Search Results

Ask a specific question about this device

The Bunkerhill BMD Algorithm is a post-processing AI-powered software intended for adults 30 years and above to assess estimated DXA-measured average areal bone mineral density of spinal bones from existing CT scans and outputs a flag for low bone density below a pre-specified threshold. It is not intended to replace DXA or any other tests dedicated to BMD measurement.

Bunkerhill BMD is an opportunistic AI-powered tool that enables:(1) retrospective assessment of bone density from CT scans acquired for other purposes, (2) assessment of bone density in conjunction with another medically appropriate procedure involving CT scans, and (3) assessment of bone density without a phantom as an independent measurement procedure

The Bunkerhill BMD application is a software only medical device (SaMD) that includes deep- learning-based computer vision and post-processing algorithms that estimates the bone mineral density from previously obtained computed tomography (CT) images.

The results from Bunkerhill BMD are not intended to be used as the primary input for clinical decision making, but rather are intended to provide information that may assist the clinician to identify 'findings of interest' within existing imaging studies.

Here's a breakdown of the acceptance criteria and study details for the BunkerHill BMD device, based on the provided FDA 510(k) clearance letter:

Acceptance Criteria and Device Performance Study for BunkerHill BMD

1. Table of Acceptance Criteria and Reported Device Performance

2. Sample Size and Data Provenance for the Test Set

• Test Set Sample Size: 371 CT studies
• Data Provenance: The studies were collected from four (4) geographically diverse sites. The retrospective nature of the study is explicitly stated ("stand-alone retrospective study").

3. Number of Experts and Qualifications for Ground Truth

The document does not explicitly state the number of experts used to establish the ground truth or their specific qualifications (e.g., "radiologist with 10 years of experience").

4. Adjudication Method for the Test Set

The document does not explicitly state the adjudication method (e.g., 2+1, 3+1, none) used for the test set.

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

No, an MRMC comparative effectiveness study involving human readers with and without AI assistance was not reported in the provided text. The study described is a standalone performance evaluation of the algorithm.

6. Standalone Performance Study

Yes, a standalone (algorithm only without human-in-the-loop performance) study was done. The document states: "Bunkerhill BMD performance was validated in a stand-alone retrospective study for overall agreement of the device output compared to the established ground truth."

7. Type of Ground Truth Used

The type of ground truth used is implied to be based on DXA-measured average areal bone mineral density of spinal bones, as the device is intended to "assess estimated DXA-measured average areal bone mineral density." The text refers to "established ground truth" in relation to this assessment.

8. Sample Size for the Training Set

The document does not provide the sample size for the training set. It only describes the test 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. It only refers to "established ground truth" for the test set evaluation.

Ask a specific question about this device

TBS iNsight is a software provided for use as a complement to both DXA analysis and clinical examination. It computes the antero-posterior spine DXA examination file and calculates a score (Trabecular Bone Score - TBS) that is compared to those of the age matched controls. The TBS is derived from the texture of the DXA image and has been shown to be related to bone microarchitecture.

TBS iNsight provides as an option an assessment of 10-year fracture risk. It provides an estimate of 10-year probability of hip fracture and 10-year probability of a major osteoporotic fracture (clinical spine, forearm, hip or shoulder fracture). This estimate is based on the WHO's FRAX® Fracture Risk Assessment Tool, after adjustment for the TBS. The tool has been validated for Caucasian and Asian men and post-menopausal women between 40 and 90 years old.

TBS provides information independent of BMD value: it is used as a complement to the data obtained from the DXA analysis and the clinical examination (questioning by the clinician about patient history, bioassay of bone resorption markers...).

The results can be used by a physician in conjunction with other clinical risk factors as an aid in the diagnosis of osteoporosis and other medical conditions leading to altered trabecular bone microarchitecture, and ultimately in the assessment of fracture risk.

The TBS score can assist the health care professional in monitoring the effect of treatments on patients across time. Overall fracture risk will depend on many additional factors that should be considered before making diagnostic or therapeutic recommendations. The software does not disease or recommend treatment regimens. Only the health care professional can make these judgments.

TBS iNsight is a software application provided for use as a complement to bone mineral density (BMD) acquired from dual energy X-ray absorptiometry (DXA) and other clinical risk factors for osteoporosis and fragility fracture. It calculates a score (Trabecular Bone Score - TBS) derived from the texture of the DXA image of the anterior-posterior (AP) lumbar spine and has been shown to be related to bone microarchitecture. The method analyzes X-ray based images acquired by DXA imaging systems and produces the TBS based on the computation of an Adapted Experimental Variogram (modified fractal-like approach). This variogram is used to measure the degree of spatial variation between pairs of data points in a spatial dimension of a region of a digital image.

The absolute TBS values for the same equivalent tissue thickness vary slightly between GE and Hologic systems. Additionally, within each system, variability in TBS values can be observed across different scan modes. To address these differences, corrections are applied that are both device-specific and mode-specific. For instance, TBS is corrected differently for GE and Hologic systems to account for inherent differences in tissue thickness assessment. Furthermore, corrections are also tailored separately for each scan mode within the same system, ensuring that TBS measurements remain consistent and reliable regardless of the scan mode used. These device-specific and mode-specific corrections are necessitated by differences in the dynamic range of tissue thickness measurements between GE and Hologic devices. The variations arise due to differences in the methodologies used to assess tissue thickness. To harmonize these discrepancies and ensure measurement accuracy, correction fits derived from ex-vivo data are applied individually to each device and scan mode. This approach ensures the accuracy and consistency of TBS measurements across all configurations.

The device is intended to be used for bone health assessment in medical facilities employing one or more DXA system(s) to which the subject device is connected. These facilities are usually hospitals, clinics, healthcare centers, radiology practices and medical imaging centers. The software is designed to be used by qualified clinical professionals (including physicians, radiologists and DXA technicians) and the physicians are solely responsible for making all final patient management decisions.

