(133 days)
Rayvolve is a computer-assisted detection and diagnosis (CAD) software device to assist radiologists and emergency physicians in detecting fractures during the review of radiographs of the musculoskeletal system. Rayvolve is indicated for adults only (≥ 22 years old). Rayvolve is indicated for radiographs of the following industry-standard radiographic views and study types.
Study Type (Anatomic Area of interest) / Radiographic Views supported: Ankle / Frontal, Lateral,Oblique Clavicle / Frontal Elbow / Frontal, Lateral Forearm / Frontal, Lateral Hip / Frontal, Frog Leg Lateral Humerus / Frontal, Lateral Knee / Frontal, Lateral Pelvis / Frontal Shoulder / Frontal, Lateral, Axillary Tibia/fibula / Frontal. Lateral Wrist / Frontal, Lateral, Oblique Hand / Frontal, Lateral Foot / Frontal, Lateral
*For the purposes of this table, "Frontal" is considered inclusive of both posteroanterior (PA) and anteroposterior (AP) views.
+Definitions of anatomic area of interest and radiographic views are consistent with the American College of Radiology (ACR) standards and guidelines.
The medical device is called Rayvolve. It is a standalone software that uses deep learning techniques to detect and localize fractures on osteoarticular X-rays. Rayvolve is intended to be used as an aided-diagnosis device and does not operate autonomously. It is intended to work in combination with Picture Archiving and communication system (PACS) servers. When remotely connected to a medical center PACS server, Rayvolve directly interacts with the DICOM files to output the prediction (potential presence of fracture). Rayvolve does not intend to replace medical doctors. The instructions for use are strictly and systematically transmitted to each user and used to train them on Rayvolve's use.
Here's a summary of the acceptance criteria and the study proving the device meets them, based on the provided text:
1. Table of Acceptance Criteria and Reported Device Performance
| Acceptance Criterion (Primary Endpoint) | Reported Device Performance | Study Type |
|---|---|---|
| Standalone Study: Characterize the detection accuracy of Rayvolve for detecting adult patient fractures (AUC, Sensitivity, Specificity) | AUC: 0.98607 (95% CI: 0.98104; 0.99058) Sensitivity: 0.98763 (95% CI: 0.97559; 0.99421) Specificity: 0.88558 (95% CI: 0.87119; 0.89882) | Standalone Bench Testing |
| MRMC Study: Diagnostic accuracy of readers aided by Rayvolve is superior to unaided readers (AUC of ROC curve comparison). H0: T-test for p (no statistical difference) > 0.05; H1: T-Test for p (statistical difference) < 0.05 | Reader AUC significantly improved from 0.84602 (unaided) to 0.89327 (aided), a difference of -0.04725 (95% CI: 0.03376; 0.061542), p=0.0041 (indicating superiority of aided reads). | Clinical Reader Study (MRMC) |
Secondary Endpoints (Standalone Study): Demonstrate Rayvolve's ability to perform across different subgroup variables (gender, age, anatomic region, machine acquisition, machine view, weight-bearing, complex & uncommon cases).
Reported Performance: Rayvolve performs with high accuracy across study types (including anatomic areas of interest, views, patient age and sex, and machine) and across potential confounders such as different X-ray manufacturers. Specific AUCs for various subgroups are provided in the document and demonstrate high performance.
Secondary Endpoints (MRMC Study): Report the sensitivity and specificity of Rayvolve-aided and unaided reads.
Reported Performance:
- Unaided Sensitivity: 0.86561 (95% CI: 0.84859, 0.88099)
- Aided Sensitivity: 0.9554 (95% CI: 0.94453, 0.96422)
- Unaided Specificity: 0.82645 (95% CI: 0.81187, 0.84012)
- Aided Specificity: 0.83116 (95% CI: 0.81673, 0.84467)
2. Sample Size Used for the Test Set and Data Provenance
- Standalone Test Set Size: 2626 radiographs
- Provenance: Data were acquired from 4 sites in the US.
- MRMC Test Set Size: 186 cases (equivalent to 186 patients)
- Provenance: Data were acquired from 4 sites in the US. All radiographs in the validation study were independent of the training data.
3. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications of Those Experts
- Number of Experts: Three US board-certified MSK radiologists.
- Qualifications: "US board-certified MSK radiologists." The document does not specify their years of experience.
