Search Results

3DXR is the AW application which is intended to perform the three-dimensional (3D) reconstruction computation of any images acquired with a 3D Acquisition mode of the X-ray interventional system for visualization under Volume Viewer. The 3D Acquisition modes are intended for imaging vessels, bones and soft tissues as well as other internal body structures. The 3D reconstructed Volume assist the physician in diagnosis, surgical planning, Interventional procedures and treatment follow-up.

3DXR is a post-processing software-only application, runs on Advantage Workstation (AW) platform [K110834], and performs 3D reconstruction for the CBCT 3D acquisition images (input) acquired from the fixed interventional X-ray system [K181403, K232344]. The reconstructed 3D volume (output) is visualized under Volume Viewer application [K041521]. The proposed 3DXR is a modification from the predicate 3DXR [included and cleared in K181403]. A new option, called CleaRecon DL, based on Deep-Learning (DL) technology, is added in the proposed subject 3DXR application.

Here's a breakdown of the acceptance criteria and the study proving the device meets them, based on the provided text:

1. A table of acceptance criteria and the reported device performance

2. Sample size used for the test set and the data provenance (e.g. country of origin of the data, retrospective or prospective)

• Engineering Bench Testing:
  • Test Set 1 (Image Pairs): Two samples (number not specified beyond "two samples").
  • Test Set 2 (Phantoms): Not specified beyond "phantoms."
• Clinical Image Quality Evaluation (Retrospective):
  • Sample Size: 110 independent exams, each from a unique patient.
  • Data Provenance: Retrospectively collected from 13 clinical sites.
    • 80 patients from the US
    • 26 patients from France
    • 4 patients from Japan
  • Patient Population: Adult patients (pediatrics excluded) undergoing interventional procedures. No inclusion criteria for age (within adult range) or sex/gender (except for prostate).

3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts (e.g. radiologist with 10 years of experience)

• Engineering Bench Testing: Not applicable; ground truth was intrinsic (images with and without applying CleaRecon DL, or phantoms with reference).
• Clinical Image Quality Evaluation:
  • Number of Experts: 13 clinicians.
  • Qualifications: "Clinicians" (specific specialties or years of experience are not mentioned, but their role implies expertise in image interpretation for interventional procedures).

4. Adjudication method (e.g. 2+1, 3+1, none) for the test set

• Clinical Image Quality Evaluation: Each of the 110 exams (with and without CleaRecon DL) was compared/evaluated at least 3 times independently by the recruited clinicians. This resulted in 490 pairs of clinicians' evaluations. This suggests a multi-reader, independent review with subsequent aggregation of results, rather than a formal consensus-based adjudication like 2+1 or 3+1 for individual cases, as the "489 reviews" and "98% of cases" suggest aggregated findings.

5. If a multi-reader multi-case (MRMC) comparative effectiveness study was done, If so, what was the effect size of how much human readers improve with AI vs without AI assistance

• A multi-reader multi-case (MRMC) study was implicitly conducted as part of the clinical image quality evaluation, with 13 clinicians reviewing 110 cases (though the "multi-case" aspect is strong, the "multi-reader" aspect is also present for each case, as each was reviewed at least 3 times independently).
• Effect Size: The study focused on the impact of the image quality on interpretation, rather than a direct measure of human reader performance improvement in diagnostic accuracy or efficiency with and without AI assistance. The results indicated:
  • "In 98% of the cases, the CBCT images reconstructed with CleaRecon DL option are evaluated as clearer than the conventional CBCT images."
  • Clinicians were asked to assess "how it impacts their confidence in image interpretation," but the specific effect size or improvement in confidence wasn't quantified.
  • No hidden or artificially created structures were identified, indicating perceived safety and reliability.
  • Therefore, while it showed a significant improvement in perceived image clarity, it did not provide a quantitative effect size for human reader improvement (e.g., in AUC, sensitivity, or specificity) with AI assistance.

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

• Yes, a standalone performance evaluation of the algorithm was done as part of the "Engineering bench testing." This involved:
  • Testing on a segregated test dataset of image pairs (with and without CleaRecon DL) where Mean Absolute Error (MAE) and Structural Similarity Index Measure (SSIM) were computed.
  • Testing on phantoms where MAE and Standard Deviation (SD) were computed relative to a reference (without simulated artifacts).
  • These tests directly assessed the algorithm's capability to reduce artifacts and improve image metrics without human interaction.

7. The type of ground truth used (expert consensus, pathology, outcomes data, etc)

• Engineering Bench Testing:
  • For the image pair comparison, the ground truth was essentially the "ideal" or "less artifact-laden" image derived from the paired comparison. This is a form of reference-based comparison where the output without the artifact or with the artifact corrected is the standard.
  • For phantoms, the ground truth was the known characteristics of the phantom (e.g., absence of artifacts in the reference image).
• Clinical Image Quality Evaluation:
  • The ground truth was established through expert evaluation/consensus (13 clinicians evaluating side-by-side images). However, it was focused on subjective image quality and the presence/absence of artifacts, rather than ground truth for a specific diagnosis or outcome.

8. The sample size for the training set

• The text states: "The CleaRecon DL algorithm was trained and qualified using pairs of images with and without streak artifacts." However, the specific sample size of the training set is not provided.

9. How the ground truth for the training set was established

• The ground truth for the training set was established using "pairs of images with and without streak artifacts." This implies that for each image with streak artifacts, there was a corresponding reference image without such artifacts, which allowed the algorithm to learn how to remove them. The method by which these "pairs of images" and their respective "with/without streak artifacts" labels were generated or confirmed is not detailed. It could involve expert labeling, simulation, or other methods.

Ask a specific question about this device

BreView is a tool to aid clinicians with the review of multi-parametric breast magnetic resonance (MR) images. The combination of acquired images, reconstructed images, annotations, and measurements performed by the clinician are intended to provide the referring physician with clinically relevant information that may aid in diagnosis and treatment planning.