The provided text does not contain a discrete table of acceptance criteria nor explicit reported device performance metrics against such criteria in a tabular format. However, based on the Performance Testing Summary & Conclusions section (pages 9-10), the acceptance criteria can be inferred from the studies described. The reported performance is woven into the narrative of the study results.

Here's an attempt to derive the information requested:

1. Table of Acceptance Criteria and Reported Device Performance

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

• Clinical Association Study 1 (Correlation with microCT):
  • Sample Size: 30 human cadaver lumbar vertebrae.
  • Data Provenance: Not explicitly stated, but "human cadaver lumbar vertebrae" implies a laboratory-based, retrospective study.
• Clinical Association Study 2 (Agreement between V4 and Predicate):
  • Sample Size: 15 human cadaver vertebrae.
  • Data Provenance: Not explicitly stated, but "human cadaver vertebrae" implies a laboratory-based, retrospective study. Measurements were performed on "multiple DXA systems."
• Analytical Reproducibility (Ex Vivo Study):
  • Sample Size: Not explicitly stated, but "dried human lumbar vertebrae," likely similar to the cadaver studies.
  • Data Provenance: Retrospective, laboratory-based.
• Analytical Reproducibility (In Vivo Study):
  • Sample Size: 132 participants.
  • Data Provenance: Not explicitly stated, but "scanned on four different DXA systems" with "repositioning between scans" indicates a prospective, controlled clinical study.
• Performance and Utility (Fracture Risk Assessment):
  • Sample Size: Over 17,000 individuals across 14 international cohorts.
  • Data Provenance: International cohorts, implying retrospective data from multiple countries.
• Performance and Utility (Treatment Monitoring):
  • Sample Size: A "large cohort" of postmenopausal women with osteoporosis. Specific number not provided.
  • Data Provenance: Not explicitly stated, but likely retrospective clinical data.

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

The document does not specify the number or qualifications of experts used to establish ground truth for the various test sets.

• For microCT correlation: Micro-computed tomography (microCT) is an objective imaging technique, so human expert interpretation for ground truth might not be directly applicable in the same way as, for example, image interpretation by radiologists.
• For in vivo studies: The ground truth for bone microarchitecture or fracture risk assessment, as implied by the use of ISCD standards and FRAX, are typically established by clinical diagnosis and follow-up, rather than a panel of experts explicitly reviewing the test set images for ground truth.

4. Adjudication Method for the Test Set

The document does not describe any adjudication methods (e.g., 2+1, 3+1) for establishing ground truth from expert readers for the test sets.

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, an MRMC comparative effectiveness study was not explicitly described in the provided text. The studies focus on the device's standalone performance, its agreement with the predicate device, and its correlation with bone microarchitecture ground truth (microCT), as well as its utility for fracture risk assessment and treatment monitoring.
• The text describes the device as a "complement to both DXA analysis and clinical examination" and states that "The results can be used by a physician in conjunction with other clinical risk factors." While it aids physicians, the document does not include a comparative study evaluating how human reader performance (e.g., diagnosis or risk assessment accuracy) improves with TBS iNsight V4 assistance versus without it.

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

• Yes, a standalone performance assessment was conducted. The "Standalone Performance Testing Summary" section (page 9) explicitly states that "Software Verification and Validation demonstrate specified requirements are met and the software functions as intended."
• Furthermore, the "Clinical Association" and "Analytical Reproducibility" studies directly evaluate the performance of TBS version 4.0 in calculating TBS values and their correlation with microCT independently, and their agreement/reproducibility compared to the predicate, without human intervention in the calculation process.

7. The Type of Ground Truth Used

• Pathology/Objective Measurement: For establishing the relationship between TBS and bone microarchitecture, micro-computed tomography (microCT) measurements were used as the ground truth (e.g., for trabecular number and separation). This is an objective measurement of bone structure.
• Clinical Diagnosis/Outcomes Data: For "fracture risk assessment" and "treatment monitoring," the ground truth implicitly refers to the clinically established reference values (e.g., those used in FRAX) and potentially actual fracture outcomes data from the cohorts. The studies confirm that TBS V4 provides "similar TBS-adjusted FRAX and BMD T-scores" and is effective for monitoring, suggesting that it aligns with known clinical indicators and outcomes.
• Predicate Device Performance: For demonstrating "agreement" and "substantial equivalence," the performance of the predicate device (TBS iNsight V3) serves as a comparative ground truth or benchmark.

8. The Sample Size for the Training Set

The document does not provide information regarding the sample size used for the training set for the TBS iNsight V4 algorithm.

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

The document does not provide information on how the ground truth for any training set was established, as the details of algorithm development and training are not included. The focus is on the performance testing of the final device version.

Ask a specific question about this device

VirtuOst uses data from computed tomography scans to estimate bone mineral density, bone strength, and a load-to-strength ratio. This information can be used by a physician to assess fracture risk, identify osteoporosis, and monitor therapy. For pediatric patients, VirtuOst provides these estimates without any classifications and should be used only when the benefit of obtaining these estimates outweighs the risk of radiation.

VirtuOst is a software-only medical device that analyzes data in computed tomography (CT) scans to measure bone mineral density (BMD), bone strength, and a load-to-strength ratio at the proximal femur and vertebral body. BMD is measured from both a 2D projection (in g/cm2) and a volumetric scan reconstruction (in mg/cm³) of the CT scan. VirtuOst measurements can be used by a physician to identify osteoporosis, assess fracture risk, and monitor therapy. The VirtuOst analysis is performed on previously physician-acquired image data and is unrelated to acquisition equipment and clinical workstations.