4. Adjudication Method for the Test Set
- MRMC Study Ground Truth Adjudication: A panel of three US board-certified MSK radiologists reviewed each case to provide ground truth binary labeling (presence or absence of fracture) and localization information. While not explicitly stated as "2+1" or "3+1", the use of a panel of three experts strongly suggests a consensus-based adjudication, where if at least two agreed, that would likely be the accepted ground truth.
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
- Yes, an MRMC comparative effectiveness study was done.
- Effect Size of Improvement with AI Assistance:
- Reader AUC significantly improved by 0.04725 (from 0.84602 to 0.89327).
- Reader sensitivity improved by 0.08979 (from 0.86561 to 0.9554).
- Reader specificity improved by 0.00471 (from 0.82645 to 0.83116).
6. If a Standalone (i.e. algorithm only without human-in-the-loop performance) was done
- Yes, a standalone performance assessment was conducted.
- The results are detailed in the table above: AUC of 0.98607, sensitivity of 0.98763, and specificity of 0.88558.
7. The type of ground truth used
- Expert Consensus: For both standalone (bench testing) and MRMC (clinical data), the ground truth was established by human experts, specifically "a panel of three US board-certified MSK radiologists" for the MRMC study, who provided binary labeling indicating the presence or absence of fracture and localization information.
8. The sample size for the training set
- The document states: "The dataset used to develop the Rayvolve deep learning algorithm is composed of labeled osteoarticular radiographs." and "Rayvolve training set has been established before the collection of the standalone and MRMC studies data."
- However, the exact sample size for the training set is not explicitly provided in the given text.
9. How the ground truth for the training set was established
- The document states: "The dataset used to develop the Rayvolve deep learning algorithm is composed of labeled osteoarticular radiographs."
- Similar to the test sets, it is implied that the ground truth for the training set was established through expert labeling, given the nature of a "labeled" dataset for deep learning in medical image analysis. However, the specific process or number/qualifications of experts for the training set ground truth are not explicitly detailed in the provided text.
{0}------------------------------------------------
Image /page/0/Picture/0 description: The image contains the logo of the U.S. Food & Drug Administration (FDA). On the left is the Department of Health & Human Services logo. To the right of that is the FDA logo, which is a blue square with the letters "FDA" in white. To the right of the blue square is the text "U.S. FOOD & DRUG ADMINISTRATION" in blue.
AZmed SAS % Patricia Massako QARA Manager 6 rue Leonard de Vinci Laval, Pays de la Loire 53000 FRANCE
June 2, 2022
Re: K220164
Trade/Device Name: Rayvolve Regulation Number: 21 CFR 892.2090 Regulation Name: Radiological computer assisted detection and diagnosis software Regulatory Class: Class II Product Code: QBS Dated: April 12, 2022 Received: April 29, 2022
Dear Patricia Massako:
We have reviewed your Section 510(k) premarket notification of intent to market the device referenced above and have determined the device is substantially equivalent (for the indications for use stated in the enclosure) to legally marketed predicate devices marketed in interstate commerce prior to May 28, 1976, the enactment date of the Medical Device Amendments, or to devices that have been reclassified in accordance with the provisions of the Federal Food, Drug, and Cosmetic Act (Act) that do not require approval of a premarket approval application (PMA). You may, therefore, market the device, subject to the general controls provisions of the Act. Although this letter refers to your product as a device, please be aware that some cleared products may instead be combination products. The 510(k) Premarket Notification Database located at https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/pmn.cfm identifies combination product submissions. The general controls provisions of the Act include requirements for annual registration, listing of devices, good manufacturing practice, labeling, and prohibitions against misbranding and adulteration. Please note: CDRH does not evaluate information related to contract liability warranties. We remind you, however, that device labeling must be truthful and not misleading.
If your device is classified (see above) into either class II (Special Controls) or class III (PMA), it may be subject to additional controls. Existing major regulations affecting your device can be found in the Code of Federal Regulations, Title 21, Parts 800 to 898. In addition, FDA may publish further announcements concerning your device in the Federal Register.