BreView is a dedicated advanced visualization review and post-processing tool that facilitates optimized post-processing of MR breast data review and assessment workflows for radiologists - including organizing images and composing reports. Adding the techniques of automatic motion correction and subtraction improves the review process. Software functionalities include:

• Ability to load 2D, 3D and 4D MR datasets as well as DICOM Secondary Captures (SCPT) .
• Optimized display of multi-parametric images within a dedicated layout
• Display customization: ability to choose layout, orientation, rendering mode
• Guided workflows for reviewing and processing MR breast exams
• Automated motion correction and/or subtraction of multi-phase datasets
• . Multi-planar reformats and maximum intensity projections (MIPs)
• Semi-automated segmentation and measurement tools
• Graph view for multi-phase datasets
• Save and export capabilities though DICOM Secondary Captures
• Data export in the form of a summary table

The provided text does not contain detailed information about specific acceptance criteria or a comprehensive study that proves the device meets acceptance criteria. The document is primarily a 510(k) premarket notification, focusing on demonstrating substantial equivalence to a predicate device rather than presenting a detailed performance study with specific acceptance metrics.

However, based on the type of information typically expected for such a submission and the limited details provided, I can infer some aspects and highlight what is missing.

Here's an analysis based on the provided text, structured to address your questions as much as possible, while also noting where information is explicitly not present:

Context: The device, "BreView," is a medical diagnostic software intended to aid clinicians in reviewing multi-parametric breast MR images. Its core functionalities include image organization, display customization, automated motion correction and/or subtraction, multi-planar reformats, semi-automated segmentation and measurement tools, graph view, and data export. It is marketed as a Class II device.

1. Table of Acceptance Criteria and Reported Device Performance

Critique: The document does not provide a specific table of acceptance criteria with corresponding performance metrics. It mentions "Performance testing (Verification, Validation)" and "Bench testing" but lacks quantifiable results against predefined acceptance thresholds.

However, we can infer some criteria from the "Comparison" table against the predicate device, although these are qualitative comparisons rather than quantitative performance data.

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

• Sample Size for Test Set: Not specified. The document mentions "Bench testing was conducted on a database of exams representative of the clinical scenarios where BreView is intended to be used, with consideration of acquisition protocols and clinical indicators." However, no numerical sample size (e.g., number of cases, number of images) is provided for this test set.
• Data Provenance: Not specified. There is no mention of the country of origin of the data or whether the data was retrospective or prospective.

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

• Number of Experts: Not specified.
• Qualifications of Experts: Not specified. The document mentions "End user evaluation of the brush," implying radiologists or similar clinical professionals, but their specific qualifications or number are not detailed.

4. Adjudication Method for the Test Set

• Adjudication Method: None stated. The document does not describe any specific adjudication method (e.g., 2+1, 3+1 consensus) used for establishing ground truth or evaluating performance.

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

• MRMC Study Conducted: No. The document does not describe a formal MRMC study where human readers' performance with and without AI assistance was evaluated. The evaluation appears to be primarily focused on the device's technological equivalence rather than its impact on human reader effectiveness.
• Effect Size of Improvement: Not applicable, as no MRMC study was described.

6. Standalone (Algorithm Only Without Human-in-the-Loop) Performance

• Standalone Performance Evaluation: Implied, but not explicitly quantified with metrics. The statement, "BreView Non-Rigid Registration algorithm was tested and found to be successfully implemented," suggests a standalone evaluation of this specific algorithm. However, no quantitative performance metrics (e.g., accuracy, precision, dice score for segmentation, registration error) are provided for any standalone algorithm. The "equivalence or better" statement for the Smart Brush tool is a comparative qualitative assessment based on "End user evaluation," not a formal standalone study result.

7. Type of Ground Truth Used

• Type of Ground Truth: The document does not explicitly define how ground truth was established for "bench testing." For functionalities like non-rigid registration and image subtraction, the "ground truth" might be implied by successful implementation against a known good algorithm (the predicate's method). For the "Smart Brush tool," the ground truth seems to be based on expert (end-user) evaluation comparing its behavior to the predicate's tool. There's no mention of pathology, clinical outcomes data, or a formal consensus process.

8. Sample Size for the Training Set

• Sample Size for Training Set: Not specified. The document only mentions "Bench testing was conducted on a database of exams representative of the clinical scenarios," which refers to a test set, not a training set. Given that this is a 510(k) submission and the device is presented as applying "the same fundamental scientific technology as its predicate device," it is possible that new training data (in the context of deep learning) was not a primary focus of this submission, if the core algorithms are established or rule-based.

9. How Ground Truth for the Training Set Was Established

• Ground Truth Establishment for Training Set: Not applicable / Not specified. Since no training set or related machine learning components requiring explicit ground truth labeling were discussed in detail, this information is not provided.

Ask a specific question about this device

FlightPlan for Embolization is a post processing software package that helps the analysis of 3D X-ray angiography images. Its output is intended to be used by physicians as an adjunct means to help visualize vasculature during the planning phase of embolization procedures. FlightPlan for Embolization is not intended to be used during therapy delivery.

The output includes segmented vasculature, and selective display of proximal vessels from a reference point determined by the user. User-defined data from the 3D X-ray angiography images may be exported for use during the guidance phase of the procedure. The injection points should be confirmed independently of FlightPlan for Embolization prior to therapy delivery.

FlightPlan for Embolization is a post-processing, software-only application using 3D X-ray angiography images (CBCT) as input. It helps clinicians visualize vasculature to aid in the planning of endovascular embolization procedures throughout the body.

A new option, called AI Segmentation, was developed from the modifications to the predicate device, GE HealthCare's FlightPlan for Embolization [K193261]. It includes two new algorithms. This Al Segmentation option is what triggered this 510(k) submission.

The software process 3D X-ray angiography images (CBCT) acquired from GE HealthCare's interventional X-ray system [K181403], operates on GEHC's Advantage Workstation (AW) [K110834] platform and AW Server (AWS) [K081985] platform, and is an extension to the GE HealthCare's Volume Viewer application [K041521].

FlightPlan for Embolization is intended to be used during the planning phase of embolization procedures.

The primary features/functions of the proposed software are:

• Semi-automatic segmentation of vasculature from a starting point determined by the user, when AI Segmentation option is not activated;
• Automatic segmentation of vasculature powered by a deep-learning algorithm, when Al Segmentation option is activated;
• Automatic definition of the root point powered by a deep-learning algorithm, when AI Segmentation option is activated;
• Selective display (Live Tracking) of proximal vessels from a point determined by the user's cursor;
• Ability to segment part of the selected vasculature;
• Ability to mark points of interest (POI) to store cursor position(s);
• Save results and optionally export them to other applications such as GEHC's Vision Applications ● [K092639] for 3D road-mapping.

