Rapid RV/LV

The Rapid RV/LV software is a medical image processing tool that utilizes machine learning algorithms to automatically measure right and left ventricular diameters from adult CTPA scans, providing the RV/LV ratio as a quantitative metric. Its purpose is to assist clinicians in evaluating potential right-heart strain by offering consistent, automated measurements and annotated images, intended for adjunctive use with overall clinical assessment.

The Rapid RV/LV device, developed by iSchemaView Inc., is a radiological computer-assisted image processing software designed to aid medical professionals in assessing right-heart strain. Its primary purpose is to automatically measure the maximal diameters of the right and left ventricles from volumetric CT Pulmonary Angiography (CTPA) acquisitions in adult patients and report the ratio of these measurements (RV/LV ratio). The device operates as a module within the integrated Rapid Platform, utilizing machine learning algorithms to analyze DICOM-formatted CTPA images. These algorithms identify the relevant ventricular locations and perform the diameter measurements. The system then provides users with annotated images visually indicating where measurements were taken, along with quantitative results including individual ventricle diameters and their ratio, outputted in both visual and JSON text file formats. The performance of the Rapid RV/LV software was validated using a separate set of 124 CTPA cases, with ground truth established by three expert clinicians. This final validation demonstrated strong agreement with expert measurements, with a primary endpoint average slope of 1.1 (95% CI: 1.0, 1.2) and average intercept of -0.2 (95% CI: -0.1, -0.3) when comparing device measurements to expert consensus. A secondary Bland-Altman analysis showed a mean bias of 0.023 (95% CI: -0.04, 0.08), and the Mean Absolute Error was 3.8mm. The AI/ML model itself was developed using an initial dataset of 516 CTPA cases from diverse scanner types, reportedly utilizing 20% of these for training and 80% for validation during its development phase. The Rapid RV/LV software is intended to be used as an adjunctive tool, providing quantitative data to support clinical assessment, and its results are not designed for stand-alone diagnostic decision-making.

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 performance (Mean Absolute Error, bias) of the Rapid RV/LV software vary across different CT scanner manufacturers (Siemens, GE, Toshiba, Philips) and acquisition protocols not explicitly detailed in the training or validation sets?; What is the specific architecture of the machine learning algorithms used for ventricle segmentation and diameter measurement, and how was the reported split of the 516 development cases (20% for training, 80% for validation) precisely implemented during algorithm development?; Given the final validation results showing an average intercept of -0.2 (95% CI: -0.1, -0.3) when comparing device measurements to expert consensus, what is the clinical significance of this statistically significant systematic difference, and are there specific patient subgroups or measurement ranges where this bias might be more pronounced?; The device is indicated for use in adults; what are the primary challenges and potential modifications required for adapting this technology for pediatric populations, considering anatomical variations and differences in typical CTPA parameters?; How does the Rapid RV/LV software handle suboptimal image quality in CTPA scans (e.g., motion artifacts, poor contrast enhancement, variations in slice thickness), and what specific failure modes or confidence indicators are provided to the user in such scenarios?