(308 days)
The Philips Iterative Reconstruction Technique (IRT) Software Application is intended to reconstruct raw data from a Philips CT Scanner to produce images containing noise levels less than or equal to images produced by standard Filtered Back Projection reconstruction. Resulting IRT images are to be used to supplement conventional Filtered Back Projection images to aid the physician in diagnosis; they are not to be used as the sole basis for diagnosis.
The Philips IRT Software Application is a software option used for the reduction of noise in an image. IRT iteratively reconstructs raw data from a Philips CT Scanner to produce images containing noise levels less than or equal to images produced by standard Filtered Back Projection (FBP) reconstruction. This feature will be used by radiologists as a supplementary method to reconstruct CT raw data, in addition to traditional FBP.
Here's a breakdown of the acceptance criteria and the study information based on the provided text for the Philips Iterative Reconstruction Technique (IRT) Software Application:
1. Table of Acceptance Criteria and Reported Device Performance
| Acceptance Criteria Category | Acceptance Criteria | Reported Device Performance |
|---|---|---|
| Objective Image Quality | Noise levels less than or equal to images produced by standard Filtered Back Projection (FBP). | "IRT iteratively reconstructs raw data from a Philips CT Scanner to produce images containing noise levels less than or equal to images produced by standard Filtered Back Projection (FBP) reconstruction." and "Resulting data confirmed that the IRT application provides equivalent or better noise reduction." and "IRT affords a reduction in noise...with no degradation in high contrast spatial resolution, CT number accuracy, or CT number uniformity." |
| Low Contrast Detectability (LCD) | Improved LCD compared to FBP. | "The resulting data confirmed the improved LCD using the IRT application." and "...an improvement in LCD..." |
| High Contrast Spatial Resolution | No degradation compared to FBP. | "...no degradation in high contrast spatial resolution..." |
| CT Number Accuracy | No degradation compared to FBP. | "...no degradation in...CT number accuracy..." |
| CT Number Uniformity | No degradation compared to FBP. | "...no degradation in...CT number uniformity." |
Study Information
The provided text describes both non-clinical and clinical image data testing.
2. Sample size used for the test set and the data provenance:
- Non-clinical testing (Objective Image Quality):
- Test Set Description: Phantom raw CT scan data.
- Sample Size: Not explicitly stated, but implies multiple phantom scans were reconstructed for comparison.
- Data Provenance: Not specified, but generally phantom data is created in a controlled lab/testing environment.
- Non-clinical testing (Low Contrast Detectability - LCD):
- Test Set Description: Images used for an observer study with a low contrast test object.
- Sample Size: "The test was repeated multiple times for each test subject, and multiple test subjects were used." (Specific numbers not given).
- Data Provenance: Not specified, but likely generated in a controlled testing environment.
- Clinical testing (Noise Reduction):
- Test Set Description: Clinical image raw data sets.
- Sample Size: Not explicitly stated.
- Data Provenance: Not specified, but implied to be retrospective as it refers to existing "clinical image raw data sets."
3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts:
-
For the Low Contrast Detectability (LCD) observer study: "a cohort of human subjects was required to identify a low contrast test object..."
- Number of experts: Not specified beyond "multiple test subjects."
- Qualifications of experts: Not explicitly stated. It only mentions "human subjects," not necessarily radiologists or experts in image perception.
-
For Objective Image Quality and Clinical Noise Reduction where ground truth is based on physical measurements or direct comparison, no external experts are mentioned for establishing ground truth.
4. Adjudication method for the test set:
- Objective Image Quality and Clinical Noise Reduction: No explicit human adjudication method is described for these. The comparison seems to be based on direct measurement (noise, CT number uniformity, high contrast spatial resolution) or visual comparison of noise levels by Philips.
- Low Contrast Detectability (LCD): The "observer study" served as the evaluation method, where human subjects (not explicitly "adjudicators" in the sense of resolving conflicting interpretations, but rather participants identifying objects) made determinations. No specific adjudication protocol (like 2+1 or 3+1) is mentioned for resolving differences among these subjects, as the data was likely used to characterize the statistical nature of detection.
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:
- A Multi-Reader Multi-Case (MRMC) comparative effectiveness study was not explicitly described in the document in the typical sense of measuring reader performance with and without AI assistance for diagnosis.
- An observer study was conducted for Low Contrast Detectability (LCD), which involves multiple human subjects. However, its stated purpose was to confirm improved LCD using IRT, not to quantify diagnostic improvement of human readers with IRT versus without it as an assistive tool to be compared against a human-only baseline. The IRT images are intended to "supplement conventional Filtered Back Projection images to aid the physician in diagnosis," suggesting an assistive role, but the study described doesn't measure this specific comparative effectiveness for diagnostic improvement.
6. If a standalone (i.e., algorithm only without human-in-the-loop performance) was done:
- Yes, standalone performance was a primary focus.
- The "Objective image quality testing" using phantoms evaluated the algorithm's performance (noise, CT number uniformity, high contrast spatial resolution) directly by comparing IRT reconstructions to FBP reconstructions.
- The "Clinical image raw data sets" were also used to compare noise reduction between IRT and FBP.
- The IRT software application is described as an "iterative reconstruction technique" which processes raw data independently to produce images.
7. The type of ground truth used:
- Objective Image Quality (Phantoms): Ground truth was established by the known physical properties and measurements from the phantoms, adhering to methodologies like IEC 61223-3-5. This is a form of physical/definitive measurement ground truth.
- Low Contrast Detectability (LCD): Ground truth was the verifiable presence or absence of the low contrast test object within the images. This is a form of definitive presence/absence ground truth.
- Clinical Noise Reduction: Ground truth for noise reduction was based on direct comparison of noise levels between IRT and FBP reconstructions of the same raw clinical data. This relies on comparative measurement ground truth against an established baseline (FBP).
8. The sample size for the training set:
- The document does not provide any information about the training set size or how the algorithm was trained. This 510(k) summary focuses on the verification and validation (testing) of the final product.
9. How the ground truth for the training set was established:
- As the document does not mention a training set, it does not provide information on how its ground truth was established.
§ 892.1750 Computed tomography x-ray system.
(a)
Identification. A computed tomography x-ray system is a diagnostic x-ray system intended to produce cross-sectional images of the body by computer reconstruction of x-ray transmission data from the same axial plane taken at different angles. This generic type of device may include signal analysis and display equipment, patient and equipment supports, component parts, and accessories.(b)
Classification. Class II.