1. Table of Acceptance Criteria and Reported Device Performance

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

• Hip Areal BMD Equivalence:
  • Test Set Size: 324 women and men (overall). Specifically, 244 women and 80 men.
  • Data Provenance: Not explicitly stated, but implies clinical data. Retrospective or prospective is not specified.
• Phantomless Calibration:
  • Development of reference data for visceral fat attenuation: 268 patients and a custom torso phantom scanned on 31 and 35 different CT scanners respectively.
  • Comparison of fat/air phantomless calibration vs. phantom calibration: 40 independent patient CT scans.
  • Data Provenance: Not explicitly stated, but implies clinical data from various CT manufacturers (GE Healthcare, Siemens Medical Solutions, Philips Healthcare, Toshiba International) at various settings (80-140 kVp). Retrospective or prospective is not specified.
• CT-to-DXA Areal BMD Conversion for Hip:
  • Test Set Size: 200 subjects scanned on both CT and DXA.
  • Data Provenance: Not explicitly stated, but implies clinical data. Retrospective or prospective is not specified.
• Slice Spacing Adjustment for Femurs:
  • Development of adjustments: Pairs of CT images (unspecified number).
  • Observational study: 2,783 women and men (1,306 with hip fracture). This was a case-cohort study design.
  • Data Provenance: Not explicitly stated, but implies clinical data. The observational study implies retrospective data.

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

The document does not describe the use of human experts to establish ground truth for the test sets. Instead, it relies on:

• Clinical reference standards: FDA-cleared DXA devices for hip areal BMD equivalence.
• External calibration phantoms: For validating phantomless calibration.
• Outcomes data: Hip fracture prediction in an observational study for slice spacing adjustment.
• Regression analysis and statistical comparison: For quantitative metrics.

4. Adjudication Method for the Test Set

Not applicable, as expert adjudication was not used to establish ground truth. Ground truth was based on objective measurements from clinical reference standards (DXA, external phantoms) or clinical outcomes.

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

No MRMC comparative effectiveness study involving human readers with and without AI assistance was mentioned. The study focused on the equivalence of the device's measurements to established clinical standards (DXA, phantom) and its ability to predict outcomes, rather than human reader performance improvement.

6. Standalone (Algorithm Only) Performance Study

Yes, the studies presented primarily demonstrate the standalone performance of the VirtuOst algorithm.

• The comparison of hip areal BMD measurements to DXA devices is a standalone assessment.
• The phantomless calibration validation compares the algorithm's output to a known standard (external phantom).
• The CT-to-DXA conversion study assesses the algorithm's ability to produce DXA-equivalent T- and Z-scores.
• The slice spacing adjustment validation and the observational study on hip fracture prediction also assess the algorithm's standalone performance in providing clinically relevant measurements.

7. Type of Ground Truth Used

The ground truth used was a combination of:

• Clinical reference standards: Measurements from FDA-cleared DXA devices for hip areal BMD.
• Physical standards: External calibration phantoms for validating phantomless calibration.
• Outcomes data: Hip fracture occurrence in a case-cohort study for evaluating the slice spacing adjustment and overall clinical utility.
• Regression analysis and statistical comparison: Used to define the relationship between the device's measurements and the chosen ground truths.

8. Sample Size for the Training Set

The document does not explicitly state the sample sizes used for training the VirtuOst algorithm. The information provided relates to testing and validation of the modified device.

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

The document does not provide details on how the ground truth for the training set was established. The studies described are primarily focused on the validation of the modified device's performance against established benchmarks and clinical data, rather than the initial development and training of the core algorithms. It is implied that the predicate device (K113725) was developed and trained prior to this submission, and this submission focuses on demonstrating that modifications have not negatively impacted safety or effectiveness.

Ask a specific question about this device

3D-SHAPER uses data from conventional Dual Energy X-Ray Absorptiometry (DXA) scans to measure the distribution of bone mineral mass at specific cross sections of the hip and allows the physician to estimate structural properties of the hip, such as CSA, CSMI, Z, Buckling Ratio, cortical sBMD, trabecular vBMD and integral vBMD.

3D-SHAPER is indicated to be used by qualified medical professionals to assist healthcare professionals in the bone health evaluation & changes, and cannot fully substitute their clinical judgement.

3D-SHAPER is a software reconstructing in 3D, the shape and bone mineral density from 2D DXA images. The software incorporates a DXA-based 3D reconstruction and measures the distribution of bone mineral mass at specific cross sections of the hip. It provides physicians with estimates of the structural properties of the hip, such as densities or geometrical parameters. The software is not suitable for use with paediatric patients or for subjects with hip prosthesis such as total hip replacement prosthesis, hip resurfacing prosthesis or an osteosynthesis system, since the software has not been tested against these conditions. 3D-SHAPER is indicated to be used by qualified medical professionals to assist healthcare professionals in the bone health evaluation & changes, and cannot fully substitute their clinical judgement.

The 3D-SHAPER® measurements include:

Cortical surface Bone Mineral Density (Cortical sBMD, mg/cm2)

Trabecular volumetric Bone Mineral Density (Trabecular vBMD, mg/cm³)

Integral volumetric Bone Mineral Density (Integral vBMD, mg/cm³), the integral compartment being the union of the cortical and trabecular compartments.

Hip structural measurements: Cross-Sectional Area (CSA, in cm²), Cross-Sectional Moment of Inertia (CSMI, in cm4), section modulus (Z, in cm3) and buckling Ratio (BR).

Cortical sBMD, trabecular VBMD and integral vBMD are calculated at the total femur. CSA, CSMI, Z and BR are calculated in cross-sections at the neck, intertrochanteric region and lower shaft.

Here's a breakdown of the acceptance criteria and study information for the 3D-SHAPER device, based on the provided document:

1. Table of Acceptance Criteria and Reported Device Performance:

The document primarily focuses on demonstrating substantial equivalence to a predicate device through correlation studies rather than establishing absolute performance metrics against predefined acceptance criteria for each measurement. However, it does outline the type of performance expected in the "Performance Bench Testing" section.

