(87 days)
The Emory Cardiac Toolbox™ 4.0 software program should be used for the quantification of myocardial perfusion for the display of wall motion and quantification of left-ventricular function parameters from SPECT & PET myocardial perfusion studies (EGS™), for the 3D alignment of coronary artery models from CT coronary angiography onto the left ventricular 3D epicardial surface, for the assessment of cardiac mechanical dyssynchrony using phase analysis, for generation of the short axis, vertical, and horizontal long axis tomograms from the SPECT raw data using either filtered backprojection (FBP) or iterative reconstruction (MLEM/OSEM), for the quantification of myocardial blood flow and coronary flow reserve, and for the decision support in interpretation (LVX) and automatic structured reporting of the study.
The product is intended for use by trained nuclear technicians and nuclear medicine or nuclear cardiology physicians. The clinician remains ultimately responsible for the final interpretation and diagnosis based on standard practices and visual interpretation of all SPECT and PET data.
The Emory Cardiac Toolbox™ 4.0 is used to display gated wall motion and for quantifying parameters of left-ventricular perfusion and function from gated SPECT & PET myocardial perfusion studies and for the evaluation of dynamic PET studies. These parameters are: perfusion, ejection fraction, end-diastolic volume, end-systolic volume, myocardial mass, transient ischemic dilatation (TID), analysis of coronary blood flow and coronary flow reserve, and assessment of cardiac mechanic dyssynchrony. In addition, the program offers the capability of providing the following diagnostic information: computer assisted visual scoring, prognostic information, and expert system image interpretation. The program can also be used for the 3D alignment of coronary artery models from CT coronary angiography onto the left ventricular 3D epicardial surface and for generation of the short axis, vertical, and horizontal long axis tomograms from the SPECT raw data either filtered backprojection (FBP) or iterative reconstruction using (MLEM/OSEM). The Emory Cardiac Toolbox can be used with any of the following Myocardial SPECT Protocols: Same Day and Two Day Sestamibi, Dual-Isotope (Tc-99m/Tl-201), Tetrofosmin, and Thallium, Rubidium-82, Rubidium-82 with CTbased attenuation correction, N-13-ammonia, FDG protocols, and user defined normal databases. This program was developed to run in the .NET operating system environment which can be executed on any PC, any nuclear medicine computer system, or through a web browser. In addition, the program can be used for the decision support in interpretation and automatic structured reporting of the study. The program processes the studies automatically, however, user verification of output is required and manual processing capability is provided.
Here's an analysis of the acceptance criteria and study information for the Emory Cardiac Toolbox™ 4.0, extracted from the provided 510(k) summary:
Acceptance Criteria and Device Performance
The 510(k) summary for Emory Cardiac Toolbox™ 4.0 primarily focuses on demonstrating substantial equivalence to predicate devices and detailing the validation studies for its new features. It does not explicitly state pre-defined acceptance criteria in a quantitative format as one might find in a clinical trial protocol. Instead, it describes successful validations and accuracy against established methods or predicate devices.
However, based on the information provided, we can infer the performance goals for certain aspects of the device.
| Feature/Parameter | Acceptance Criteria (Inferred) | Reported Device Performance |
|---|---|---|
| IDL to .NET Conversion Accuracy (for 14 perfusion, function, and dyssynchrony variables) | To demonstrate accuracy in deriving "similar values" when converting from the IDL to the .NET operating system. | Successfully achieved "accuracy greater than 99%." |
| Coronary Blood Flow Validation | Comparable performance to cleared predicate devices (INVIA's CORRIDOR 4DM V2010 and cfrQuant). | Validation conducted successfully in 44 patient studies, contributing to the claim of safety and effectiveness, and substantial equivalence to predicate devices. |
| Decision Support Validation (LVX) | Comparable performance to cleared predicate devices (Emory Cardiac Toolbox v2.0 K992450 (PERFEX) and Rcadia's COR Analyzer K110071). | Validation conducted successfully in 126 studies, contributing to the claim of safety and effectiveness, and substantial equivalence to predicate devices. |
Study Details
2. Sample Sizes Used for the Test Set and Data Provenance
- IDL to .NET Conversion: 301 patients. Data provenance is not explicitly stated (e.g., country, retrospective/prospective).
- Coronary Blood Flow Validation: 44 patient studies. Data provenance is not explicitly stated.
- Decision Support Validation (LVX): 126 studies. Data provenance is not explicitly stated.
- Previous Versions Referenced (ECT 2.0, 2.1, 2.6, 3.1):
- ECT 2.0: 217 patients for LV functional parameter evaluation (in-house trials), 80 patients for a multicenter trial validation. Computer assisted visual scoring (20 patients), prognosis (504 patients), expert system (461 patients), and coronary fusion (9 patients).
- ECT 2.1: Rb-82 normal limits (n=176), PET tools for perfusion-metabolism match-mismatch (n=90).
- ECT 2.6: N-13-ammonia normal limits (n=144), alignment method for 3D CT coronary artery onto LV 3D epicardial surface (phantom and 8 patients).
