (Model : PHT-75CHS) is an advanced 4-in-1 digital X-ray imaging system that incorporates PANO, CEPH(optional), CBCT and MODEL Scan imaging capabilities into a single system. Green X (Model : PHT-75CHS), a digital radiographic imaging system, acquires and processes multi-FOV diagnostic images for dentists. Designed explicitly for dental radiography. Green X is a complete digital X-ray system equipped with imaging viewers, an X-ray generator and a dedicated SSXI detector.

The digital CBCT system is based on a CMOS digital X-ray detector. The CMOS CT detector is used to capture 3D radiographic images of the head, neck, oral surgery, implant and orthodontic treatment. Green X (Model : PHT-75CHS) can also acquire 2D diagnostic image data in conventional PANO and CEPH modes.

The provided text describes the Green X (Model: PHT-75CHS) dental X-ray imaging system and its substantial equivalence to a predicate device. However, it does not contain detailed information about a study proving the device meets acceptance criteria for an AI feature with specific performance metrics such as sensitivity, specificity, or AUC calculated on a test set, nor does it describe an MRMC study.

The document discusses improvements and additions to the device, including "Endo mode," "Double Scan function," "Insight PAN 2.0," and the availability of FDK and CS reconstruction algorithms. It mentions some quantitative evaluations for these features, primarily focusing on image quality metrics and stitching accuracy, but not clinical performance metrics typical for AI algorithms (e.g., detection of specific pathologies).

Based on the provided text, here's an attempt to answer the questions, highlighting where information is missing for AI-specific criteria:

Acceptance Criteria and Device Performance (Based on available information):

1. Sample size used for the test set and the data provenance:

• Test Set Sample Size: Not explicitly stated for any of the described evaluations (Endo mode, Double Scan, Insight PAN 2.0, FDK/CS algorithms, or general image quality). The evaluations seem to be based on a limited number of clinical images/test phantoms rather than large-scale patient datasets.
• Data Provenance: Not specified. It indicates "clinical images generated in Endo mode" and "3D clinical consideration" for Double Scan, and "clinical evaluation" for Insight PAN 2.0. There is no mention of country of origin or whether the data was retrospective or prospective.

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

• Endo Mode: "A US licensed dentist" evaluated the clinical images. The number of dentists is not specified (could be one or multiple). No specific years of experience or sub-specialty are mentioned beyond "US licensed dentist."
• Double Scan Function: "3D clinical consideration and evaluation" was performed. No specific number or qualifications of experts are mentioned.
• Insight PAN 2.0: "Clinical evaluation was performed." No specific number or qualifications of experts are mentioned.
• Other evaluations: The document refers to "satisfying standard criteria" (IEC 61223-3-5, IEC 61223-3-4) and measurements on phantoms, which typically do not involve expert ground truth in the same way clinical AI performance studies do.

3. Adjudication method (e.g. 2+1, 3+1, none) for the test set:

• None specified. The evaluations appear to involve a single "US licensed dentist" for Endo mode, and "clinical evaluation" without detailing the adjudication process for other features. This is not a typical AI performance study setup where multiple readers independently review and a consensus process might be employed.

4. If a multi-reader multi-case (MRMC) comparative effectiveness study was done:

• No, an MRMC comparative effectiveness study was not done. The document describes performance evaluations of the device's features (e.g., image quality, stitching accuracy, clinical utility) but not a comparative study where human readers' performance with and without AI assistance is measured. Thus, no effect size for human improvement is reported.

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

• This is not explicitly an AI-only device where the "algorithm" performs diagnostic tasks autonomously. The features described (Endo mode, Double Scan, Insight PAN 2.0, reconstruction algorithms) are functionalities of an X-ray imaging system that produce images for human interpretation. The "evaluations" described are largely for image quality metrics and technical performance, not for algorithmic detection or classification of disease. Therefore, a standalone performance study in the context of an AI diagnostic aid is not applicable in the way it might be for, say, an algorithm that flags lesions.

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

• Primary Ground Truth:
  • Phantom Measurements: For quantitative image quality metrics (Noise, Contrast, CNR, MTF, line pair resolution, low contrast resolution) according to IEC standards.
  • Calculated Metrics: For stitching accuracy (SSIM, RMSE).
  • Clinical Evaluation: For confirming "diagnostic quality" (Endo mode) and "clinical efficacy" (Double Scan, Insight PAN 2.0), which relies on expert judgement of the generated images, rather than independent pathology or outcomes data. It functions more as a qualitative assessment of the image's utility.

7. The sample size for the training set:

• Not applicable/Not provided. The document describes a traditional X-ray imaging system with new features, some of which might involve algorithms (e.g., stitching algorithm, reconstruction algorithms) but doesn't explicitly state that these features are "AI" in the sense of requiring a large, labeled training dataset of images to learn to perform a diagnostic task. If these features involve machine learning (e.g., for image enhancement or reconstruction), the training data for those specific algorithms is not detailed.

8. How the ground truth for the training set was established:

• Not applicable/Not provided for the reasons stated above.

Summary of the Device and Evaluation Context:

The FDA 510(k) clearance process for the Green X (Model: PHT-75CHS) system focuses on demonstrating substantial equivalence to a predicate device. The performance evaluations described are primarily related to the physical and technical performance of the X-ray imaging system and its new functionalities (Endo mode, Double Scan, Insight PAN 2.0, FDK/CS algorithms). These evaluations confirm that the device produces images of sufficient quality, that spatial and contrast resolutions meet standards, and that new features like image stitching are accurate.

Crucially, this is not a submission for an AI/ML-driven diagnostic medical device that would typically involve large, diverse test sets, multiple expert readers, detailed ground truth establishment (like pathology or clinical outcomes), and comparative effectiveness studies to measure how much AI improves human reader performance for a specific diagnostic task (e.g., detecting a particular disease from the image). The "performance data" provided relates to the image acquisition capabilities and processing algorithms of the imaging system itself, which are fundamental to any diagnostic interpretation by a human professional rather than an algorithmic diagnosis or detection.

Ask a specific question about this device