Here's a breakdown of the acceptance criteria and the study details for the GE Medical Systems SCS's FlightPlan for Embolization device, based on the provided text:

Acceptance Criteria and Device Performance

Study Details

1. Sample Size Used for the Test Set and Data Provenance:

• Test Set Sample Size: 207 contrast-injected CBCT scans, each from a unique patient.
• Data Provenance: The scans were acquired during the planning of embolization procedures from GE HealthCare's interventional X-ray system. The text indicates that these were from "clinical sites" and were "representative of the intended population" but does not specify countries of origin. The study appears to be retrospective, using existing scans.

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

• Vessel Extraction: 3 board-certified radiologists.
• Root Definition: 1 GEHC advanced application specialist.

3. Adjudication Method for the Test Set:

• Vessel Extraction: Consensus of 3 board-certified radiologists. (Implies a qualitative agreement, not a specific quantitative method like 2+1).
• Root Definition: The acceptable area was manually defined by the annotator (the GEHC advanced application specialist).

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

• No, an MRMC comparative effectiveness study was not explicitly described in terms of human readers improving with AI vs. without AI assistance. The non-clinical testing focused on the algorithms' performance against ground truth and the clinical assessment used a Likert scale to evaluate the proposed device with the AI option, rather than a direct comparison of human reader performance with and without AI.

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

• Yes, a standalone performance evaluation was conducted for both the vessel extraction and root definition algorithms. The reported success rates of 93.7% and 95.2% are measures of the algorithms' performance against established ground truth without a human in the loop for the primary performance metrics.

6. The Type of Ground Truth Used:

• Vessel Extraction: Expert consensus (3 board-certified radiologists).
• Root Definition: Manual definition by an expert (GEHC advanced application specialist), defining an "acceptable area."

7. The Sample Size for the Training Set:

• The document states that "contrast injected CBCT scans acquired from GE HealthCare's interventional X-ray system [K181403] were used for designing and qualifying the algorithms." However, it does not specify the sample size for the training set. It only mentions that a test set of 207 scans was "reserved, segregated, and used to evaluate both algorithms."

8. How the Ground Truth for the Training Set Was Established:

• The document does not explicitly state how the ground truth for the training set was established. It only details the ground truth establishment for the test set.

Ask a specific question about this device

FlightPlan for Liver is a post processing software package that helps the analysis of 3D X-ray angiography images of the liver. Its output is intended as an adjunct means to help visualize vasculature and identify arteries leading to the vicinity of hypervascular lesions in the liver. This adjunct information may be used by physicians to aid them in their evaluation of hepatic arterial anatomy during the planning phase of embolization procedures.

FlightPlan for Liver with the Parenchyma Analysis option is a post-processing, software-only application using 3D X-ray angiography images (CBCT) as input. It helps physicians visualize and analyze vasculature to aid in the planning of endovascular embolization procedures in the liver. It was developed from modifications to the predicate device, GE's FlightPlan for Liver [K121200], including the addition of 2 new algorithms supporting the Parenchyma Analysis option. The Parenchyma Analysis option is what triggered this 510k. The subject device also includes a feature, Live Tracking, that was cleared in the reference device, GE's FlightPlan for Embolization. The software operates on GE's Advantage Workstation [K110834] platform and AW Server [K081985] platform and is an extension to the GE's Volume Viewer application [K041521].

Here's a breakdown of the acceptance criteria and study information for FlightPlan for Liver, based on the provided text:

1. Table of Acceptance Criteria and Reported Device Performance

The document states that "The test results of both of the algorithms met their predefined acceptance criteria" for the deep learning-based Liver Segmentation algorithm and the non-deep learning Virtual Parenchyma Visualization (VPV) algorithm. However, the specific quantitative acceptance criteria and their corresponding reported performance values are not explicitly detailed in the provided text.

The clinical assessment also "demonstrated that the proposed device FlightPlan for Liver with the Parenchyma Analysis option met its predefined acceptance criteria," but again, the specifics are not provided.

Therefore, a table cannot be fully constructed without this missing information.

Missing Information:

• Specific quantitative acceptance criteria for Liver Segmentation algorithm (e.g., Dice score, IoU, boundary distance).
• Specific quantitative reported performance for Liver Segmentation algorithm.
• Specific quantitative acceptance criteria for VPV algorithm (e.g., accuracy of distal liver region estimation).
• Specific quantitative reported performance for VPV algorithm.
• Specific quantitative or qualitative acceptance criteria for the clinical assessment using the 5-point Likert scale.

2. Sample Sizes and Data Provenance

For the Deep Learning Liver Segmentation Algorithm:

• Test Set Sample Size: Not explicitly stated, but derived from a "database of contrast injected CBCT liver acquisitions."
• Data Provenance: Clinical sites in the USA and France. The data was prospective (implied by "clinical sites") and for the purpose of training and testing.

For the Non-Deep Learning Virtual Parenchyma Visualization (VPV) Algorithm:

• Test Set Sample Size: "a test set of proximal CBCT acquisitions." The exact number is not provided.
• Data Provenance: From the USA and France.

For the Clinical Testing:

• Test Set Sample Size: "A sample of 3D X-ray angiography image pairs." The exact number is not provided.
• Data Provenance: From France and the USA.

3. Number of Experts and Qualifications for Ground Truth

For the Deep Learning Liver Segmentation Algorithm:

• Number of Experts: Not specified.
• Qualifications: Not specified. The ground truth method itself (how segmentation was established for training and testing) is not fully detailed beyond using existing "contrast injected CBCT liver acquisitions."

For the Non-Deep Learning Virtual Parenchyma Visualization (VPV) Algorithm:

• Number of Experts: Not specified.
• Qualifications: Not specified.
• Ground Truth Basis: "selective contrast injected CBCT exams from same patients." It's implied that these were expert-interpreted or based on a recognized clinical standard, but the specific expert involvement is not detailed.

For the Clinical Testing:

• Number of Experts: Not specified.
• Qualifications: "interventional radiologists." No experience level (e.g., years of experience) is provided.

4. Adjudication Method for the Test Set

The document does not explicitly mention an adjudication method (like 2+1 or 3+1 consensus) for any of the test sets.