- ECT 3.1: SPECT reconstruction (phantom, animal, and 4 patients for development; 10 patients for prospective validation); Phase analysis (90 normal patients for development; 75 additional patients for prospective validation).
The provided text does not specify the country of origin for the data or if the studies were retrospective or prospective, except for two instances in ECT 3.1: "prospective validation in 10 patients" for SPECT reconstruction and "prospective validation in 75 additional patients" for phase analysis.
3. Number of Experts Used to Establish the Ground Truth for the Test Set and Their Qualifications
The document does not explicitly state the number of experts or their specific qualifications (e.g., "radiologist with 10 years of experience") for establishing ground truth for the test sets described for version 4.0.
However, for the original Emory Cardiac Toolbox™ 2.0, it references an article "by Vansant et al Emory Cardiac Toolbox™ (CEquar®, EGS™) Version 2.0, Ref. 510(k) #: K992450 and Version 2.1, Ref. 510(k) #: K014033)" for initial program accuracy. The details of how ground truth was established in those studies would likely be found in those referenced 510(k)s or publications.
For the decision support system (LVX), given its intended use "to assist a trained physician to analyze nuclear cardiology images" and that the "physician should integrate all of the patients' clinical and diagnostic information... prior to making his final interpretation," it implies that the physicians using the system acted as the ultimate arbiters of truth, but how "ground truth" for the validation study itself was established for the 126 studies is not detailed.
4. Adjudication Method for the Test Set
The document does not explicitly describe an adjudication method (like 2+1 or 3+1 consensus) for establishing ground truth for the test sets mentioned in the 4.0 validation or earlier versions.
5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study Was Done
The document does not indicate that a Multi-Reader Multi-Case (MRMC) comparative effectiveness study was done to quantify how much human readers improve with AI vs. without AI assistance. The new features for version 4.0 (coronary blood flow, coronary flow reserve, and decision support) were primarily validated by demonstrating their derivation accuracy or by comparing their results to predicate devices, rather than through a comparative reader study with and without AI assistance.
6. If a Standalone (i.e., algorithm only without human-in-the-loop performance) Was Done
Yes, standalone performance was assessed for the new features. The validation of the "accuracy greater than 99%" for the IDL to .NET conversion, the "coronary blood flow validation" in 44 patients, and the "decision support validation" in 126 studies describe the algorithm's performance in generating specific measurements or interpretations. While user verification is required and manual processing is possible, the reported validations focus on the program's ability to process studies and produce results, which is a form of standalone evaluation. The device is intended to provide "decision support," meaning its output is fed to a human for final interpretation, but its internal validation steps would typically evaluate its standalone computational accuracy.
7. The Type of Ground Truth Used
- IDL to .NET Conversion: The ground truth would likely be the values derived from the previous, established IDL version of the software. The goal was to show the new .NET version produced "similar values."
- Coronary Blood Flow and Coronary Flow Reserve: The ground truth for these quantitative measurements would typically be established against other validated quantitative methods for measuring blood flow and reserve, implicitly by comparison/equivalence to the predicate devices (INVIA's CORRIDOR 4DM V2010 and cfrQuant).
- Decision Support System (LVX): For decision support systems, ground truth is often clinical outcomes, invasive procedures (like angiography for coronary artery disease), or expert consensus on image interpretation. The text states equivalence to PERFEX and COR Analyzer, implying the ground truth methods used to validate those predicates (which are not detailed here) would be relevant. The phrasing "physician should integrate all of the patients' clinical and diagnostic information... prior to making his final interpretation" suggests that the ultimate "truth" is the comprehensive clinical diagnosis.
8. The Sample Size for the Training Set
The document does not explicitly provide a sample size for a training set directly associated with the new features of version 4.0. The studies described for version 4.0 (IDL to .NET conversion, coronary blood flow, decision support) are presented as validation studies, which typically use independent test sets, not training sets.
However, the text mentions "development" for earlier versions:
- ECT 2.1: "development and validation of Rb-82 normal limits (n=176)"
- ECT 2.6: "development and validation of N-13-ammonia normal limits (n=144)"
- ECT 3.1: "development (phantom, animal, and patients n=4)" for SPECT reconstruction and "development in 90 normal patients" for phase analysis.
These "development" activities would involve data used to build or refine algorithms, akin to training data, but they are tied to earlier versions of the product.
9. How the Ground Truth for the Training Set Was Established
As no specific training sets are described for the version 4.0 updates, the method for establishing ground truth for such sets is not mentioned. For the "development" phases of earlier versions (Rb-82 normal limits, N-13-ammonia normal limits, phase analysis in normal patients), the ground truth for "normal limits" would typically be derived from a healthy population cohort, likely using established clinical criteria to define "normal." For animal and phantom studies, the "ground truth" is typically the known physical properties or induced conditions.
§ 892.1200 Emission computed tomography system.
(a)
Identification. An emission computed tomography system is a device intended to detect the location and distribution of gamma ray- and positron-emitting radionuclides in the body and produce cross-sectional images through computer reconstruction of the data. This generic type of device may include signal analysis and display equipment, patient and equipment supports, radionuclide anatomical markers, component parts, and accessories.(b)
Classification. Class II.