Search Results

cvi42 is intended to be used for viewing, post-processing, qualitative and quantitative evaluation of cardiovascular magnetic resonance (MR) images and computed tomography (CT) images in a Digital Imaging and Communications in Medicine (DICOM) Standard format.

lt enables:

· Importing cardiac MR & CT Images in DICOM format.

• Supporting clinical diagnostics by qualitative analysis of cardiac MR & CT images using display functionality such as panning, windowing, zooming, navigation through series/slices and phases, 3D reconstruction of images including multiplanar reconstructions of the images.

• Supporting clinical diagnostics by quantitative measurement of the heart and adjacent vessels in cardiac MR & CT images, specifically signal intensity, distance, area, volume, and mass.

• Supporting clinical diagnostics by using area and volume for measuring cardiac function and derived parameters cardiac output and cardiac index in long axis and short axis cardiac MR & CT images.

• Flow quantifications based on velocity encoded cardiac MR images (including two and four dimensional flow analysis).

• Strain analysis of cardiac MR images by providing measurements of 2D LV myocardial function (displacement, velocity, strain, strain rate, time to peak, and torsion).

· Supporting clinical diagnostics of cardiac CT images including quantitative measurements of calcified plaques in the coronary arteries (calcium scoring), specifically Agatston and volume and mass calcium scores, visualization and quantitative measurement of heart structures including coronaries, femoral, aortic, and mitral valves.

· Evaluating CT and MR images of blood vessels. Combining digital image processing and visualization tools such as multiplanar reconstruction (MPR), thin/thick maximum intensity projection (MIP), inverted MIP thin/thick, volume rendering technique (VRT), curved planar reformation (CPR), processing tools such as bone removal (based on both single energy and dual energy) table removal and evaluation tools (vessel centerline calculation, lumen calculation, stenosis calculation) and reporting tools (lesion location, lesion characteristics) and key images. The software package is designed to support the physician in confirming the presence of physician identified lesion in blood vessels and evaluation, documentation and follow up of any such lesions.

cvi42 shall be used by qualified medical professionals, experienced in examining and evaluating cardiovascular MR or CT images, for the purpose of obtaining diagnostic information as part of a comprehensive diagnostic decision-making process. cvi42 is a software application that can be used as a stand-alone product or in a networked environment.

The target population for cvi42 and its manual workflows is not restricted; however, cvi42's semiautomated machine learning algorithms, included in the MR Function and CORE CT modules, are intended for an adult population. Further, image acquisition by a cardiac MR or CT scanner may limit the use of the software for certain sectors of the general public.

cvi42 shall not be used to view or analyze images of any part of the body except the cardiac images acquired from a cardiovascular magnetic resonance or computed tomography scanner.

cvi42 Software Application ("cvi42") is a software as a medical device (SaMD) that is intended for evaluating CT and MR images of the cardiovascular system. Combining digital image processing, visualization, quantification, and reporting tools, cvi42 is designed to support physicians in the evaluation and analysis of cardiovascular imaqing studies.

cvi42 uses machine learning techniques to aid in semi-automatic contouring of regions of interest in cardiac MR or CT images.

The data used to train these machine learning algorithms were sourced from multiple clinical sites from urban centers and from different countries. When selecting data for training, the importance of model generalization was considered and data was selected such that a good distribution of patient demographics, scanner, and image parameters were represented. The separation into training versus validation datasets is made on the study level to ensure no overlap between the two sets. As such, different scans from the same study were not split between the training and validation datasets. None of the cases used for model validation were used for training the machine learning models.

cvi42 has a graphical user interface which allows users to analyze cardiac MR & CT images qualitatively and quantitatively.

cvi42 accepts uploaded data files previously acquired by MR or CT scanners or other data collection equipment but does not interface directly with such equipment. Its functionality is independent of the type of vender acquisition equipment. The analysis results are available onscreen and can be saved with the software for future review.