Please be advised that FDA's issuance of a substantial equivalence determination does not mean that FDA has made a determination that your device complies with other requirements of the Act or any Federal statutes and regulations administered by other Federal agencies. You must comply with all the Act's requirements, including, but not limited to: registration and listing (21 CFR Part 807); labeling (21 CFR Part 801); medical device reporting of medical device-related adverse events) (21 CFR 803) for
{1}------------------------------------------------
devices or postmarketing safety reporting (21 CFR 4, Subpart B) for combination products (see https://www.fda.gov/combination-products/guidance-regulatory-information/postmarketing-safety-reportingcombination-products); good manufacturing practice requirements as set forth in the quality systems (QS) regulation (21 CFR Part 820) for devices or current good manufacturing practices (21 CFR 4, Subpart A) for combination products; and, if applicable, the electronic product radiation control provisions (Sections 531-542 of the Act); 21 CFR 1000-1050.
Also, please note the regulation entitled, "Misbranding by reference to premarket notification" (21 CFR Part 807.97). For questions regarding the reporting of adverse events under the MDR regulation (21 CFR Part 803), please go to https://www.fda.gov/medical-device-safety/medical-device-reportingmdr-how-report-medical-device-problems.
For comprehensive regulatory information about mediation-emitting products, including information about labeling regulations, please see Device Advice (https://www.fda.gov/medicaldevices/device-advice-comprehensive-regulatory-assistance) and CDRH Learn (https://www.fda.gov/training-and-continuing-education/cdrh-learn). Additionally, you may contact the Division of Industry and Consumer Education (DICE) to ask a question about a specific regulatory topic. See the DICE website (https://www.fda.gov/medical-device-advice-comprehensive-regulatoryassistance/contact-us-division-industry-and-consumer-education-dice) for more information or contact DICE by email (DICE@fda.hhs.gov) or phone (1-800-638-2041 or 301-796-7100).
Sincerely.
Jessica Lamb, Ph.D. Assistant Director Imaging Software Team DHT8B: Division of Radiological Imaging Devices and Electronic Products OHT8: Office of Radiological Health Office of Product Evaluation and Quality Center for Devices and Radiological Health
Enclosure
{2}------------------------------------------------
Indications for Use
510(k) Number (if known) K220164
Device Name Rayvolve
Indications for Use (Describe)
Rayvolve is a computer-assisted detection and diagnosis (CAD) software device to assist radiologists and emergency physicians in detecting fractures during the review of the musculoskeletal system. Rayvolve is indicated for adults only ( ≥ 22 years old ). Rayvolve is indicated for radiographs of the following industry-standard radiographic views and study types.
Study Type (Anatomic Area of interest) / Radiographic Views supported: Ankle / Frontal, Lateral,Oblique Clavicle / Frontal Elbow / Frontal, Lateral Forearm / Frontal, Lateral Hip / Frontal, Frog Leg Lateral Humerus / Frontal, Lateral Knee / Frontal, Lateral Pelvis / Frontal Shoulder / Frontal, Lateral, Axillary Tibia/fibula / Frontal. Lateral Wrist / Frontal, Lateral, Oblique Hand / Frontal, Lateral Foot / Frontal, Lateral
*For the purposes of this table, "Frontal" is considered inclusive of both posteroanterior (PA) and anteroposterior (AP) views.
+Definitions of anatomic area of interest and radiographic views are consistent with the American College of Radiology (ACR) standards and guidelines.
Type of Use (Select one or both, as applicable)
X Prescription Use (Part 21 CFR 801 Subpart D)
Over-The-Counter Use (21 CFR 801 Subpart C)
CONTINUE ON A SEPARATE PAGE IF NEEDED.
This section applies only to requirements of the Paperwork Reduction Act of 1995.
DO NOT SEND YOUR COMPLETED FORM TO THE PRA STAFF EMAIL ADDRESS BELOW.
The burden time for this collection of information is estimated to average 79 hours per response, including the time to review instructions, search existing data sources, gather and maintain the data needed and complete and review the collection of information. Send comments regarding this burden estimate or any other aspect of this information collection, including suggestions for reducing this burden, to:
Department of Health and Human Services Food and Drug Administration Office of Chief Information Officer Paperwork Reduction Act (PRA) Staff PRAStaff(@fda.hhs.gov
"An agency may not conduct or sponsor, and a person is not required to respond to, a collection of information unless it displays a currently valid OMB number."
{3}------------------------------------------------
Image /page/3/Picture/0 description: The image shows a document header with the AZMed logo on the left. The text "510(k) SUMMARY" is in the center. The text "Page 1 on 15" is on the right side of the image.