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

The document mentions two main studies:

• Correlation Study with Predicate Device:
  • Sample Size: 740 men and women (94% women).
  • Data Provenance: Not explicitly stated (e.g., country of origin). The study is retrospective in the sense that existing DXA scans were analyzed by both the 3D-SHAPER and the predicate device.
• Performance Bench Testing (Correlation to QCT):
  • Sample Size: Not explicitly stated for this particular correlation study with QCT. It refers to "subject-specific models of the femoral shape and measurements compared to Quantitative Computed Tomography (QCT)." This implies data from human subjects but the exact number isn't provided.
  • Data Provenance: Not explicitly stated. Likely retrospective, as it compares measurements to an established imaging modality (QCT).

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

• Correlation Study with Predicate Device: Ground truth for this study was the measurements produced by the predicate device (Hologic Hip Structural Analysis software, K061561). No human experts were involved in establishing this specific ground truth; it relied on a previously cleared, validated software.
• Performance Bench Testing (Correlation to QCT): Ground truth was established by Quantitative Computed Tomography (QCT) measurements. QCT is a quantitative imaging technique that provides volumetric bone mineral density and structural information. No human experts are described as manually establishing this ground truth; it's a direct comparison to QCT output.

4. Adjudication Method:

• Correlation Study with Predicate Device: There was no adjudication method described. The study compared the output of the 3D-SHAPER software directly against the output of the predicate device.
• Performance Bench Testing (Correlation to QCT): There was no adjudication method described. The study compared the output of the 3D-SHAPER software directly against QCT measurements.

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

No, a multi-reader multi-case (MRMC) comparative effectiveness study was not done. The study described focuses on the comparison of the 3D-SHAPER software's measurements against a predicate software and against QCT, not on how human readers' performance improves with or without AI assistance.

6. Standalone Performance:

Yes, a standalone performance study (algorithm only without human-in-the-loop) was done. The document consistently describes the 3D-SHAPER software's ability to measure structural properties from DXA scans and correlates these measurements with a predicate device and QCT. This is considered standalone performance.

7. Type of Ground Truth Used:

• Correlation Study with Predicate Device: The ground truth was the output of a legally marketed predicate device (Hologic Hip Structural Analysis software).
• Performance Bench Testing (Correlation to QCT): The ground truth was Quantitative Computed Tomography (QCT) measurements.

8. Sample Size for the Training Set:

The document does not provide any information regarding the sample size used for the training set. It focuses on the validation of the device.

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

The document does not provide any information on how the ground truth for the training set was established, as the training process itself is not detailed.

Ask a specific question about this device

The Automated Bone Mineral Density Software Module (ABMD) is a post-processing AI-powered software intended to measure bone mineral density (BMD) from existing CT scans by averaging Hounsfield units in the trabecular region of vertebral bones. ABMD is not intended to replace DXA or any other tests dedicated to BMD measurement. It is solely designed for measuring BMD in existing CT scans ordered for reasons other than BMD measurement. In summary, ABMD is an opportunistic AI-powered tool that enables: (1) retrospective assessment of bone density from CT scans acquired for other purposes, (2) assessment of bone density in conjunction with another medically appropriate procedure involving CT scans, and (3) assessment of bone density without a phantom as an independent measurement procedure.

The Automated Bone Mineral Density (ABMD) Software is a software module that estimates bone mineral density in the vertebral bones by averaging Hounsfield Units (HU) in the trabecular area. ABMD Software is a post-processing software that works on existing CT scans. ABMD Software measurements are to be reviewed by radiologists and should be used by healthcare providers in conjunction with clinical evaluation.

Here's a breakdown of the acceptance criteria and study information for the ABMD Software based on the provided text:

1. Table of Acceptance Criteria and Reported Device Performance

Ask a specific question about this device

The proposed device: GE Lunar DXA Bone Densitometers with the enCORE version 18 software is intended for medical purposes to measure bone density, bone mineral content, and fat and lean tissue content by x-ray transmission measurements through the bone and adjacent tissues.

The changes proposed in this Premarket Notification will be used with the existing GEHC DXA Bone Densitometers, which utilize the enCORE software.

The GEHC DXA Bone Densitometers are composed of a scanner and a computer. The scanner comprises the x-ray source and detector, the patient scan table, the mechanical drive system, and the lowest level portions of the control system. The scanner is in communication with the computer, which is a standard PC. The computer runs the enCORE software, and thus controls the scanner, acquires scan data from the scanner, stores and analyzes the data, and interacts with the human operator.

GEHC DXA Bone Densitometers are used healthcare facilities and hospitals to measure bone mineral density (BMD) and body composition (%fat, fat mass, lean mass) using a technique called Dual-energy X-ray Absorptiometry or DXA. DXA measures the attenuation of x-rays of two different energy levels after they pass through the body of a subject. As bone, fat tissue, and lean tissue absorb the different energy x-rays at different rates, the relative attenuation of each x-ray energy is measured and used to calculate the composition of each pixel.

The provided text describes the GEHC DXA Bone Densitometers with enCORE version 18. This device is a bone densitometer that measures bone mineral density (BMD) and body composition using Dual-energy X-ray Absorptiometry (DXA). The submission is a 510(k) premarket notification, indicating that the device claims substantial equivalence to existing legally marketed predicate devices.

1. Table of Acceptance Criteria and Reported Device Performance

The document does not explicitly present a table of acceptance criteria with specific numerical thresholds for performance. Instead, it describes various software features and states that bench performance testing confirmed that design outputs met design input requirements and that results are "comparable" or "precise and accurate."