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

• The document describes a "Clinical Testing" section where "interventional radiologists using a 5-point Likert scale" assessed image pairs. This suggests a reader study.
• However, it does not explicitly state that it was an MRMC comparative effectiveness study comparing human readers with AI assistance vs. without AI assistance. The study assessed whether the device "met its predefined acceptance criteria and helps physicians visualize and analyze... and aids in the planning..." which seems to be an evaluation of the device's utility rather than a direct comparison of reader performance with and without the AI.
• Therefore, no effect size of human readers improving with AI vs. without AI assistance is reported.

6. Standalone (Algorithm Only) Performance

• Yes, standalone performance was done for both new algorithms.
  • The "deep learning-based Liver Segmentation algorithm" was evaluated, although specific performance metrics are not given.
  • The "non-deep learning Virtual Parenchyma Visualization algorithm's performance was evaluated [...] compared to selective contrast injected CBCT exams from same patients used as the ground truth."
• This indicates that the algorithms themselves were tested for their inherent performance.

7. Type of Ground Truth Used

For the Deep Learning Liver Segmentation Algorithm:

• Based on "contrast injected CBCT liver acquisitions." The precise method for establishing the "correct" segmentation (e.g., manual expert tracing, pathology, outcome data) is not detailed. It's implicitly clinical data.

For the Virtual Parenchyma Visualization (VPV) Algorithm:

• "selective contrast injected CBCT exams from same patients used as the ground truth." This implies a gold standard of directly observed, selective angiography, which is a clinical reference.

For the Clinical Testing:

• The "ground truth" here is the perception and evaluation of interventional radiologists using a 5-point Likert scale regarding the device's utility ("helps physicians visualize," "aids in the planning"). This is a form of expert consensus/subjective assessment of utility.

8. Sample Size for the Training Set

For the Deep Learning Liver Segmentation Algorithm:

• "a database of contrast injected CBCT liver acquisitions from clinical sites in the USA and France was used for the training and testing." The exact sample size for training is not specified, only that a database was used.

For the Virtual Parenchyma Visualization (VPV) Algorithm:

• The VPV algorithm is described as "non-deep learning," so it would not have a traditional "training set" in the same way a deep learning model would. It likely relies on predefined anatomical/physiological models or rules.

9. How Ground Truth for the Training Set Was Established

For the Deep Learning Liver Segmentation Algorithm:

• The document states "a database of contrast injected CBCT liver acquisitions [...] was used for the training." However, it does not explicitly detail how the ground truth labels (i.e., the correct liver segmentations) were established for these training images. This is a critical piece of information for a deep learning model. It's usually through expert manual annotation for segmentation tasks.

Ask a specific question about this device

Hepatic VCAR is a CT image analysis software package that allows the analysis and visualization of Liver CT data derived from DICOM 3.0 compliant CT scans. Hepatic VCAR is designed for the purpose of assessing liver morphology, including liver lesion, provided the lesion has different CT appearance from surrounding liver tissue; and its change over time through automated tools for liver lobe, liver segments and liver lesion segmentation and measurement. It is intended for use by clinicians to process, review, archive, print and distribute liver CT studies.

This software will assist the user by providing initial 3D segmentation, visualization, and quantitative analysis of liver anatomy. The user has the ability to adjust the contour and confirm the final segmentation.

Hepatic VCAR is a CT image analysis software package that allows the analysis and visualization of Liver CT data derived from DICOM 3.0 compliant CT scans. Hepatic VCAR was designed for the purpose of assessing liver morphology, including liver lesion, provided the lesion has different CT appearance from surrounding liver tissue; and its change over time through automated tools for liver, liver lobe, liver segments and liver lesion segmentation and measurement.

Hepatic VCAR is a post processing software medical device built on the Volume Viewer (K041521) platform, and can be deployed on the Advantage Workstation (AW) (K110834) and AW Server (K081985) platforms, CT Scanners, and PACS stations or cloud in the future.
This software will assist the user by providing initial 3D segmentation, vessel analysis, visualization, and quantitative analysis of liver anatomy. The user has the ability to adjust the contour and confirm the final segmentation.
In the proposed device, two new algorithms utilizing deep learning technology were introduced. One such algorithm segments the liver producing a liver contour editable by the user; another algorithm segments the hepatic artery based on an initial user input point. The hepatic artery segmentation is also editable by the user.

The provided text describes the 510(k) summary for Hepatic VCAR, a CT image analysis software package. The submission outlines the device's intended use and the validation performed, particularly highlighting the introduction of two new deep learning algorithms for liver and hepatic artery segmentation.

Here's an analysis of the acceptance criteria and study proving the device meets them, based on the provided text:

1. Table of Acceptance Criteria and Reported Device Performance

The document doesn't explicitly define "acceptance criteria" through numerical thresholds for performance metrics. Instead, it states that "Verification and validation including risk mitigations have been executed with results demonstrating Hepatic VCAR met the design inputs and user needs with no unexpected results or risks."

For the new deep learning algorithms, the performance is described qualitatively:

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

• Test Set Sample Size: "A representative set of clinical sample images" was used for the clinical assessment. The exact number of cases/images is not specified in the provided text.
• Data Provenance: The provenance of the data (e.g., country of origin, retrospective or prospective) is not specified.

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

• Number of Experts: For the clinical assessment of the deep learning algorithms, 3 board certified radiologists were used.
• Qualifications of Experts: They are described as "board certified radiologists." The number of years of experience is not specified.
• For the "ground truth" used in "bench tests," the text states "ground truth annotated by qualified experts," but the number and specific qualifications of these experts are not explicitly detailed beyond "qualified experts."

4. Adjudication Method for the Test Set

The text states that the "representative set of clinical sample images was assessed by 3 board certified radiologists using 5-point Likert scale." It does not specify an explicit adjudication method (e.g., 2+1, 3+1 consensus) for establishing the "ground truth" or assessing the device's performance based on the radiologists' Likert scale ratings. The Likert scale assessment sounds more like an evaluation of clinical acceptability/usability rather than establishing ground truth for quantitative segmentation accuracy.

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

A formal MRMC comparative effectiveness study comparing human readers with AI assistance vs. without AI assistance is not explicitly described in the provided text. The clinical assessment mentioned ("assessed by 3 board certified radiologists using 5-point Likert scale") appears to be an evaluation of the device's capability rather than a direct comparison of human performance with and without AI assistance. Therefore, no effect size for human reader improvement is provided.

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

Yes, standalone performance was assessed for the algorithms. The text states:

• "Bench tests that compare the output of the two new algorithms with ground truth annotated by qualified experts show that the algorithms performed as expected."
  This indicates an evaluation of the algorithm's direct output against an established ground truth before human interaction/adjustment.