K220164
510(k) SUMMARY
{4}------------------------------------------------
Image /page/4/Picture/0 description: The image contains the logo for AZmed. The logo consists of a blue hexagon with a white stylized "AZ" inside of it. To the right of the hexagon is the text "AZMED" in a dark blue sans-serif font.
Content
| 1 | Submitter | 3 | |
|---|---|---|---|
| 2 | Device identification | 3 | |
| 3 | Predicate device | 3 | |
| 4 | Device description | 4 | |
| 5 | Intended use/Indication for use | 4 | |
| 6 | Substantial equivalence Discussion | 5 | |
| 7 | Performance data | 8 | |
| 7.1 | Software verification and validation testing | 8 | |
| 7.2 | Biocompatibility testing | 8 | |
| 7.3 | Electrical safety and electromagnetic compatibility (EMC) | 8 | |
| 7.4 | Bench Testing | 8 | |
| 7.4.1 | Acceptance criteria / Endpoints | 8 | |
| 7.4.2 | Data & Patient information | 9 | |
| 7.4.3 | Collected images | 10 | |
| 7.4.4 | Results | 10 | |
| 7.5 | Animal testing | 13 | |
| 7.6 | Clinical data | 13 | |
| 8 | CONCLUSION | 15 |
{5}------------------------------------------------
Image /page/5/Picture/0 description: The image shows the logo for AZ Med. The logo consists of a blue hexagon with the letters "AZ" inside it, and the word "MED" to the right of the hexagon. The letters "AZ" are stylized and appear to be connected. The word "MED" is in a simple, sans-serif font.
1 Submitter
K220164
AZmed 6 rue Léonard de Vinci 53000, Laval, France
Contact Person: Patricia Massako – Quality and regulatory Manager Phone: +336 25 84 69 88
Prepared date: 22nd April 2022
2 Device identification
| Name of theDevice | Common orUsual Name | Regulatorysection | Classification | ProductCode | Panel |
|---|---|---|---|---|---|
| Rayvolve | Rayvolve | 21 CFR892.2090 | Class II | QBS | 90(Radiology) |
Predicate device 3
The legally marketed device for which AZmed is claiming equivalence is identified as follows:
| Manufacturer | Product Name | 510K Number |
|---|---|---|
| Imagen Technologies | FractureDetect (FX) | K193417 |
{6}------------------------------------------------
Image /page/6/Picture/0 description: The image shows a logo for AZMED on the left side of the image. In the middle of the image, the text "510 k summary" is present. On the right side of the image, the text "Page 4 on 15" is present. The image appears to be a header or footer of a document.
Device description ব
The medical device is called Rayvolve. It is a standalone software that uses deep learning techniques to detect and localize fractures on osteoarticular X-rays. Rayvolve is intended to be used as an aided-diagnosis device and does not operate autonomously. It is intended to work in combination with Picture Archiving and communication system (PACS) servers. When remotely connected to a medical center PACS server, Rayvolve directly interacts with the DICOM files to output the prediction (potential presence of fracture). Rayvolve does not intend to replace medical doctors. The instructions for use are strictly and systematically transmitted to each user and used to train them on Rayvolve's use.
The dataset used to develop the Rayvolve deep learning algorithm is composed of labeled osteoarticular radiographs. The osteoarticular radiographs have been collected from multiple centers (different types of medical imaging centers: public hospitals, private clinics, generalist medical imaging centers, and musculoskeletal medical imaging centers) from different countries (France, Israel, Germany, Switzerland, Belgium, United-Kingdom, Argentina, Brazil, and Nigeria) to have the largest diversity and variety. This diversity allows the Rayvolve algorithm to have a high generalization ability.
5 Intended use/Indication for use
Rayvolve is a computer-assisted detection and diagnosis (CAD) software device to assist radiologists and emergency physicians in detecting fractures during the review of radiographs of the musculoskeletal system. Rayvolve is indicated for adults only (≥ 22 years old). Rayvolve is indicated for radiographs of the following industry-standard radiographic views and study types.:
| Study type (AnatomicArea of interest) | Radiographic Viewssupported* |
|---|---|
| Ankle | Frontal, Lateral, Oblique |
| Clavicle | Frontal |
| Elbow | Frontal, Lateral |
| Forearm | Frontal, Lateral |
| Hip | Frontal, Frog Leg Lateral |
| Humerus | Frontal, Lateral |
{7}------------------------------------------------
| Image: AZMED logo | 510 k summary | Page 5 on 15 |
|---|---|---|
| ------------------- | --------------- | -------------- |
| Knee | Frontal, Lateral |
|---|---|
| Pelvis | Frontal |
| Shoulder | Frontal, Lateral, Axillary |
| Tibia/fibula | Frontal, Lateral |
| Wrist | Frontal, Lateral, Oblique |
| Hand | Frontal, Lateral |
| Foot | Frontal, Lateral |
Table 1 Rayvolve indication for use - Anatomical regions
*For this table, "Frontal" is considered inclusive of both posteroanterior (PA) and anteroposterior (AP) views.
