Search Results

The CT 5300 is a Computed Tomography X-Ray System intended to produce images of the head and body by computer reconstruction of x-ray transmission data taken at different angles and planes. These devices may include signal analysis and display equipment, patient and equipments and accessories. The CT 5300 is indicated for head, whole body, cardiac and vascular X-ray Computed Tomography applications in patients of all ages.

These scanners are intended to be used for diagnostic imaging and for low dose CT lung cancer screening for the early detection of lung nodules that may represent cancer*. The screening must be performed within the established inclusion criteria of programs / protocols that have been approved and published by etther a governmental body or professional medical society.

• Please refer to clinical literature, including the results of the National Lung Screening Trial (N Engl J Med 2011; 365:395-409) and subsequent literature, for further information.

The proposed device is a whole-body computed tomography (CT) X-Ray System featuring a continuously rotating x-ray tube, detectors, and gantry with multi-slice capability. The acquired x-ray transmission data is reconstructed by computer into cross-sectional images of the body taken at different angles and planes. This system also includes signal analysis and display equipment, patient and equipment supports, components, and accessories. The CT 5300 has a 72 cm bore and includes a detector array that provides 50 cm scan field of view (FOV). The main components (detection system, the reconstruction algorithm, and the x-ray system) that are used in the proposed device have the same fundamental design characteristics and are based on comparable technologies as the current marketed predicate Philips Incisive CT K212441(April 27, 2022).

The key system modules and functionalities are:

1. Gantry
   The Gantry consists of 4 main internal units:
   a. X-Ray Tube – produces X-rays necessary for scanning.
   b. High voltage generator - produces high voltage power supply to X-ray tube, consists of system Interface Unit, Power Block Unit and Anode Drive Unit.
   c. A-plane: adjusts the slice thickness during axial scan and monitor the changes of X-ray.
   d. DMS (Data Management System) – absorbs X-ray radiation by detectors and converts it to digital readout.
2. Patient Table (Couch)
   The Couch is used to position the patient. Carries the patient in and out through the Gantry bore synchronized with the scan.
3. Console
   The console is used to operate the system and monitor the scan. The Operator console includes a computer, monitors and CTBOX.
4. CT on Trailer Kit
   The CT 5300 installed and secured on a trailer requires locking motion parts during trailer transportation and unlocking motion parts before CT operations. Besides being installed in hospital, the CT may also be installed on trailer to be transported to designated locations for use within a professional healthcare environment.

The CT 5300 on Trailer Kit has the same fundamental design characteristics and technologies as the current marketed Philips Incisive CT on trailer (K211168 - November 22, 2021). The CT on Trailer configuration is identical to the K211168 trailer configuration. The CT system should only be used in designated locations for use with appropriate radiation controls and safety measures.

In addition to the above components and the software operating them, each system includes hardware and software for data acquisition, display, manipulation, storage and filming as well as post-processing into views other than the original axial images.

Upgrades Kit is available to upgrade earlier Incisive CT installations to latest version.

The provided document is a 510(k) Premarket Notification from Philips Healthcare (Suzhou) Co., Ltd. for their CT 5300 Computed Tomography X-Ray System. The purpose of this document is to demonstrate "Substantial Equivalence" to a legally marketed predicate device (Philips Incisive CT, K212441) and thus does not include acceptance criteria or detailed study results for standalone AI/ML device performance.

Instead, the document focuses on showing that the new CT 5300 system (which incorporates an updated version of the "Precise Image" and "Precise Position" algorithms) is as safe and effective as the predicate device.

However, based on the limited information provided regarding the "Precise Image" algorithm update, I can attempt to extract the relevant details and structure them according to your request, while highlighting what information is not present.

Please note: This document's primary goal is to establish substantial equivalence of the entire CT system, not to provide a detailed clinical validation study for a new AI/ML algorithm that is the subject of separate regulatory submissions (like a De Novo or full PMA). The "Precise Image" and "Precise Position" modifications are treated as part of the overall system's equivalence demonstration.

Device: CT 5300 (incorporating updated Precise Image and Precise Position algorithms)

Study Proving Device Meets Acceptance Criteria (as described for the specific algorithms):

The document states: "Non-Clinical verification and or validation tests have been performed... Non-Clinical verification and or validation test results demonstrate that the proposed device... Meets the acceptance criteria and is adequate for its intended use."

Specifically for "Precise Image": "Precise Image (K210760) was modified for use in the CT 5300 system. With no changes to the algorithm architecture, new models were introduced to enable the reconstruction of a new organ type (cardiac), support more slice thickness and increment combinations, a new scan mode (high resolution head), and more clinical scenarios for body and head. All models were adequately trained and successfully compared using half-dose Precise Image reconstructions with full-dose iDose4 reconstructions. The comparative image quality assessment using phantoms demonstrated acceptable performance for the new models used in Precise Image. Additionally, a comparative image evaluation by two US Board Certified Radiologists of 126 image set pairs (including cases with pathology) comprising 31 unique patients representing the newly supported reconstructions. The comparative image assessment demonstrated that half-dose images processed by Precise Image in CT 5300, including both new and original existing models, are of equal or greater diagnostic quality compared to full dose images processed by iDose4. The comparative external image assessment confirms the validity of successful bench testing and clinical image quality evaluations, and when taken together, demonstrate Precise Image in CT 5300 to be as safe and effective as the predicate, and thus substantially equivalent to Precise Image (K210760) in predicate Incisive CT (K212441)."