7. The Type of Ground Truth Used

Based on the document:

• For the "bench tests" of the new deep learning algorithms, the ground truth was "ground truth annotated by qualified experts." This suggests expert consensus or expert annotation was used.
• For the clinical assessment by 3 radiologists using a Likert scale, it's more of a qualitative assessment of the device's capability rather than establishing a definitive ground truth for each case.

8. The Sample Size for the Training Set

The sample size for the training set (used to train the deep learning algorithms) is not specified in the provided text.

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

The text states that the deep learning algorithms were trained, but it does not explicitly describe how the ground truth for the training set was established. It only mentions that the ground truth for bench tests was "annotated by qualified experts."

Ask a specific question about this device

CardIQ Flow is an aiding tool for the clinicians to perform analysis of stationary, dynamic, or gated cardiac position emission tomography (PET) images. The measurements include perfusion, end-diastolic volumes, stroke volume, ejection fraction, myocardial mass, and transient ischemic dilatation. Also included is a feature that enables clinicians to visualize reformatted data and make a comparison between stress and rest series. Software is indicated for determining regional and global absolute rest and stress myocardial perfusion in cc/min/g, Coronary Flow Reserve and their combination into the Coronary (CFC) Map in patients with suspected or known coronary artery disease (CAD) in order to assist clinical interpretation of PET perfusion images by quantification of their severty. It is intended for use by trained professionals, such as nuclear medicine or nuclear cardiology physicians, or cardiologists with appropriate training. The clinician remains ultimately responsible for the final assessment and diagnosis based on standard practices, clinical judgment and interpretation of cardiac images or quantitative data.

The basis for this submission is the introduction of a new software device, CardIQ Flow which integrates the functionalities of the predicate devices HeartSee and FullCard Analysis for the display and analysis of Cardiac PET images. It is built within the framework of the Volume Viewer (K041521) 3D platform, intended to aid clinicians in assessing the myocardial perfusion, myocardial perfusion, myocardial blood flow and wall motion. It is used for the evaluation of patients with known or suspected coronary artery or ischemic heart disease.

This application can be used for:

· Reformatting axial data to the standard axis of the heart. Axial data may have been reconstructed using iterative or FBP algorithms with scatter and attenuation correction, if available.

• Analyzing the myocardial perfusion by the Vertical Long Axis (VLA), Horizontal Long Axis (HLA) and Short Axis (SA) obliques generated from the selected input data.

· Comparing myocardial perfusion in rest and stress studies as well as metabolism in viability studies.

• · Viewing wall motions via synchronized gated beating slices.
• · Investigating the entire left ventricle in one 2D Polar Plot that displays all short axis slices.
• · Comparing myocardial perfusion of corrected and uncorrected datasets.

· Calculation and evaluation of Myocardial Blood Flow (MBF), Coronary Flow Reserve (CFR), and Coronary Flow Capacity (CFC).

· Printing a report designed for referring physicians.

Here's a breakdown of the acceptance criteria and study information for CardIQ Flow based on the provided document:

1. Acceptance Criteria and Reported Device Performance:

The document doesn't explicitly state quantitative acceptance criteria for device performance. Instead, it focuses on the device performing "substantially equivalent" to its predicates and successfully completing design control activities. The technological comparison table highlights similarities in workflow and measurements.

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

The document does not specify the sample size used for the test set or the data provenance (e.g., country of origin, retrospective/prospective). It generally refers to "performance testing (Verification, Validation)" but does not provide details about the datasets used in these tests.

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

The document does not provide information on the number of experts used or their qualifications for establishing ground truth for any test sets.

4. Adjudication Method for the Test Set:

The document does not specify any adjudication method used for the test set.

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

The document does not mention a multi-reader multi-case (MRMC) comparative effectiveness study, nor does it quantify any effect size of human readers improving with AI vs. without AI assistance. The focus is on demonstrating substantial equivalence to existing devices, not on comparing performance with and without AI assistance for human readers.

6. Standalone (Algorithm Only Without Human-in-the-Loop Performance) Study:

The document implies a standalone performance assessment through "performance testing (Verification, Validation)" and by stating the device "performs substantially equivalent to the predicate devices." The nature of the comparison to predicates (HeartSee and FullCard Analysis, which are also software tools for analysis) suggests that CardIQ Flow's computational and analytical outputs were directly compared. However, it does not explicitly detail a dedicated "standalone" study separate from human interaction, but rather the software's ability to produce quantitative measurements and visualizations.

7. Type of Ground Truth Used:

The document does not explicitly state the type of ground truth used for performance testing (e.g., expert consensus, pathology, outcomes data). Given the device's function in quantifying myocardial perfusion and other cardiac parameters based on PET images, it's highly probable that ground truth would involve comparisons to established clinical methods, potentially expert interpretation, or potentially phantom studies for validation of quantitative accuracy. However, this is not detailed.

8. Sample Size for the Training Set:

The document does not provide information on the sample size used for the training set.

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

The document does not provide information on how the ground truth for the training set was established.

Ask a specific question about this device

FlightPlan for Embolization is a post processing software package that helps the analysis of 3D X-ray angiography images. Its output is intended to be used by physicians as an adjunct means to help visualize vasculature during the planning phase of embolization procedures. FlightPlan for Embolization is not intended to be used during therapy delivery.

The output includes segmented vasculature, and selective display of proximal vessel and distal vessels from a reference point determined by the user. User-defined data from the 3D X-ray angiography images may be exported for use during the guidance phase of the procedure. The injection points should be confirmed independently of FlightPlan for Embolization prior to therapy delivery.

FlightPlan for Embolization is a post processing software application which operates on the Advantage Workstation (AW) [K110834] platform and AW Server [K081985] platform. It is an extension to the Volume Viewer application [K041521] modified from FlightPlan for Liver (K121200) and is designed for processing 3D X-ray angiography images to help visualize vasculature

The primary features of the software are: semi-automatic segmentation of vascular tree from a starting point determined by the user; selective display (Live Tracking) of proximal vessel and distal vessels from a point determined by the user's cursor; ability to segment part of the vasculature; ability to mark points of interest (POI) to store cursor position; save results and export to other applications such as Vision Applications [K092639] for 3D road-mapping.

