(120 days)
IDEAL IQ is a software option intended for use on GE MR systems. IDEAL IQ is capable of producing transverse, sagittal, coronal, and oblique images of internal structures of the body, including but not limited to, the musculoskeletal, breast, abdominal, and neurological systems. Specific anatomical regions that may be imaged include the abdomen, breast, spine, joints, and extremities.
IDEAL IQ is an acquisition and reconstruction technique that simultaneously obtains independent images of hydrogen nuclei that resonate at different frequencies to provide separation of water and triglyceride fat. IDEAL IQ generates images of separated water and triglyceride fat, relative triglyceride fat fraction map, and tissue transverse magnetization relaxation. In the liver, the relative triglyceride fat fraction map is quantitative; it reflects the proton density (number of protons per unit volume) of triglyceride fat, divided by the sum of the proton density of triglyceride fat and the proton density of water, on a voxel-by-voxel basis.
When interpreted by a trained physician, these images provide information that can aid in diagnosis.
IDEAL IQ is a software application offered as an option for GE MR scanners. The IDEAL IQ imaging technique (IDEAL: Iterative Decomposition of water and fat with Echo Asymmetry and Least-squares estimation) is a triglyceride fat and water separation technique that acquires multiple images of the anatomy at separate echo times to calculate the phase differences and determine triglyceride fat and water content per pixel. It exploits the resonance frequency differences between triglyceride fat and water, measured as phase differences in multiple echoes, to resolve triglyceride fat and water. It provides reliable and uniform water-fat separation in the presence of B0 field inhomogeneity and improves the accuracy of water-fat separation by estimating and correcting for T2* decay between echoes and by more accurately modeling triglyceride fat's spectral profile as multiple peaks rather than a single peak. It produces images showing the separated water and triglyceride fat signals, and the tissue transverse magnetization relaxation. Finally, IDEAL IO processes decomposed water and triglyceride fat images to generate a relative triglyceride fat-fraction map. Such a representation of a proton density triglyceride fat fraction is intrinsically insensitive to B1 and coil-sensitivity heterogeneity. The IDEAL IO method uses a low flip angle excitation to reduce any TI bias in the relative triglyceride fat fraction map images caused by differences in the TI of triglyceride fat and water. With the confounding effects of T2*, multiple spectral peaks of triglyceride fat, and T1 differences reduced, the images from IDEAL IQ reflect the spatial distribution of relative concentration of triglyceride fat in a voxel.
The provided text describes the "IDEAL IQ Software Option" for GE MR scanners. However, the document does not contain a table of acceptance criteria or a detailed study that explicitly proves the device meets specific performance criteria through a clinical trial with reported outcomes against predefined metrics.
The submission focuses on establishing substantial equivalence to a predicate device (IDEAL Software Option, K072998) and outlining non-clinical tests for safety and internal quality assurance.
Here's an attempt to answer your questions based on the available information, noting where information is not present:
1. A table of acceptance criteria and the reported device performance
Not provided in the document. The submission states: "Clinical image comparisons demonstrate that the IDEAL IQ software option maintains the same imaging performance results as the predicate software option IDEAL." This implies the acceptance criterion was "maintaining the same performance as the predicate device," but no specific metrics or reported performance values are given.
2. Sample size used for the test set and the data provenance
Not explicitly provided. The document mentions "several in vivo human studies showing correlation to MR Spectroscopy" for quantification in the liver. It also refers to "clinical image comparisons." However, no specific sample sizes for these studies or their data provenance (e.g., country of origin, retrospective/prospective) are detailed.
3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts
Not explicitly provided. The document mentions "When interpreted by a trained physician, these images provide information that can aid in diagnosis." This implies expert interpretation, but the number or qualifications of these experts are not specified for any studies mentioned.
4. Adjudication method for the test set
Not provided.
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
No MRMC comparative effectiveness study is described for this device. The IDEAL IQ is described as a software option for acquisition and reconstruction to generate specific image types (separated water and fat images, fat-fraction maps). It is not presented as an AI-powered diagnostic aid that improves human reader performance. Its purpose is to provide quantitative information (e.g., relative triglyceride fat-fraction map) that aids diagnosis.
6. If a standalone (i.e., algorithm only without human-in-the-loop performance) was done
Yes, the device has standalone performance. The IDEAL IQ software itself generates the separated water and fat images, the relative triglyceride fat-fraction map, and tissue transverse magnetization relaxation maps. The "quantification in the liver" is shown through "correlation to MR Spectroscopy," suggesting a direct comparison of the algorithm's output (fat-fraction map) to a reference standard (MR Spectroscopy). The key output, the "relative triglyceride fat-fraction map," is inherently quantitative, reflecting the proton density of triglyceride fat per voxel.
7. The type of ground truth used
For quantification in the liver (fat-fraction map), the ground truth used was MR Spectroscopy.
For the general "clinical image comparisons" to demonstrate "same imaging performance results as the predicate software option IDEAL," the nature of the ground truth is not specified but would likely involve expert visual assessment and comparison for image quality and diagnostic utility, though this is inferred.
8. The sample size for the training set
The document describes IDEAL IQ as a software option for image acquisition and reconstruction that exploits physical phenomena (resonance frequency differences between fat and water). It is not a machine learning or AI algorithm that typically has a "training set" in the conventional sense. Therefore, a training set size is not applicable or provided.
9. How the ground truth for the training set was established
As there is no traditional "training set" for this type of image reconstruction algorithm, this question is not applicable. The algorithm's principles are based on established physics and signal processing.
§ 892.1000 Magnetic resonance diagnostic device.
(a)
Identification. A magnetic resonance diagnostic device is intended for general diagnostic use to present images which reflect the spatial distribution and/or magnetic resonance spectra which reflect frequency and distribution of nuclei exhibiting nuclear magnetic resonance. Other physical parameters derived from the images and/or spectra may also be produced. The device includes hydrogen-1 (proton) imaging, sodium-23 imaging, hydrogen-1 spectroscopy, phosphorus-31 spectroscopy, and chemical shift imaging (preserving simultaneous frequency and spatial information).(b)
Classification. Class II (special controls). A magnetic resonance imaging disposable kit intended for use with a magnetic resonance diagnostic device only is exempt from the premarket notification procedures in subpart E of part 807 of this chapter subject to the limitations in § 892.9.