Search Results

Clarify DL is a deep learning-based image reconstruction method intended to reduce noise and improve image quality of Nuclear Medicine bone SPECT scintigraphy images. Clarify DL may be used for patients of all ages.

Clarify DL is a software-only device for the reconstruction of NM bone SPECT scintigraphy images obtained using supported GE HealthCare SPECT and SPECT-CT systems. Clarify DL is designed to reduce image noise while maintaining contrast to enable increased Contrast-to-Noise Ratio (CNR) and Contrast Recovery Coefficient (CRC). Clarify DL is deployed within the software of the predicate Xeleris V Processing and Review System and StarGuide devices.

The provided text describes the acceptance criteria and the study conducted for Clarify DL, a deep learning-based image reconstruction method for Nuclear Medicine bone SPECT scintigraphy images.

Acceptance Criteria and Reported Device Performance

The core acceptance criteria for Clarify DL revolve around its ability to reduce image noise while maintaining contrast, thereby improving image quality. The specific metrics assessed are:

The text explicitly states that "All readers attested that their assessments of Clarify DL demonstrated acceptable diagnostic results" for overall image quality, image resolution, and noise level. Furthermore, the additional engineering bench testing "substantiated all performance claims," including improvement in CNR and CRC.

Study Details

The evaluation of Clarify DL involved both non-clinical (bench) testing and clinical reader evaluation.

Clinical Reader Evaluation (Test Set):

1. Sample size used for the test set: The text mentions "clinical bone SPECT and SPECT-CT exams obtained using supported SPECT CT systems" from "11 hospitals." The exact number of exams (cases) used for the test set is not specified.
2. Data provenance: The clinical exams came from "11 hospitals, including 5 US hospitals from 4 different states (i.e Ohio, Kansas, Rhode Island, and California) and 6 non-US hospitals from Europe and Asia." This indicates a prospective or retrospective collection of real-world clinical data from a diverse geographical range. The studies are described as "clinical bone SPECT and SPECT-CT exams."
3. Number of experts used to establish the ground truth for the test set and their qualifications: The clinical reader evaluation involved "expert NM physicians" but the number of experts is not specified. Their qualifications are given as "expert NM physicians." No specific years of experience are provided.
4. Adjudication method for the test set: The text states, "The clinical reader evaluation included review by expert NM physicians to assess overall Image Quality, Image Resolution, and Noise Level. The results of the clinical reader evaluation support the determination of substantial equivalence. All readers attested that their assessments of Clarify DL demonstrated acceptable diagnostic results." This implies a consensus or individual assessment model, but a specific adjudication method (e.g., 2+1, 3+1) is not explicitly mentioned. Given that "all readers attested," it suggests a strong agreement or individual positive assessments were sufficient.
5. Multi-reader multi-case (MRMC) comparative effectiveness study: The text does not explicitly state that an MRMC comparative effectiveness study was done to compare human readers with and without AI assistance. The clinical reader evaluation assessed the Clarify DL images directly.
6. Standalone (algorithm-only without human-in-the-loop performance): Yes, standalone performance was evaluated through "additional engineering bench testing". This included image-level metrics (SSIM, MSE, PSNR) and lesion-level metrics (CNR, CRC) to quantitatively assess the algorithm's performance independent of human interpretation.
7. Type of ground truth used: For the clinical reader evaluation, the ground truth was expert consensus/interpretation by "expert NM physicians" regarding image quality, resolution, and noise level. For the bench testing, the "ground truth" implicitly referred to a reference image or quantitative metric against which the performance improvements (e.g., noise reduction, contrast enhancement) of Clarify DL were measured.
8. Sample size for the training set: The text does not specify the sample size for the training set used for the deep learning model.
9. How the ground truth for the training set was established: The text does not explicitly describe how the ground truth for the training set was established. It only states that Clarify DL uses an "OSEM-based Iterative Image Reconstruction with integrated deep learning model trained to reduce noise."

Ask a specific question about this device