Here's a breakdown of the acceptance criteria and study details for the cvi42 Software Application, based on the provided document:

1. Table of Acceptance Criteria and Reported Device Performance

For cvi42 Auto (MR-CMR Function, CORE CT Coronary, and CORE CT-Calcium):

For CORE CT (CT Function Module):

2. Sample Size for the Test Set and Data Provenance

For cvi42 Auto (MR-CMR Function, CORE CT Coronary, and CORE CT-Calcium):

• Sample Size: n = 235 anonymized patient images acquired from major vendors of MR and CT imaging devices.
  • 70 samples for Coronary Analysis
  • 102 samples for Calcium analysis
  • 63 samples for SAX Function contouring
  • 63 samples for each of 2-CV, 3-CV, and 4CV LAX function contouring
  • 252 samples for Function Classification
• Data Provenance: Images were acquired from major vendors of MR and CT imaging devices. At least 50% of the data came from a U.S. population. The document does not specify if the data was retrospective or prospective, but the phrasing "were used for the validation" implies retrospective use of existing data.

For CORE CT (CT Function Module):

• Sample Size: Not explicitly stated, but the validation data was sourced from 9 different sites.
• Data Provenance: Sourced from 9 different sites, with 90% of the data sampled from US sources. The document does not specify if the data was retrospective or prospective.

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

For CORE CT (CT Function Module):

• Number of Experts: Three expert readers.
• Qualifications: "Expert readers" – specific qualifications (e.g., years of experience, board certification) are not detailed in the provided text.

For cvi42 Auto (MR-CMR Function, CORE CT Coronary, and CORE CT-Calcium), the document does not explicitly state the number of experts used to establish ground truth for the test set. It does mention expert readers for the comparison in the CORE CT section.

4. Adjudication Method for the Test Set

For CORE CT (CT Function Module):

• The "reference standard" was "established from three expert readers." The specific adjudication method (e.g., majority vote, specific consensus process) is not detailed, but it implies a consensus or agreement among these three experts.

For cvi42 Auto (MR-CMR Function, CORE CT Coronary, and CORE CT-Calcium):

• The document does not explicitly state the adjudication method for establishing ground truth for these modules.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study was done

• No, an MRMC comparative effectiveness study was not explicitly stated to have been done to measure human reader improvement with AI vs. without AI assistance. The performance tests described are primarily focused on the standalone performance of the AI algorithms (Machine Learning Derived Outputs) compared to a ground truth or a reference standard established by experts.

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

• Yes, standalone performance was assessed. The sections titled "Validation of Machine Learning Derived Outputs" and "CORE CT: CT Function" describe the evaluation of the algorithms' performance (e.g., classification accuracy, MAE, Dice coefficient, HD, EF bias) against pre-defined acceptance criteria and a reference standard made by experts, without human-in-the-loop assistance for the AI's output generation. This is a standalone assessment of the algorithms.

7. The Type of Ground Truth Used

• Expert Consensus: For the CORE CT module, the ground truth (reference standard) used for evaluation was established by "three expert readers." This implies an expert consensus or expert-derived ground truth.
• For other modules (cvi42 Auto), the document states that performance was evaluated against "pre-defined acceptance criteria" but does not explicitly describe how the ground truth for those criteria was established, though it likely involved expert annotations or established clinical metrics.

8. The Sample Size for the Training Set

• The document states: "The data used to train these machine learning algorithms were sourced from multiple clinical sites from urban centers and from different countries." However, the specific sample size for the training set is not provided for any of the modules. It only mentions that the data was selected for good distribution of patient demographics, scanner, and image parameters.

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 was established. It only states that the training data "were sourced from multiple clinical sites from urban centers and from different countries." It also notes that "the separation into training versus validation datasets is made on the study level to ensure no overlap between the two sets." This suggests that the training data would have had associated ground truth data (e.g., expert annotations, clinical measurements) to enable supervised learning, but the method of establishing that ground truth is not detailed.

Ask a specific question about this device