Search Results

uCT 780 is a computed tomography x-ray system intended to produce cross-sectional images of the body by computer reconstruction of x-ray transmission data taken at different angles and planes and indicated for the whole body (including head, neck, cardiac and vascular).

uCT 780 is intended to be used 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 either a governmental body or professional medical society.

uWS-CT-Dual Energy Analysis is a post-processing software package that accepts UIH CT images acquired using different tube voltages and/or tube currents of the same anatomical location. The various materials of an anatomical region of interest have different attenuation coefficients, which depend on the used energy. These differences provide information on the chemical composition of the scanned body materials and enable images to be generated at multiple energies within the available spectrum. uWS-CT-Dual Energy Analysis software combines images acquired with low and high energy spectra to visualize this information.

The Computed Tomography X-ray System, uCT 780, is intended to produce cross-sectional images of the patient by computer reconstruction of X-ray transmission data taken at different angles and planes. These images may be obtained either with or without contrast.

The uWS-CT-Dual Energy Analysis is a post-processing software package that accepts UIH CT images acquired using different tube voltages and/or tube currents of the same anatomical location. The various materials of an anatomical region of interest have different attenuation coefficients, which depend on the used energy. These differences provide information on the chemical composition of the scanned body materials. CT dual energy analysis application combines images acquired with low and high energy spectra to visualize this information. In this submission, the DETLA (Deep Recon) feature of the proposed device is modified and there is no other significant change compared with K230162.

Here's a breakdown of the acceptance criteria and the study proving the device meets them, based on the provided text:

1. Table of Acceptance Criteria and Reported Device Performance

Device: DELTA (Deep Recon) algorithm for uCT 780

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

• Sample Size for Test Set:
  • Validation Dataset: 20 cases (5 for each body part: abdomen, chest, head, cardiac).
  • Test Dataset: 40 cases (10 for each body part: abdomen, chest, head, cardiac).
• Data Provenance: Not explicitly stated (e.g., country of origin). The data is from "sample clinical data" and "different head, chest, abdomen and cardiac protocol, common diseases, different scanning dose and a wide range of population." It's described as "clinical datasets," implying real patient data. It is not explicitly stated whether it is retrospective or prospective, but the phrasing "sample clinical data" suggests it could be retrospective.

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

• Number of Experts: 2
• Qualifications: Board-certified U.S. radiologists.

4. Adjudication Method for the Test Set

The text states that the image quality was scored by the 2 radiologists, implying individual assessments rather than a formal adjudication method like 2+1 or 3+1. There is no mention of a process to resolve disagreements between the two radiologists' scores.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study was done, and its effect size

While a reader study was performed with multiple readers (2 radiologists) and multiple cases, it did not explicitly state it as an MRMC comparative effectiveness study in the typical sense of measuring improvement in human performance with AI vs. without AI assistance.

Instead, the study evaluated the image quality of images reconstructed by DELTA (AI-based) compared to FBP (standard) directly. The radiologists scored the image quality of the images themselves, not their diagnostic performance with or without the DELTA reconstructions.

Therefore, no effect size of how much human readers improve with AI vs. without AI assistance is provided.

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

Yes, a significant part of the evaluation was a standalone algorithm performance assessment through Bench Testing - Phantom Test. This involved objective measurements of image quality metrics (IQ, noise power spectrum, mean CT value, LCD, spatial resolution) using various phantoms, without human interpretation involved.

The "Clinical Assessment" also indirectly assesses the algorithm's standalone performance by having radiologists score the output images of the algorithm where the algorithm is the primary variable (DELTA vs. FBP).

7. The Type of Ground Truth Used

• Bench Test (Phantom Testing): The ground truth for the bench tests would be the known physical properties and measurements from the phantoms themselves (e.g., precise HU values in known materials, known low-contrast objects, quantifiable spatial resolution targets).
• Clinical Assessment (Reader Study): The ground truth for the clinical assessment was based on expert consensus (or at least expert scoring) from the two board-certified U.S. radiologists. They performed a five-point scale evaluation of image quality aspects (noise, structure fidelity, and overall image quality), which serves as the "ground truth" for the perceived image quality. There is no mention of pathology or outcomes data being used as ground truth for the clinical evaluation of image quality.

8. The Sample Size for the Training Set

The sample size for the training set is not explicitly given. The document mentions an increased dataset size for the DELTA algorithm: "The new dataset was increased to 127 cases compared with the original dataset." This "new dataset" would likely include training data, but it's not specified how many of these 127 cases were specifically used for training. It also states, "The performance testing for DELTA (Deep Recon) algorithm was performed on 60 subjects during the product development." The phrase "during the product development" might imply the training phase, but it's ambiguous.

However, it explicitly states: "Training dataset do not contain any validation dataset or test dataset of DELTA algorithm."

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

The document does not explicitly describe how the ground truth for the training set (e.g., the "new dataset increased to 127 cases") was established. For deep learning reconstruction algorithms like DELTA, the "ground truth" for training often involves:

• Reference standard images: High-quality, higher-dose CT scans (or simulated ideal images) that the low-dose or noisy input images are trained to match or approximate.
• Image quality metrics: Objective computational metrics might be used to label or guide the training process to optimize for reduced noise, improved spatial resolution, etc.

Without further information, the exact method for establishing ground truth for the training data remains unspecified.

Ask a specific question about this device