Here's an attempt to structure the information into a table based on the provided text, recognizing that precise criteria are often implied rather than explicitly stated with quantifiable targets in this type of summary:

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

• DXAVision/Adult TBLH: The study involved a "heterogeneous sample of men and women with a diverse range of age, BMI, BMD, and body fat." No specific numerical sample size is provided.
• Integrated TBS iNsight: "internal engineering DXA data sets" were used. No specific numerical sample size.
• CoreScan: An "anthropomorphic phantom" was used. This is not human data.
• Small Body Composition ROIs: No specific sample size or provenance is mentioned for the verification results.
• Advanced Analytics: No specific sample size or provenance is mentioned.
• Software Verification and Validation: No specific sample size or provenance is mentioned besides "software verification and validation testing."

The provenance of human data, based on the description, is internal to GE Healthcare ("internal engineering DXA data sets"). It is implied to be retrospective, as it refers to existing data sets.

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

The document does not explicitly mention the number or qualifications of experts used to establish ground truth for any of the described tests. The studies focus on comparing the updated software features to previous cleared versions or literature, or phantom data, rather than independent expert-adjudicated ground truth.

4. Adjudication Method for the Test Set

No adjudication method is mentioned for any of the described tests. The evaluations appear to be based on comparisons to predicate technology, internal software testing, or phantom measurements.

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 is mentioned. The device's improvements are focused on streamlined workflow, integration of features, and accuracy/precision of measurements, not on direct human reader improvement with AI assistance.

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

Yes, the studies described are primarily standalone algorithmic performance evaluations. For example, the CoreScan bench testing used an anthropomorphic phantom, and the TBS iNsight validation compared algorithmic outputs. The DXAVision and Adult TBLH studies compare algorithm results to literature-reported precision, implying a standalone assessment of the algorithm's measurement precision.

7. The Type of Ground Truth Used

• DXAVision/Adult TBLH: The "ground truth" used for comparison appears to be the precision of "duplicate scans reported in the literature" for existing iDXA and Prodigy devices. This implies a comparison to established performance benchmarks rather than an independently adjudicated 'ground truth' for each case.
• Integrated TBS iNsight: The "ground truth" is the results from the "previously cleared TBS iNsight application."
• CoreScan: The "ground truth" is derived from the known characteristics of an "anthropomorphic phantom."
• Small Body Composition ROIs: "Accuracy of bone and body composition" is confirmed, suggesting comparison to a known standard or previously validated measurements, but the specific nature of this ground truth is not detailed.
• Advanced Analytics: No explicit ground truth is mentioned, as this feature is about user-defined equations using existing data.

8. The Sample Size for the Training Set

The document does not provide details about training sets for any of the software features. This submission is for updates to existing software (enCORE version 18) and integration of previously cleared applications, suggesting development was not based on a new, distinct training phase described in this summary.

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

Since no training set details are provided, the method for establishing ground truth for a training set is not described. The focus of this 510(k) submission is on changes and integrations based on verified algorithms and features, rather than the initial development of a new AI model requiring a training set.

Ask a specific question about this device

X-ray Bone Densitometer designed to estimate the bone mineral density of patients when medically indicated by their physicians.

• Provides an estimate of bone mineral density at various anatomical sites (Spine, Femur, Forearm). These values can then be compared to an adult reference population at the sole discretion of the physician.
• Provides an assessment of relative risk based on the patient's T-score value using the categories of fracture risk defined by the World Health Organization (WHO).
• Provides an assessment of 10-year fracture risk using WHO FRAX model.
• Provides a standardized bone density report using data from the densitometer and physician-generated assessments based on the patient's demographics, which can assist the physician in communicating scan results to the patient and the patient's referring physician.

The Aria X-ray Bone Densitometer is designed to be a value product version of the predicate device, the Prodigy (K982267 and K161682). Like its predicate Prodigy device, the proposed Aria device is composed of a scanner and a computer that runs the software. The scanner comprises the x-ray source and detector, the patient scan table, the mechanical drive system, and the lowest level portions of the control system. The scanner is in communication with the computer, which is a standard PC. The computer runs the enCORE software, and thus controls the scanner, acquires scan data from the scanner, stores and analyzes the data, and interacts with the human operator.

Aria X-ray Bone Densitometer functions with the same software as the one that is FDA cleared under GE Lunar DXA Bone Densitometers with enCORE version 17 (K161682).

Here's a breakdown of the acceptance criteria and the study conducted for the Aria device, based on the provided text:

Based on the provided information, the Aria device is an X-ray Bone Densitometer. The study presented is a bench testing to demonstrate precision and accuracy, not a clinical study involving human subjects or AI algorithms. As such, several requested items like "Multi-reader multi-case (MRMC) comparative effectiveness study" and "Human reader improve with AI vs without AI assistance" are not applicable.

1. Table of Acceptance Criteria and Reported Device Performance

2. Sample Size and Data Provenance for the Test Set

• Sample Size: The text mentions "a wide range of phantoms" for precision and accuracy tests, and "a set of phantoms with different BMD and tissue composition properties." However, specific numerical sample sizes for these phantom sets are not provided.
• Data Provenance: The tests were conducted during "product development" as "bench testing." The data provenance is from phantom measurements, not human participants. No country of origin is explicitly stated for the phantom data, but it's implied to be internal testing by GE Healthcare (USA). The study is retrospective in the sense that it's based on controlled phantom measurements rather than prospective patient recruitment.

3. Number of Experts and Qualifications for Ground Truth

• Not applicable. The ground truth for the test set was established using phantoms with known bone mineral density (BMD) and tissue composition properties, not by human experts. The accuracy was verified by showing an "excellent correlation to previously released Prodigy product," implying a comparison to an established device's performance.

4. Adjudication Method for the Test Set

• Not applicable. Since the ground truth was phantom-based and not expert-driven, there was no adjudication method involving multiple experts.

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

• No, a multi-reader multi-case (MRMC) comparative effectiveness study was not done. The presented study is a bench test, and the device is a bone densitometer, not an AI-assisted diagnostic tool for human interpretation. Therefore, there's no "effect size of how much human readers improve with AI vs without AI assistance" to report.

