Search Results

DASI Dimensions is a standalone, non-invasive, clinical decision support software solution that is intended for use by cardiologists and radiologists in context of the aortic stenosis population.

DASI Dimensions provides the end-user with pre-defined images and measurements (diameters, lengths, angles, areas, and perimeters) of cardiovascular structures.

The clinician receiving the results the responsibility for interpreting and validating all information and making all patient treatment decisions.

DASI Dimensions is not intended to replace the clinician's decision or device's instructions for use.

DASI Dimensions is prescription use only.

DASI Dimensions is an image post-processing software system intended for clinical decision support in the context of pre-procedural planning of Transcatheter Aortic Valve Replacement (TAVR) procedures. The software provides users with a report of generated dimensions of cardiac structures. DASI Dimensions software is not operated by physicians.

DASI Dimensions is an off-site software system that receives cardiologists input via upload of patient multiphase DICOM computed tomography angiography (CTA) chest image files through DASI Simulations web portal. The processed report is then available for viewing and downloading. The report is generated using proprietary algorithms that (a) detect key aortic root control points with the assistance of a static deep learning artificial intelligence (Al) model and (b) calculate anatomical measurements relevant for pre-TAVR evaluation. DASI Simulations engineers perform quality checks at both steps before releasing the report to the end user via the portal.

DASI Dimensions does not contact with the patient, nor does it control any life sustaining devices. The information provided by DASI Dimensions is not diagnostic, nor does it determine recommended medical care.

The cardiologists and/or radiologists receiving the responsibility for interpreting and validating all information and making all patient treatment decisions.

DASI Dimensions:

· Supports quantification of cardiac structures for pre-procedural planning of aotic stenosis patients in consideration for potential TAVR procedures.

· Provides the measurement of different structures of the heart, e.g., aortic root, aortic valve.

The DASI Dimensions (V1.0) device's acceptance criteria and the study proving it meets these criteria are outlined below.

1. Table of Acceptance Criteria and Reported Device Performance

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

• AI Control Point Detection Validation Study: The document does not specify the exact sample size for the "testing dataset" used in this study, only mentions "All testing, validation, and training cases were independent of each other."
• Automatic Measurements Validation Study: The document doesn't specify the sample size used for this study.
• Performance Validation Study (TP134 and TR134): Cohort of patient CT images (n=40).
• Operator Variability Study (TP136 and TR136): Dataset of CTAs (n=14).
• Control Point Sensitivity Study (TP140 and TR140): Dataset of CTAs (n=14).

Data Provenance:
"Relevant studies incorporated data with various distributions of ethnicity, gender, and age. All data was collected through the DASI Simulations secure server, AWS, through internal workstations operated by trained DASI Dimensions operators." The document implies these were retrospective data collected from clinical operations rather than prospective trials, as it explicitly states, "No prospective clinical trials were conducted in support of this Traditional 510(k)." The country of origin for the data is not explicitly stated.

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

• Number of Experts: "The reference standard was derived from 2 qualified truthers each CTA, whose measurements were averaged for each case."
• Qualifications of Experts: They are described as "qualified truthers" and "qualified DASI Dimensions operators (ground truth)" or "clinicians." Specific details like years of experience or board certification (e.g., radiologist, cardiologist) are not provided, but the context of "manual control points generated by qualified DASI Dimensions operators" suggests they are trained personnel in using such software and interpreting medical images.

4. Adjudication Method for the Test Set

• Adjudication Method: "If there was a significant variance between the initial two truthers, an adjudicator was involved." This indicates a 2+1 adjudication method.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study Was Done

No, a multi-reader multi-case (MRMC) comparative effectiveness study explicitly comparing human readers with and without AI assistance was not conducted. The studies focused on validating the device's measurements against ground truth established by human experts or the variability between operators using the device. The document explicitly states: "No prospective clinical trials were conducted in support of this Traditional 510(k)."

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

Yes, standalone performance was evaluated, particularly for the AI model:

• AI Control Point Detection Validation Study: "In the AI testing dataset the AI generated control points were compared to manual control points generated by qualified DASI Dimensions operators (ground truth)." This indicates a standalone evaluation of the AI's ability to detect control points.
• Automatic Measurements Validation Study: This study compared the "automatic measurement outputs generated using control points" (which were presumably AI-generated or system-computed) to manual measurements by clinicians. This also reflects a standalone assessment of the automated measurement capabilities.

It's important to note that the device's overall workflow involves "DASI Simulations engineers perform quality checks at both steps before releasing the report to the end user," and it is described as a "semi-automatic AI enhanced workflow." However, the validation studies described, particularly the AI Control Point Detection and Automatic Measurements, evaluate the software's automated functionalities independently against a human-established ground truth.

7. The Type of Ground Truth Used

The ground truth used was expert consensus / expert manual measurements.

• "The reference standard was derived from 2 qualified truthers each CTA, whose measurements were averaged for each case. If there was a significant variance between the initial two truthers, an adjudicator was involved."
• "Al generated control points were compared to manual control points generated by qualified DASI Dimensions operators (ground truth)."
• "Automatic measurement outputs ... compared to the measurements generated manually by clinicians using current standard of care methods."
• "Outputs were compared to qualified clinician truther generated control points."

8. The Sample Size for the Training Set

The document mentions "training, validation, and testing datasets" for the AI model (W112 and TR132) but does not explicitly state the sample size for the training set. It only clarifies that, "All testing, validation, and training cases were independent of each other and were not used in any other validation studies."

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, but it is highly probable that it was established using a similar method to the test set, i.e., through qualified human experts (truthers/operators/clinicians) making manual annotations and measurements, potentially with adjudication for discrepancies. This is inferred from the overall approach to ground truth described for the validation studies, where human expert input is consistently used as the reference standard.

Ask a specific question about this device