For "Precise Position": "Precise Position (originally cleared in K203514) was modified for use in the CT 5300 with no change to the design of the AI algorithm, the body joints detection algorithm including CNN architecture, model parameters, inference pipeline, pre- and post-processing is same as what is used in the predicate Incisive CT. The original model was trained using a broad dataset and performance data using clinical images demonstrate the model can further support more exams (cardiac, spine, runoff)."

1. Table of Acceptance Criteria and Reported Device Performance

2. Sample Size Used for the Test Set and Data Provenance

• Precise Image:

  • Clinical Image Evaluation: 126 image set pairs "comprising 31 unique patients".
  • Data Provenance: Not explicitly stated, but "two US Board Certified Radiologists" implies the data used for the reading study was likely relevant to US clinical practice. It is mentioned as "retrospective clinical data" in the "Summary of Clinical Data" section (Page 10), but further details on geographical origin or specific institutions are not provided. The study is described as a "comparative image evaluation," which is by nature retrospective.
  • Phantom Study: Not specified (number of phantoms/scans).

• Precise Position:

  • Clinical Images: "performance data using clinical images" was used, but the specific sample size of images/patients for this evaluation is not provided.
  • Data Provenance: Not explicitly stated, but the "original model was trained using a broad dataset," suggesting varied provenance. The evaluation here uses "clinical images" and is implicitly retrospective.

3. Number of Experts Used to Establish Ground Truth for the Test Set and Qualifications

• Precise Image:

  • Number of Experts: Two (2)
  • Qualifications: "US Board Certified Radiologists"
  • Experience: Not specified (e.g., number of years of experience).

• Precise Position: Not explicitly stated for performance evaluation, but the "original model was trained using a broad dataset" which often implies some form of expert annotation or clinical data used as ground truth during training. For the evaluation of its extended support, it's mentioned that the model was tested with "performance data using clinical images," but details on expert review for this test set are not provided.

4. Adjudication Method for the Test Set

• Precise Image: The document describes a "comparative image evaluation by two US Board Certified Radiologists." It does not specify an adjudication method (e.g., 2+1, 3+1, or consensus reading). It's possible they read independently and their findings were compared, or they may have reached a consensus without formal arbitration by a third party.

• Precise Position: No details on expert review or adjudication method for the specified performance evaluation are provided.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study was Done, What was the Effect Size of How Much Human Readers Improve with AI vs Without AI Assistance

• No, a traditional MRMC comparative effectiveness study was not performed as described for human readers with/without AI assistance, in the sense of measuring diagnostic accuracy improvement.
• Instead, for "Precise Image," a comparative image quality assessment was performed between two different reconstruction methods: "half-dose Precise Image reconstructions" (new algorithm) vs. "full-dose iDose4 reconstructions" (predicate algorithm). The goal was to show non-inferiority or superiority of the image quality from the new algorithm at a reduced dose, rather than measuring reader performance improvement with AI assistance.
• The effect size related to reader performance is not applicable in this context as the study was about image quality comparison, not AI-assisted human reading. The conclusion was that the image quality was "equal or greater," which implicitly means readers can perform at least as well with the new images.

6. If a Standalone (i.e., algorithm only without human-in-the-loop performance) was Done

• Yes, for "Precise Image," a standalone image quality assessment was performed using phantoms. This is described as "The comparative image quality assessment using phantoms demonstrated acceptable performance for the new models used in Precise Image." This evaluates the algorithm's output (image quality) independent of human interpretation of complex clinical cases.
• For "Precise Position," it is a "body joints detection algorithm." While it's part of a workflow to assist patient positioning, a standalone performance metric (e.g., accuracy of joint detection) is not explicitly provided in this document, only that the "original model was trained using a broad dataset" and "performance data using clinical images demonstrate the model can further support more exams."

7. The Type of Ground Truth Used

• Precise Image:

  • Phantom Study: Ground truth would be based on the known characteristics of the phantom and imaging parameters, allowing for objective image quality metrics (e.g., spatial resolution, noise, contrast).
  • Clinical Image Evaluation: The ground truth for comparative diagnostic quality seems to be based on the "full-dose iDose4 reconstructions" themselves and inclusion of "cases with pathology." This implies that the full-dose images (processed by the predicate's iDose4 algorithm) were considered the reference truth against which the new algorithm's images were judged for diagnostic quality. The mention of "cases with pathology" likely means these cases had confirmed diagnoses, which serve as the ultimate ground truth for comparison.

• Precise Position: The ground truth for the training set would likely be human-annotated or verified positions of body joints on CT images. For the evaluation, it's implied that the "clinical images" and the intended "support more exams" served as the basis for performance verification.

8. The Sample Size for the Training Set

• Precise Image: The document states "All models were adequately trained." However, the specific sample size (number of images/patients) used for training the "new models" for Precise Image is not provided.
• Precise Position: The document states "The original model was trained using a broad dataset." The specific sample size (number of images/patients) used for training is not provided.

9. How the Ground Truth for the Training Set was Established

• Precise Image: The document states "All models were adequately trained." For deep learning image reconstruction like Precise Image, training typically involves paired low-dose and standard-dose (or high-quality) images, where the standard-dose images serve as the ground truth reference for the algorithm to learn how to enhance low-dose images. This is inferred but not explicitly stated in the provided text.
• Precise Position: The document states "The original model was trained using a broad dataset." For a "body joints detection algorithm," ground truth for training would typically be established by expert (e.g., radiologist or trained annotator) manual annotation of anatomical landmarks or joint locations on a large dataset of CT images. This is inferred but not explicitly stated in the provided text.

Ask a specific question about this device