Here's a breakdown of the acceptance criteria and the study proving the device meets them, based on the provided FDA 510(k) summary for FlightPlan for Embolization:

1. Table of Acceptance Criteria and Reported Device Performance

The provided document does not explicitly list quantitative acceptance criteria in a dedicated section for "acceptance criteria." However, it describes validation activities that implicitly define the performance considered acceptable. Based on the "Summary of Non-Clinical Tests" and "Summary of Clinical Tests," the following can be inferred:

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

• Non-Clinical Test Set:

  • Sample Size: A "database of XACT exams." The specific number of cases is not provided.
  • Data Provenance: Not explicitly stated, but clinical scenarios are considered "representative" and include "consideration of acquisition parameters, image quality and anatomy." It can be inferred that these are existing clinical data, likely retrospective.

• Clinical Test Set:

  • Sample Size: "A sample of 3D X-ray angiography images representative of clinical practice." The specific number of cases is not provided.
  • Data Provenance: Not explicitly stated, but described as "representative of clinical practice" and "most common anatomic regions where embolization procedures are performed." It can be inferred that these are retrospective clinical cases.

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

• Clinical Test Set:
  • Number of Experts: Four
  • Qualifications: "board certified interventional radiologists." No information on years of experience is provided.

4. Adjudication Method for the Test Set

• Clinical Test Set: The document states that the assessment was done "using a 5-point Likert scale." This implies an individual scoring system by each radiologist, but it does not specify an adjudication method (e.g., 2+1, 3+1 consensus). It sounds like individual assessments were performed and then aggregated or analyzed.

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

• No, an MRMC comparative effectiveness study was not explicitly stated to have been done in the traditional sense of comparing human readers with AI vs. without AI assistance.
  • The clinical study described is an assessment of the device's helpfulness by radiologists, rather than a comparative study measuring improvement in diagnostic accuracy or efficiency for humans using the AI vs. not using it. The statement "demonstrated that the proposed device (FlightPlan for Embolization) helps physicians" is an outcome of an assessment, not a quantitative effect size from an MRMC study comparing assisted vs. unassisted performance.

6. If a Standalone (Algorithm Only Without Human-in-the-Loop Performance) was done

• Yes, a form of standalone performance was assessed in the "Summary of Non-Clinical Tests." The "Engineering...validated FlightPlan for Embolization algorithms' capability to automatically segment and selectively display vascular structures..." using a database of XACT exams. This implies an evaluation of the algorithm's output against some reference (likely manual segmentations or ground truth established by experts) without direct human interaction at the time of assessment.

7. The Type of Ground Truth Used

• Non-Clinical Test Set: The ground truth for the non-clinical validation of segmentation and display capabilities is implicitly based on expert-defined correct segmentations or "satisfactory quality" as determined by engineering validation. The document does not explicitly state "expert consensus" or "pathology" for this.
• Clinical Test Set: The ground truth for the clinical assessment relies on the judgment of "four board certified interventional radiologists" using a 5-point Likert scale to determine if the device "helps physicians." This is a form of expert assessment/opinion as ground truth regarding the device's utility/helpfulness.

8. The Sample Size for the Training Set

• Not provided. The document does not disclose information about the training data used for the FlightPlan for Embolization algorithms.

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

• Not provided. Since the training set details are not mentioned, how its ground truth was established is also not available in this document.

Ask a specific question about this device

Vision 2, TrackVision 2 and HeartVision 2 software applications are intended to enable users to load 3D datasets and overlay and register in real time these 3D datasets with radioscopic or radiographic images of the same anatomy in order to support catheter/device guidance during interventional procedures.

The Stereo 3D option enables physicians to visualize needles, points, and segments on a 3D model/space using a stereotaction of radioscopic or radiographic images at a significantly lower dose than use of a full cone beam CT acquisition. This information is intended to assist the physician during interventional procedures.

Vision Applications (K092639) includes Vision 2, TrackVision 2 and HeartVision 2 applications. Vision Applications can load and dynamically fuse in real-time live 2D X-ray images from the X-ray system with 3D models from X-Ray (DICOM 3D XA), CT or MR system.

Stereo 3D is a new option and the subject of this submission. The Stereo 3D option is designed to be used with the Vision 2 and TrackVision 2 applications which are part of Vision Applications. The Stereo 3D option enables the user to reconstruct 3D objects from radioscopic or radiographic images.

The Stereo 3D option intends to provide an alternative to intra operative CBCT (cone beam CT) usually performed for the same purpose: to localize needles and markers within the 3D anatomy. The Stereo 3D option provides a method to reconstruct 3D contrasted objects (point and segments) from a pair of 2D X-ray images, e.g. acquisition of fluoroscopic images from different C-arm positions (2 different projections). The 3D object reconstruction is then fused in the 3D space with the 3D model used at the fusion of the x-ray image.

Stereo 3D has a workflow that is significantly guided, to support clear and easy use of the reconstruction procedure. The workflow contains the following 4 high level steps:
Image acquisition and registration adjustment
Automatic or manual object identification
Quality assessment of 3D reconstruction
Display of the reconstructed point(s) and segment(s) on a cross section of the 3D model.
The second step (object identification) can be done manually or automatically:

Manual point(s) or segment(s) identification:
After the acquisition and registration of the two x-ray images acquired at two different C-arm positions, the user has to manually select points on the two x-ray images which correspond to the object to reconstruct (e.g. endograph markers and needles).

Automatic Mode for needles (only with Trackvision 2):
The user first selects a planned trajectory with a needle inserted
After the acquisition of the two X-ray images, and the registration adjustment phase, the needle will automatically be detected and reconstructed.

The provided document refers to the "Stereo 3D option for Vision Applications" (K152352). This device is an enhancement to existing GE Vision Applications (Vision 2, TrackVision 2, and HeartVision 2) and aims to reconstruct 3D objects (needles, points, and segments) from 2D X-ray images.

Based on the provided text, the device did not involve a study to establish acceptance criteria for its performance in terms of diagnostic accuracy or reader improvement. Instead, the submission relies on non-clinical tests to demonstrate substantial equivalence to predicate devices and adherence to relevant standards.

Here's a breakdown of the requested information based on the provided text:

1. Table of Acceptance Criteria and Reported Device Performance

The document does not explicitly state quantitative acceptance criteria for device performance in terms of accuracy or clinical effectiveness for the Stereo 3D option. Instead, it focuses on demonstrating compliance with standards and successful completion of various engineering and usability tests.