6. Standalone (Algorithm Only) Performance Study

• Yes, a standalone performance study was done. The precision and accuracy bench testing evaluates the algorithm's (and the overall device's) ability to accurately and precisely estimate bone density from phantom measurements without human intervention in the measurement process. The results "demonstrate that Aria can accurately and precisely estimate bone density."

7. Type of Ground Truth Used

• The ground truth used was phantom-based with established, known values for bone mineral density (BMD) and tissue composition properties. These values were validated against the performance of the predicate Prodigy product.

8. Sample Size for the Training Set

• Not explicitly stated and likely not applicable in the traditional sense for a machine learning model. The Aria device is presented as a "value product version" using the "same software as the one that is FDA cleared under GE Lunar DXA Bone Densitometers with enCORE version 17 (K161682)." This suggests that the core algorithms might have been developed and "trained" (if applicable for this type of system) with data associated with the enCORE software, but the document does not provide details on a specific training set size for Aria's development.

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

• Not explicitly stated and likely not applicable in the traditional sense for a machine learning model. Given that the Aria uses the same software (enCORE v17) as its predicate, the ground truth for any underlying algorithm "training" would have been established during the development of that software. Typically for bone densitometry, this involves:
  • Phantom studies: Similar to the testing described, using phantoms with known physical properties.
  • Clinical studies: Correlating device measurements against other established methods or clinical outcomes, although this document states clinical studies were "not required" for Aria's substantial equivalence.

Ask a specific question about this device

Optional Atypical Femur Fracture (AFF) software uses Femur images to visualize focal reaction or thickening along the lateral cortex of the femoral shaft which may be accompanied by a transverse radiolucent line. This software provides measurements of the lateral and medial cortex width and quantifies focal thickening of the femoral shaft. The beaking index can be displayed and trended across serial scans.

Optional sarcopenia software calculates values based on published definitions and thresholds using measured appendicular lean mass in combination with patient demographics and entered values of muscle strength and physical performance. These values may be useful to health care professionals in their management of sarcopenia.

GE Lunar DXA Bone Densitometers with enCORE version 17 are composed of a scanner and a computer. The scanner comprises the x-ray source and detector, the patient scan table, the mechanical drive system, and the lowest level portions of the control system. The scanner is in communication with the computer, which is a standard PC. The computer runs the enCORE software, and thus controls the scanner, acquires scan data from the scanner, stores and analyzes the data, and interacts with the human operator.

The enCORE software runs on the following list of GE Lunar DXA Bone Densitometers:
Lunar Prodigy Series: (Prodigy, Prodigy Compact, Prodigy Pro, Prodigy Pro Compact, Prodigy Primo, Prodigy Primo Compact, Prodigy Advance, Prodigy Advance Compact) Lunar iDXA Series: (iDXA, iDXA Advance, iDXA Pro, iDXA Forma, Lunar iDXA) DPX Series: DPX-NT & DPX-Bravo. The GE Lunar DXA Bone Densitometers with enCORE version 17 measure the bone mineral density (BMD), lean and fat tissue mass and calculate derivative values of bone mineral content (BMC), area, soft tissue mass, regional soft tissue mass, total soft tissue mass, fat free mass, regional/total soft tissue mass ratio, % fat, region % fat, total body % fat, Android % fat, Gynoid % fat, Android/Gynoid ratio (A/G ratio) and Body Mass Index (BMI).

The enCORE software is used on GE Lunar DXA bone densitometers. Release 17 of the enCORE software (enCORE 17 or enCORE 17.xx) includes some feature enhancements.

GE Lunar DXA Bone Densitometers with enCORE software were modified to include 2 new software features as well as additional cybersecurity enhancements:
A- The software will now expand upon its previously cleared Hip Axis Length (K011917) & AHA - Advanced Hip Assessment (K072664) by the addition of an optional feature for Atypical Femur Fracture (AFF): AFFs are stress or insufficiency fractures that occur in the subtrochanteric or diaphyseal regions of the femur and may be associated with long term bisphosphonate use. The new AFF feature brings minor changes to the anatomical sites to extend the DXA femur scan to include the distal femur shaft, measure the lateral and medial cortex width and quantify focal thickening of the lateral cortex along the femoral shaft. The beaking index can be displayed and trended across serial scans. The new AFF indication for use is equivalent to the Hologic Single Energy (SE) Femur Exams (K130277).

B- The software will now expand upon its previously cleared Total Body Composition (K071570) and RSMI - Relative Skeletal Muscle Index (K113286) by the addition of an optional feature for Sarcopenia calculation: Sarcopenia is a gradual loss of muscle mass and strength associated with aging and is a factor in the occurrence of frailty, falls, and fractures. Sarcopenia definitions have been published by four leading clinical working groups:
• International Working Group on Sarcopenia1:
• European Working Group on Sarcopenia in Older People2
• Asian Working Group for Sarcopenia3
• The Foundation for the National Institutes of Health Biomarkers Consortium Sarcopenia Project4

All definitions include DXA appendicular lean mass (ALM), a value measured from DXA total body scans, in combination with other measurements of muscle strength and physical performance (e.g. grip strength and gait speed). The new sarcopenia software feature allows the user to select one of the four definitions, extend the total body user interface so user can input muscle strength and physical performance data along with DXA ALM, and compare individual values against clinical thresholds for assessing sarcopenia.
The Sarcopenia feature provides a calculated sarcopenia classification using one of the four definitions above.

C- The enCORE version 17 software incorporates latest Microsoft security patches and other security enhancements.

Here's a breakdown of the acceptance criteria and study information for the GE Lunar DXA Bone Densitometers with enCORE version 17, based on the provided document.

It's important to note that the document is a 510(k) summary, which focuses on demonstrating substantial equivalence to a predicate device rather than a comprehensive clinical study report. Therefore, some of the requested information (especially regarding detailed study methodology, ground truth establishment, and MRMC studies) is not explicitly provided.

