(167 days)
The Definium Tempo Select is intended to generate digital radiographic images of the skull, spinal column, chest, abdomen, extremities, and other body parts in patients of all ages. Applications can be performed with the patient sitting, standing, or lying in the prone or supine position and the system is intended for use in all routine radiography exams. Optional image pasting function enables the operator to stitch sequentially acquired radiographs into a single image.
This device is not intended for mammographic applications.
The Definium Tempo Select Radiography X-ray System is designed as a modular system with components that include an Overhead Tube Suspension (OTS) with a tube, an auto collimator and a depth camera, an elevating table, a motorized wall stand, a cabinet with X-ray high voltage generator, a wireless access point and wireless detectors in exam room and PC, monitor and control box with hand-switch in control room. The system generates diagnostic radiographic images which can be reviewed or managed locally and sent through a DICOM network for applications including reviewing, storage and printing.
By leveraging platform components/ design, Definium Tempo Select is similar to the predicate device Discovery XR656 HD (K191699) and the reference device Definium Pace Select (K231892) with regards to the user interface layout, patient worklist refresh and selection, protocol selection, image acquisition, and image processing based on the raw image. This product introduces a new high voltage generator which has the same key specifications as the predicate. A wireless detector used in referenced product Definium Pace Select is introduced. Image Pasting is improved with individual exposure parameter adjustable on images on both Table and Wall Stand Mode. Tube auto angulation is added for better auto positioning based on current auto-positioning. Camera Workflow is introduced based on existing depth camera. OTS is changed with 4 axis motorizations. An update was made to the previously cleared Tissue Equalization feature under K013481 to introduce a Deep Learning AI model that provides more consistent image presentations to the user which reduces additional workflow to adjust the image display parameters. The other minor changes including PC change, Wall Stand change and Table change.
The provided FDA 510(k) clearance letter and summary for the Definium Tempo Select offers some, but not all, of the requested information regarding the acceptance criteria and the study proving the device meets them. Notably, specific quantitative acceptance criteria for the AI Tissue Equalization feature are not explicitly stated.
Here's a breakdown of the available information and the identified gaps:
1. Table of Acceptance Criteria and Reported Device Performance
Note: The 510(k) summary does not explicitly list quantitative acceptance criteria for the AI Tissue Equalization algorithm. Instead, it states that "The verification tests confirmed that the algorithm meets the performance criteria, and the safety and efficacy of the device has not been affected." Without specific performance metrics or thresholds, a direct comparison in a table format is not possible for the AI component.
For the overall device, the acceptance criteria are implicitly performance metrics that ensure it functions comparably to the predicate device, as indicated by the "Equivalent" and "Identical" discussions in Table 1 (pages 7-11). However, these are primarily functional and technical equivalency statements rather than performance metrics for the AI feature.
Therefore, this section will focus on the AI Tissue Equalization feature as it's the part that underwent specific verification using a clinical image dataset.
AI Tissue Equalization Feature:
| Acceptance Criteria (Implied) | Reported Device Performance |
|---|---|
| Provides more consistent image presentations to the user. | "The verification tests confirmed that the algorithm meets the performance criteria, and the safety and efficacy of the device has not been affected." |
| "The image processing algorithm uses artificial intelligence to dynamically estimate thick and thin regions to improve contrast and visibility in over-penetrated and under-penetrated regions." | |
| "The algorithm is the same but parameters per anatomy/view are determined by artificial intelligence to provide better consistence and easier user interface in the proposed device." | |
| Reduces additional workflow to adjust image display parameters. | Achieved (stated as a benefit of the AI model). |
| Safety and efficacy are not affected. | Confirmed through verification tests. |
Missing Information:
- Specific quantitative metrics (e.g., AUC, sensitivity, specificity, image quality scores, expert rating differences) that define "more consistent image presentations" are not provided.
- The exact thresholds or target values for these metrics are not stated.
2. Sample Size Used for the Test Set and Data Provenance
- Test Set Sample Size: Not explicitly stated as a number of images or cases. The document refers to "clinical images retrospectively collected across various anatomies...and Patient Sizes."
- Data Provenance: Retrospective collection from locations in the US, Europe, and Asia.
3. Number of Experts Used to Establish the Ground Truth for the Test Set and Their Qualifications
Missing Information. The document does not specify:
- The number of experts involved in establishing ground truth.
- Their qualifications (e.g., specific subspecialty, years of experience, board certification).
- Whether experts were even used to establish ground truth for this verification dataset, as the purpose was to confirm the AI met performance criteria rather than to directly compare its diagnostic accuracy against human readers or a different ground truth standard.
4. Adjudication Method for the Test Set
Missing Information. No adjudication method (e.g., 2+1, 3+1) is described for the test set.
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 explicitly mentioned or described in the provided document. The verification tests focused on the algorithm meeting performance criteria, not on comparing human reader performance with or without AI assistance.
- Effect Size: Not applicable, as no MRMC study was described.
6. If a Standalone Study (Algorithm Only Without Human-in-the-Loop Performance) Was Done
Yes, implicitly. The "AI Tissue Equalization algorithms verification dataset" was used to perform "verification tests" to confirm that "the algorithm meets the performance criteria, and the safety and efficacy of the device has not been affected." This suggests a standalone evaluation of the algorithm's output (image presentation consistency) against specific, albeit unstated, criteria. While human review of the output images was likely involved, the study's stated purpose was to verify the algorithm itself.
7. The Type of Ground Truth Used (Expert Consensus, Pathology, Outcomes Data, etc.)
Implied through image processing improvement, not diagnostic ground truth. For the AI Tissue Equalization feature, the "ground truth" is not in the traditional clinical diagnostic sense (e.g., disease presence confirmed by pathology). Instead, it appears to be related to the goal of "more consistent image presentations" and improving "contrast and visibility in over-penetrated and under-penetrated regions." This suggests the ground truth was an ideal or desired image presentation quality rather than a disease state. It's likely based on existing best practices for image processing and subjective assessment of image quality by experts, or perhaps a comparative assessment against the predicate's tissue equalization.
Missing Information: The precise method or criteria for this ground truth (e.g., a panel of radiologists rating image quality, a quantitative metric for contrast/visibility) is not specified.
8. The Sample Size for the Training Set
Missing Information. The document describes the "verification dataset" (test set) but does not provide any information on the sample size or composition of the training set used to develop the Deep Learning AI model for Tissue Equalization.
9. How the Ground Truth for the Training Set Was Established
Missing Information. As the training set size and composition are not mentioned, neither is the method for establishing its ground truth. It can be inferred that the training process involved data labeled or optimized to achieve "more consistent image presentations" by dynamically estimating thick and thin regions, likely through expert-guided optimization or predefined image processing targets.
§ 892.1680 Stationary x-ray system.
(a)
Identification. A stationary x-ray system is a permanently installed diagnostic system intended to generate and control x-rays for examination of various anatomical regions. This generic type of device may include signal analysis and display equipment, patient and equipment supports, component parts, and accessories.(b)
Classification. Class II (special controls). A radiographic contrast tray or radiology diagnostic kit intended for use with a stationary x-ray system only is exempt from the premarket notification procedures in subpart E of part 807 of this chapter subject to the limitations in § 892.9.