2. Sample size used for the test set and the data provenance

The document explicitly states: "The Stereo 3D option for Vision Applications did not require clinical studies to assess safety and effectiveness and, thus, to establish the substantial equivalence."

Therefore, there is no mention of a "test set" in the context of clinical data, sample size, or data provenance (country of origin, retrospective/prospective). The assessment was based on non-clinical testing, including "bench testing" and "simulated clinical workflows using step-by-step procedures."

3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts

Not applicable, as no clinical test set with human expert ground truth was used for assessing the device's performance. The "usability validation testing" involved "licensed and/or clinically trained healthcare providers or users," but this was for confirming usability, not establishing ground truth for reconstructive accuracy.

4. Adjudication method (e.g. 2+1, 3+1, none) for the test set

Not applicable, as no clinical test set necessitating adjudication was used.

5. If a multi reader multi case (MRMC) comparative effectiveness study was done, If so, what was the effect size of how much human readers improve with AI vs without AI assistance

No MRMC comparative effectiveness study was done. The submission explicitly states no clinical studies were required.

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

The document mentions "bench testing" and "performance testing" to confirm the device's functionality and "substantiate Stereo 3D's product claims." It also notes: "Stereo 3D contains the algorithms used to detect the 2D needles on the image and to reconstruct points, needles and segments in 3D from fluoroscopic images." This implies that the algorithms themselves were tested, which can be seen as a form of standalone testing in a controlled, non-clinical environment (e.g., using phantoms). However, specific metrics of standalone algorithmic performance (e.g., accuracy of 3D reconstruction against synthetic ground truth) or detailed study designs for this have not been provided beyond general statements about "performance being confirmed."

7. The type of ground truth used (expert consensus, pathology, outcomes data, etc.)

For the non-clinical "bench testing" and "performance testing," the ground truth likely involved phantom data with known 3D object positions and measurements. The document states: "engineering bench testing was able to be performed using existing phantoms, methods, and performance metrics."

8. The sample size for the training set

The document does not mention a "training set" in the context of machine learning or AI models with a large dataset. The "Stereo 3D option" is described as containing "algorithms used to detect the 2D needles... and to reconstruct points, needles and segments in 3D." It is based on "established GE technology" and the 3D reconstruction technology of an earlier predicate device ("Innova 3D"). This suggests that if there was any "training" in a modern AI sense, it happened as part of the development of the underlying algorithms, which are considered "established technology," and no specific training set size or methodology is provided for this submission.

9. How the ground truth for the training set was established

Not applicable, as no "training set" in a modern AI context is described or detailed for this submission. The technology is based on "established GE technology," implying algorithms developed and potentially validated previously, likely using phantom data or engineered models to establish ground truth for calibration and development of reconstruction techniques.

Ask a specific question about this device

Stroke VCAR is a CT image analysis software package that provides information to physicians to assist them in the analysis and visualization of Brain CT data derived from DICOM 3.0 compliant CT scans. Stroke VCAR is designed for the purpose of segmenting and assessing intracerebral and intracranial hemorrhages in the brain using semi-automated tools on non-contrast CT exams. Additionally Stroke VCAR provides a set of workflow tools for the segmentation and visualization of aneurysms in the brain from contrast enhanced CT exams. It is intended for use by clinicians to process, review, archive, print and distribute CT studies.

This software will assist the user by providing initial 3D segmentation, measurements and visualization of hemorrhages and aneurysm in the brain. The user has the ability to adjust, review and has to confirm the final segmentation.

Stroke VCAR is intended to provide 2D and 3D processing, review and analysis of CT images originally acquired to evaluate the cerebral vascular system and/or intracranial bleeding. The combination of the acquired images, reconstructed images, and measurements performed by the clinician using Stroke VCAR are intended to provide the referring physician clinically relevant information for the purpose of diagnosis, treatment planning and follow-up.

Here's an analysis of the provided text regarding the Stroke VCAR device, focusing on acceptance criteria and the supporting study:

1. Table of Acceptance Criteria and Reported Device Performance

The provided text does not explicitly state numerical acceptance criteria in terms of performance metrics (e.g., accuracy, sensitivity, specificity, Dice score, etc.) for the Stroke VCAR device. Instead, the "Summary of Clinical tests" section details observed benefits and qualitative feedback. The closest "performance" reported relates to workflow time improvement.

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

The text states: "A clinical evaluation study using consented clinical images was conducted..."

• Sample Size for Test Set: The specific number of cases (clinical images) used in the study is not explicitly stated. It mentions "consented clinical images," but gives no numerical quantity.
• Data Provenance (e.g., country of origin, retrospective or prospective): "Consented clinical images" suggests real-world data, likely retrospective, but the country of origin is not specified.

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

• Number of Experts: Three board-certified neuroradiologists were used in Part 2a and Part 2b of the study.
• Qualifications of Experts: They were identified as "board certified neuroradiologists" and "considered experts."

4. Adjudication Method for the Test Set

The document does not explicitly state an adjudication method (such as 2+1 or 3+1 consensus) for establishing the ground truth or evaluating the device's output. It mentions "three board certified neuroradiologists" but not how their assessments were combined or reconciled.

5. If a Multi Reader Multi Case (MRMC) Comparative Effectiveness Study was done, If so, what was the effect size of how much human readers improve with AI vs without AI assistance

The study described does not appear to be a traditional MRMC comparative effectiveness study in the sense of comparing reader performance with and without AI assistance on diagnostic accuracy metrics.

Instead, the study's stated goals were:

• "Productivity benefits of automated vs. manual hematoma segmentation (Part 1)"
• "General user Qualitative feedbacks of the Stroke VCAR edition tools (Part 2a and Part 2b)"

While it compares automated vs. manual workflow time, it doesn't quantify diagnostic accuracy improvement for human readers using the AI. Therefore, an effect size of human reader improvement with vs. without AI assistance is not reported.

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

The provided text does not explicitly describe a standalone performance study of the algorithm without human-in-the-loop. The description states: "This software will assist the user by providing initial 3D segmentation, measurements and visualization of hemorrhages and aneurysm in the brain. The user has the ability to adjust, review and has to confirm the final segmentation." This indicates that the device is intended for semi-automated use where the human user confirms the results, implying that standalone performance may not be the primary evaluation focus here.