Acceptance Criteria and Reported Device Performance

The document doesn't explicitly state numerical acceptance criteria in a pass/fail format for the new Atypical Femur Fracture (AFF) and Sarcopenia features. Instead, it demonstrates equivalence to predicate devices and conformance to published definitions and measurement capabilities.

1. Technology: Dual-energy X-ray Absorptiometry (DXA) for proposed device is equivalent to GE Lunar DXA for AHA/HAL. For imaging distal shaft, it is equivalent to the Hologic predicate using single-energy image.
2. Exam Site: Must extend to Proximal Femur and distal shaft (similar to Hologic predicate).
3. Values Displayed: Quantitative measurements of cortical thickness (beaking index) for the larger femoral scan area, extending beyond previous AHA capabilities.
4. Method of Deformity Assessment: Combines visual assessment with quantitative measurements of cortical width (similar to Hologic's visual).
   Functional: Visualize focal reaction/thickening, measure lateral/medial cortex width, quantify focal thickening, display/trend beaking index. | Technology: Employs Dual-energy X-ray Absorptiometry (DXA), extending existing GE Lunar DXA femur scans to image the distal shaft. This is deemed equivalent to the predicate devices.
   Exam Site: Images both the Proximal Femur and distal shaft, identical to the Hologic predicate (K130277).
   Values Displayed: The enCORE Version 17 AFF feature measures cortical width (beaking index) over a larger femoral scan area compared to the Advanced Hip Assessment (AHA), demonstrating extended capability while remaining equivalent in principle of measurement.
   Method of Deformity Assessment: Combines a visual assessment with quantitative measurements of cortical width, which is considered equivalent to the predicate methods. The software "provides measurements of the lateral and medial cortex width and quantifies focal thickening of the femoral shaft. The beaking index can be displayed and trended across serial scans." |
   | Sarcopenia Software | Equivalence to Predicate Devices:
5. Technology: Dual-energy X-ray Absorptiometry (DXA) for proposed device must be identical to GE Lunar DXA predicates for Total Body composition and RSMI.
6. Anatomical Sites: Must be identical (Regional and Whole Body) to GE Lunar DXA predicates.
7. Values Displayed: Must provide BMD, lean/fat tissue mass, BMC, Area, Soft Tissue Mass (regional/total), Fat Free Mass, Ratios, %Fat (regional/total), RSMI, and new calculated values based on published definitions (FNIH, AWGS, EWGSOP, IWGS).
   Functional: Calculate values based on published definitions and thresholds using measured appendicular lean mass (ALM) in combination with patient demographics and entered values of muscle strength and physical performance. | Technology: Employs Dual-energy X-ray Absorptiometry (DXA), identical to its predicate devices for total body composition and RSMI.
   Anatomical Sites: Measures Regional and Whole Body, identical to its predicate devices.
   Values Displayed: Provides all values from the predicate devices (BMD, lean/fat tissue mass, BMC, etc.) and additionally calculates RSMI, FNIH, AWGS, EWGSOP, and IWGS values, directly incorporating the four published sarcopenia definitions. The software "calculates values based on published definitions and thresholds using measured appendicular lean mass in combination with patient demographics and entered values of muscle strength and physical performance." |
   | General Software Enhancements (including Cybersecurity) | Functional: Incorporate latest Microsoft security patches and other security enhancements. Maintain safety and effectiveness. | The enCORE version 17 software incorporates the latest Microsoft security patches and other security enhancements. No adverse impact on safety or effectiveness was reported. |

Study Information

1. Sample sizes used for the test set and the data provenance:

   • Test Set Sample Size: The document does not specify a distinct "test set" sample size for the AFF or Sarcopenia features from a clinical data perspective. It primarily relies on non-clinical evaluation (bench testing) for software verification and validation.
   • Data Provenance: Not explicitly stated for specific test data used. However, the basis for the Sarcopenia calculations is "published definitions and thresholds" from international working groups. For AFF, it extends existing DXA femur scans, implying the use of standard DXA imaging. As no new clinical studies were required, there is no mention of country of origin or retrospective/prospective data collection for a dedicated test set.

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

   • This information is not provided in the document, as no specific clinical test set requiring expert ground truth for adjudication is detailed. The Sarcopenia feature relies on published definitions from various clinical working groups as its "ground truth" for calculations. For AFF, the ground truth for measurement accuracy would likely be established through phantom or simulated data testing, not expert consensus on pathology.

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

   • None explicitly stated, as no clinical test set requiring human adjudication is described.

4. 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, an MRMC comparative effectiveness study was not done. The submission explicitly states: "The subject of this premarket submission, GE Lunar DXA Bone Densitometers with enCORE version 17, did not require clinical studies to support substantial equivalence." These features are presented as measurement and calculation tools, not as diagnostic AI aids that would improve human reader performance in an MRMC study.

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

   • Yes, in essence. The "non-clinical evaluation" and "bench testing" of the "AFF" software option and the "optional sarcopenia calculator software" represent standalone algorithm testing. The software performs measurements (cortical width) and calculations (sarcopenia classifications) and presents them to the user. The testing confirmed that "the design outputs met the design input requirements" and demonstrated "safety and effectiveness to the predicate devices."

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

   • For AFF (measurements): Implied ground truth from established anatomical measurements, potentially verified through phantom studies or comparison to existing validated measurement techniques. The document mentions "minor changes to the anatomical sites to extend the DXA femur scan," implying reliance on anatomical correctness and quantitative measurement accuracy.
   • For Sarcopenia (calculations): The "ground truth" for the calculations is based on published definitions and thresholds from four leading clinical working groups (International Working Group on Sarcopenia, European Working Group on Sarcopenia in Older People, Asian Working Group for Sarcopenia, and The Foundation for the National Institutes of Health Biomarkers Consortium Sarcopenia Project). The software's performance is validated against these established definitions.

