Autoplaque

Autoplaque 3.0 is a non-invasive, workstation-based software application that utilizes deep learning algorithms for post-processing of coronary CT angiographic images to analyze coronary anatomy, quantify plaque, and assess luminal stenosis. It provides an interactive tool for cardiologists and radiologists by performing automated segmentation and measurements, aiding in the determination of coronary pathology and treatment paths while requiring users to be aware of limitations like interpolated data.

Autoplaque 3.0 is a workstation-based, stand-alone software designed as an optimized non-invasive application for cardiologists and radiologists to analyze coronary anatomy and pathology from Computed Tomography (CT) Angiographic images. Its primary purpose is to aid in viewing and analyzing cardiac CT data to determine the presence and extent of coronary plaques and luminal stenoses, thereby assisting in the determination of treatment paths. The software allows for standardized characterization of plaque and stenosis from DICOM images, featuring tools for reviewing coronary vessels in various Multiplanar Reformatting (MPR) views, measuring vessel and lumen dimensions, and quantifying plaque parameters. A key technological advancement in Autoplaque 3.0 is its deep learning-based automated segmentation of vessels, plaque, and lumen, which remains reviewable and editable by the clinician. Clinical validation studies demonstrated the performance of Autoplaque 3.0. Comparisons with expert reader measurements on a US patient population (201 patients, 781 lesions) and a non-US patient population (175 patients, 1081 lesions) showed excellent agreement for total plaque volume, calcified plaque volume, non-calcified plaque volume, and diameter stenosis. The software also showed significant efficiency, with an average per-patient plaque analysis time of 5.7 seconds in one study, compared to 25-30 minutes for manual expert analysis. Furthermore, a comparative study with its predicate device (30 lesions) confirmed excellent correlation and agreement for plaque and stenosis measurements, with Autoplaque 3.0 performing analysis much faster. The software is compatible with both Windows and Mac operating systems, but users must be appropriately trained and remain aware of limitations, such as the potential for interpolated data to obscure pathology near or smaller than scanning resolution.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. How does the deep learning algorithm in Autoplaque 3.0 specifically address the limitations of interpolated data, especially in detecting pathology smaller than scanning resolution?; What are the specific differences in performance and accuracy of Autoplaque 3.0 when running on Mac OS versus Windows OS, given the different operating system environments?; The related Lancet study (Appendix 7D) mentions a prognostic value for myocardial infarction based on plaque volume; how has this prognostic capability been validated or integrated within Autoplaque 3.0's intended use for aiding treatment path determination beyond anatomical analysis?; What training protocol is recommended for clinicians to become "appropriately trained" in Autoplaque 3.0's functions, capabilities, and particularly its limitations, including the interpretation of AI-assisted segmentations?; Given the variability in CT scanner models and acquisition protocols across clinical sites (as mentioned in the performance data summary), how does Autoplaque 3.0 ensure consistent and robust performance across this range of input data quality and characteristics?