7. The Type of Ground Truth Used (expert consensus, pathology, outcomes data, etc.)

Given the involvement of "three board certified neuroradiologists" who "were considered experts" in the clinical evaluation, the ground truth for evaluating the device's output (specifically for "diagnostic value" and potentially the accuracy of the automated segmentations) was likely based on expert consensus or expert interpretation of the clinical images. There's no mention of pathology or outcomes data as ground truth.

8. The Sample Size for the Training Set

The provided text does not specify the sample size for the training set used to develop the Stroke VCAR software. The "Summary of Non-Clinical Tests" mentions that "Stroke VCAR optimizes the segmentation algorithms for Hematoma segmentation and Aneurysm segmentation," implying training, but no details on the training data are given.

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

The provided text does not describe how the ground truth for the training set was established.

Ask a specific question about this device

Innova EPVision 2.0 software application is intended to enable users to load 3D datasets and overlay and register in real time these 3D datasets with radioscopic or radiographic images of the same anatomy in order to support catheter/device guidance during interventional procedures.

Innova EPVision 2.0 software application is intended to enable users to load, overlay and register in real time 3D datasets with radioscopic or radiographic images of the same anatomy. Electrophysiological signal information is imported and used to color-code these 3D datasets in order to support catheter/device during cardiac electrophysiology guidance interventional procedures.

Innova EPVision 2.0 is the new version of Innova EPVision software, which is part of the Innova Vision Applications [K092639] software. Innova EPVision 2.0, as all Innova Vision Applications image processing algorithms, is executed on a hardware called Advantage platform Workstation (AW) [K110834].
It can perform the following functions:

• Superimpose the segmented DICOM 3D XA, CT, MR dataset on radioscopic or radiographic image of the same anatomy, obtained on an Innova Fluoroscopic X-ray system [K113034].
• . Register the segmented DICOM 3D XA. CT. MR dataset with radioscopic or radiographic images obtained on an Innova Fluoroscopic X-ray system for interventional procedures.
• . Image stabilization features such as ECG gated display or motion tracking in the image.
• . Capability to load planning data, deposited on the 3D model in Volume Viewer [K041521], such as 3D landmarks, ablations lines, and to display them on the 3D-2D fused image to support the physician during procedures.
• . Marking points of interest of different size and color during the procedures.
• . The frequently used functions are also available from tableside on the Innova Central Touch Screen to provide efficient workflow during the interventional procedures.

Innova EPVision 2.0 can perform additionally the following functions:

• Import electrophysiology (EP) data digitized and processed . on the CardioLab system [K111200] and use them to colorcode EP recording points on 3D model of the visualized anatomy in order to support catheter/device guidance during cardiac electrophysiology interventional procedures.
• . Catheter tip detection to help locate the catheter tip on the 2D X-ray image. The user can modify or correct the automatically proposed tip location anytime.
• . 3D side viewer allowing the user to freely rotate the 3D model independently from the Fluoro image and the gantry angulation.

Innova EPVision 2.0, as Innova EPVision targets clinical indication for interventional cardiology procedures and in particular cardiac electrophysiology procedures.

Here's an analysis of the provided text regarding the acceptance criteria and study for the Innova EPVision 2.0 device:

1. Table of Acceptance Criteria and Reported Device Performance

The provided 510(k) summary does not explicitly state specific quantitative acceptance criteria or detailed performance metrics for the Innova EPVision 2.0. Instead, it relies on demonstrating substantial equivalence to a predicate device (K092639, Innova Vision Applications) and adherence to various standards (NEMA PS 3.1 - 3.20 (2011) DICOM Set, IEC 60601-1-4 (2000), IEC 62304 (2006), IEC 62366 (2007)).

The "reported device performance" is described in terms of verification and validation activities designed to ensure the device works as required and meets user needs and intended use.

Summary of Device Performance (as described):

2. Sample size used for the test set and the data provenance

The document does not specify a "test set" in the context of clinical data for performance evaluation. Instead, it discusses verification and validation tests as part of the software development lifecycle. These tests would involve a variety of inputs and scenarios, but the exact sample sizes (e.g., number of test cases, number of images) for these engineering tests are not provided.

There is no mention of data provenance (e.g., country of origin, retrospective/prospective) because no clinical studies were deemed necessary or performed to support substantial equivalence.

3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts

This information is not applicable or provided as no clinical studies were performed that would require expert-established ground truth on a test set. The ground truth for the engineering verification and validation tests would be the expected behavior of the software as defined by its requirements and design specifications, not clinical expert consensus.

4. Adjudication method for the test set

This information is not applicable or provided since no clinical validation study involving adjudication of a test set was performed.

5. If a multi-reader multi-case (MRMC) comparative effectiveness study was done, If so, what was the effect size of how much human readers improve with AI vs without AI assistance

No MRMC comparative effectiveness study was done. The submission explicitly states: "The subject of this premarket submission, Innova EPVision 2.0, did not require clinical studies to support substantial equivalence." Therefore, there is no reported effect size of human readers improving with or without AI assistance.

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

The document describes the device as a software application that "enables users to load, overlay and register... in order to support catheter/device guidance." It also mentions "Catheter tip detection to help locate the catheter tip on the 2D X-ray image. The user can modify or correct the automatically proposed tip location anytime." This clearly indicates a human-in-the-loop design.

While the "Catheter tip detection" component could have an internal standalone performance evaluation during development, the submission does not report a standalone (algorithm only) performance study in the context of a regulatory submission outcome. The overall device is intended to be used with a human interventionalist.

7. The type of ground truth used (expert consensus, pathology, outcomes data, etc.)

For the engineering verification and validation tests, the ground truth would be the expected software behavior and functionality based on design input (system requirements and specifications). For example, a test for "Superimpose the segmented DICOM 3D XA, CT, MR dataset on radioscopic or radiographic image" would have a ground truth defined by the correctness of the overlay as per the design specifications. There is no mention of clinical ground truth (expert consensus, pathology, or outcomes data) being used for the regulatory submission's performance evaluation because no clinical studies were conducted.

8. The sample size for the training set

This information is not provided. The document makes no mention of machine learning model training or a training set. The device is described as inheriting functions and employing the "same fundamental scientific technology" as its predicate, implying a rule-based or traditional image processing approach rather than a machine learning approach requiring distinct training data that would be relevant to this 510(k) submission.

9. How the ground truth for the training set was established

This information is not provided as no training set is mentioned in the context of this 510(k) submission.

Ask a specific question about this device