Search Results

The Viatronix V3D-Cardiac is intended to be used for the display and 2D/3D visualization of medical image data derived from CT of the human heart. The goal of this non-invasive, image analysis software package is to assist physician in diagnosing of cardiovascular disease to include, coronary artery disease, functional problem of left ventricle, myocardium anomaly, other heart diseases, and follow-up for stent/graft placement, bypasses and plaque imaging. The image post-processing tools are applied real time on-line by the physician's interactive demand. There is no image processing stage required prior to the start of the using of the V3D-Cardiac. The Viatronix V3D-Cardiac provides a set of fully or semi-automated tools, including, rib cage removal, initial coronary vessel tree segmentation, selected vessel segmentation, vessel crosssectional size measurements, and left ventricle volume functional parameter computation. The V3D-Cardiac will also provide automated setting and display of conventional cardiac imaging planes based on a single user input: selection of the aorta valve. The automatic 2D/3D view correlation is available for vessel analysis views. User can virtually fly thru the vessel lumen in the endoluminal 3D view. Manual tools for adjusting location of center of mitral valve for left ventricle region segmentation are provided. The V3D-Cardiac is intended for use by radiologist, clinicians and referring physicians to acquire, process, render, measure, evaluate, archive, print and distribute DICOM 3.0 compliant coronary artery, left ventricle and other heart anatomy images, utilizing PC hardware.

The Viatronix V3D-Cardiac is a software device for evaluating scanned images of heart. It is designed to aid the physician in analyzing the heart anatomy and detecting anomaly based on images from a CT scan. The heart anatomy includes coronary arteries, cardiac chambers, aorta root, cardiac valve, myocardium, and other parts of the heart. The goal is to simplify the physician's work as much as possible by providing fully or semi-automated tools for segmenting and measuring coronary artery and left ventricle and displaying reformatted images for visualizing complicated heart anatomy. It is an additional image processing option specific to heart imaging procedure added to our V3D visualization system product line, which pre-market clearance was granted by the FDA vide K002780, K013146, K020658, K022789, K032483, K033361, and K040126. It is a general software module, designed for use as a part of our V3D visualization system core technology. The V3D visualization system consists of V3D processor and V3D viewer in multiple computer configuration or V3D processor and V3D viewer in a stand alone one computer configuration. Upon receipt of contrast enhanced, multi-slice CT scan images of human heart in a DICOM format, the V3D processor converts the DICOM image data into an internally recognized volume data format using our core software technology. If there are more than one phase images available, the V3D-processor shall automatically extract each phase image into a separate volume and label the phase percentage information based on the header information from DICOM images. The V3D-Cardiac is an organ specific V3D viewer application. The V3D-Cardiac provides interactive orthogonal and multi-planar reformatted 2D and 3D images. User can evaluate those images for normality or malformation in specified part of heart obtained from scanned CT images.

Here's an analysis of the provided text regarding the acceptance criteria and study for the Viatronix V3D-Cardiac, revision 1.0 device:

No specific acceptance criteria or detailed study results are explicitly stated in the provided text. The submission focuses on demonstrating substantial equivalence to a predicate device rather than meeting pre-defined performance metrics. The "acceptance criteria" can be inferred as showing that "evaluation results of both predicate device and V3D-Cardiac were same and no significant differences were detected in the results of evaluation."

1. Table of Acceptance Criteria and Reported Device Performance

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

• Sample Size: Not explicitly stated. The text mentions "Scanned images datasets of actual patients were selected retrospectively" and "Tests and validations on actual patient data were performed per established protocol." There is no specific number provided for the patient studies used in the comparative evaluation.
• Data Provenance: "Scanned images datasets of actual patients were selected retrospectively." No specific country of origin is mentioned.

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

• Number of Experts: Not specified.
• Qualifications of Experts: Not specified. The studies involved comparison by "the physician" (singular, generic) and the output was "evaluated by a trained physician," but no details on their number or specific qualifications (e.g., years of experience, subspecialty) are provided.

4. Adjudication Method for the Test Set

• Adjudication Method: Not specified. The comparison was made between the predicate device and the V3D-Cardiac results, presumably by a physician. There's no mention of a multi-reader adjudication process (e.g., 2+1, 3+1).

5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study

• Was it done?: No, a formal MRMC comparative effectiveness study is not described. The study design directly compares the V3D-Cardiac's output against the predicate device's output on the same datasets, implying a human-in-the-loop comparison for equivalence, not a study to quantify human reader improvement with AI assistance.
• Effect Size of Human Readers Improvement with AI vs. without AI assistance: Not applicable, as this type of MRMC study was not performed. The study aimed at demonstrating equivalence to a predicate device's functionality, not an enhancement to human performance.

6. Standalone Performance Study (Algorithm Only)

• Was it done?: Yes, in a sense. The comparison was described as: "Same image series were loaded into the Viatronix V3D-Cardiac application and the results of evaluation and quantification of coronary arteries and ejection fraction for left ventricle were recorded." This was then compared to the predicate device's recorded results. While "human-in-the-loop" is mentioned for using the V3D-Cardiac ("The image post-processing tools are applied real time on-line by the physician's interactive demand"), the evaluation of the results against the predicate device can be considered showcasing the algorithm's performance indirectly, as it's the output of the algorithm that's being compared. However, it's not a standalone performance measured against a true "ground truth" derived independently of a device, but rather against another device's output. The internal validation against V3D-Vascular for vessel measurements could also be considered an algorithm-to-algorithm comparison (standalone).

7. Type of Ground Truth Used

• Type of Ground Truth: The primary "ground truth" for the main equivalence study was implicitly the output/evaluation from the predicate device (GE Medical System, CardIQ Xpress, revision 6.12.3). The study compared the V3D-Cardiac's evaluation results to those obtained from the predicate device.
  • For internal validation of vessel measurements, the "ground truth" was the measurements created from V3D-Vascular, revision 2.0.
  • No independent "gold standard" like pathology or long-term outcomes data is mentioned as a ground truth source.

8. Sample Size for the Training Set

• Sample Size: Not provided. The document highlights software development processes consistent with standards, suggesting testing and validation, but does not detail the size or nature of data used for training any automated components (e.g., segmentation models). Since the device was cleared in 2008, it's less likely to rely on deep learning models that require very large "training sets" in the modern sense.

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

• Ground Truth Establishment: Not provided. Given the nature of medical imaging software at the time of this submission (2008), automated components would likely use rule-based algorithms or traditional image processing techniques rather than machine learning requiring labeled training data. If any "training" data (e.g., for algorithm tuning) was used, the method for establishing its ground truth is not described.

Ask a specific question about this device