7. The sample size for the training set:

   • Not applicable / not provided. The document describes software enhancements that perform measurements and calculations based on known anatomical structures and published clinical definitions. There is no indication of a machine learning or AI model that required a specific training set of images or data to learn from for these new features. The software's functionality appears to be rule-based or algorithmically determined parameters rather than learned from a training set.

8. How the ground truth for the training set was established:

   • Not applicable / not provided, as there is no mention of a training set for machine learning. The "ground truth" for the software's functionality is its adherence to anatomical principles and published clinical definitions.

Ask a specific question about this device

TBS iNsight is a software provided for use as a complement to both DXA analysis and clinical examination. It computes the antero-posterior spine DXA examination file and calculates a score (Trabecular Bone Score - TBS) that is compared to those of the age-matched controls. The TBS is derived from the texture of the DXA image and has been shown to be related to bone microarchitecture.

TBS iNsight provides as an option an assessment of 10-year fracture risk. It provides an estimate of 10-year probability of hip fracture and 10-year probability of a major osteoporotic fracture (clinical spine, forearm, hip or shoulder fracture). This estimate is based on the WHO's FRAX® Fracture Risk Assessment Tool, after adjustment for the TBS. The tool has been validated for Caucasian and Asian men and post-menopausal women between 40 and 90 years old.

TBS provides information independent of BMD value: it is used as a complement to the data obtained from the DXA analysis and the clinical examination (questioning by the clinician about patient history, bioassay of bone resorption markers . . . ).

The results can be used by a physician in conjunction with other clinical risk factors as an aid in the diagnosis of osteoporosis and other medical conditions leading to altered trabecular bone microarchitecture, and ultimately in the assessment of fracture risk.

The TBS score can assist the health care professional in monitoring the effect of treatments on patients across time.

Overall fracture risk will depend on many additional factors that should be considered before making diagnostic or therapeutic recommendations. The software does not disease, or recommend treatment regimens. Only the health care professional can make these judgments.

TBS iNsight is a software provided for use as a complement to to both DXA analysis and clinical examination. It computes the AP spine DXA examination file and calculates the Trabecular Bone Score (TBS) that is compared to those of the age-matched controls. The TBS is derived from the texture of the DXA image and has been shown to be related to bone microarchitecture.

It provides as an option an assessment of 10-year fracture risk, based on the WHO's FRAX® Tool, after adjustment for the TBS.

The results can be used by a physician in conjunction with other clinical risk factors as an aid in the diagnosis of osteoporosis and other medical conditions leading to altered trabecular bone microarchitecture, and ultimately in the assessment of fracture risk.

The TBS score can assist the health care professional in monitoring the effect of treatments on patients across time.

Here's an analysis of the acceptance criteria and study information for the TBS iNsight device, based on the provided text:

Acceptance Criteria and Device Performance

The document does not explicitly present a table of acceptance criteria with corresponding performance metrics in the format requested. However, it describes that the device's performance was evaluated through clinical studies against the predicate devices and known clinical associations. The core acceptance criteria appear to be substantial equivalence to the predicate devices and the demonstration that TBS is a valid clinical risk factor for fracture and can improve FRAX probability estimates.

Implied Acceptance Criteria and Reported Device Performance:

Study Details:

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

   • Clinical Aspects FRAX adjusted for TBS: "3 published studies and a report provided by the Frax group to Medimaps." The specific sample sizes for these studies are not provided in this document. The provenance of these studies is implied to be international given the reference to WHO's FRAX® and the Frax group. The nature (retrospective/prospective) of these studies is not specified.
   • Clinical Aspects TBS iNstats:
     • Canadian cohort: "29,000 Canadian women" (retrospective, likely).
     • Studies for US applicability: "Two studies" but their sample sizes are not provided. The provenance is implied to be US for the applicability demonstration.
   • Normative Data: A "US clinical study" including "Non-Hispanic white US women and men. Non-Hispanic black US women and men and Mexican women and men, aged 20 to 85 years." The data was "provided from the Nhanes IV database." The specific sample size for this US clinical study is not provided within this summary. NHANES IV data is retrospective/cross-sectional.

2. Number of Experts Used to Establish the Ground Truth for the Test Set and Their Qualifications:
   The document does not specify the number or qualifications of experts used to establish ground truth for the clinical studies mentioned. It refers to "published studies" and the "Frax group," implying established methodologies and expert involvement, but direct details are missing.

3. Adjudication Method for the Test Set:
   The document does not specify any adjudication methods (e.g., 2+1, 3+1) used for the clinical studies or for establishing ground truth.

4. 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 provide an effect size for human readers improving with AI vs. without AI assistance. The device is described as a "complement" to DXA analysis and clinical examination, implying assistance but not a direct comparison of human performance with and without the tool.

5. Standalone (Algorithm Only Without Human-in-the-Loop Performance) Study:
   The document does not explicitly state if a standalone algorithm performance study was done. The focus is on its role as a "complement" to clinical examination and DXA analysis, and its ability to adjust FRAX scores, which inherently involve human interpretation and clinical data.

6. Type of Ground Truth Used:
   The ground truth implicitly used in the clinical studies relates to:

   • Fracture events: The studies evaluated the device's ability to predict fracture risk, so actual fracture outcomes (e.g., hip fracture, major osteoporotic fracture) served as ground truth.
   • Bone microarchitecture: TBS is described as "related to bone microarchitecture," suggesting that underlying microarchitectural properties (likely measured through other means in the mentioned "published studies") were part of the ground truth for validating the TBS score itself.
   • Normative population data: For the normative data study, the "Nhanes IV database" served as the reference for US population TBS values.

7. Sample Size for the Training Set:
   The document does not disclose the sample size used for training the TBS iNsight algorithm. It details evaluation studies, not algorithm development or training specifics.

8. 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, as it does not discuss the training phase of the algorithm.

Ask a specific question about this device