- Definitions of an anatomic area of interest and radiographic views are consistent with the American College of Radiology (ACR) standards and guidelines.
Substantial equivalence Discussion 6
The comparison chart below provides evidence to facilitate the substantial equivalence determination between Rayvolve to the predicate device (K193417) concerning the intended use, technological characteristics, and principle of operation.
| FractureDetect(FX) | Rayvolve Subjectdevice | Comparison | |
|---|---|---|---|
| Number | K193417 | TBD | / |
| Applicant | Imagen Technologies | AZmed | / |
| Device Name | FractureDetect (FX) | Rayvolve | / |
| ClassificationRegulation | 892.2090 | 892.2090 | Same |
| Product Code | QBS | QBS | Same |
| Intended use/Indication foruse | FractureDetect (FX)is a computer-assisted detection anddiagnosis (CAD)software device toassist clinicians indetecting fracturesduring the review ofradiographs of themusculoskeletalsystem. FX isindicated for adultsonly. | Rayvolve is acomputer-assisteddetection anddiagnosis (CAD)software device toassist radiologistsand emergencyphysicians indetecting fracturesduring the review ofradiographs of themusculoskeletalsystem. Rayvolve isindicated for adultsonly (≥ 22 years old). | Same |
| ImageModality | X-ray | X-ray | Same |
| Study Type(AnatomicAreas ofInterest) | AnkleClavicleElbowFemurForearmHipHumerusKneePelvisShoulderTibia / FibulaWrist | AnkleClavicleElbowForearmHipHumerusKneePelvisShoulderTibia / FibulaWristHandFoot | Rayvolve covers 2 moreanatomical regions than FX;but the intended use issimilar since both FX andRayvolve are intended toidentifyfracturesinradiographs, and all thoseanatomicalregionsareincluded in the definition ofanatomic area of interest andradiographicviewsconsistentlywiththeCollegeAmericanofRadiology (ACR) standardand guidelines. |
| ClinicalFinding | Fracture | Fracture | Same |
{8}------------------------------------------------
{9}------------------------------------------------
| Image: AZMed logo | 510 k summary | Page 7 on 15 |
|---|---|---|
| ------------------- | --------------- | -------------- |
| PatientPopulation | Adults ≥ 22 years ofage | Adults ≥ 22 years ofage | Same |
|---|---|---|---|
| Intended User | Clinicians | Clinicians (MSK andnon-MSK radiologistand emergencyphysicians) | Same |
| MachineLearningMethodology | Supervised DeepLearning | Supervised DeepLearning | Same |
| Platform | Secure localprocessing anddelivery of DICOMimages | Secure localprocessing anddelivery of DICOMimages | Same |
| Image Source | DICOM node (e.g.,imaging device,intermediate DICOMnode, PACS system,etc.) | DICOM node (e.g.,imaging device,intermediate DICOMnode, PACS system,etc.) | Same as FX |
| ImageViewing | PACS system, imageannotations toggledon or off | PACS system, imageannotations made ona copy of the originalimage | Same, with a copy of theoriginal image |
| Privacy | HIPAA compliant | HIPAA compliant | Same |
AZmed claims the substantial equivalence of Rayvolve with the predicate FX based on the functional principle of the software algorithms, the same technological characteristics, and the intended purpose of the software algorithm.
{10}------------------------------------------------
Image /page/10/Picture/0 description: The image shows a document header. On the left is the "AZmed" logo. In the center of the header is the text "510 k summary". On the right side of the header is the text "Page 8 on 15".
Performance data 7
Software verification and validation testing 7.1
The device's software development, verification, and validation have been carried out following FDA guidelines. The software was tested against the established software design specification for each test plan to assure the device's performance as intended. The device hazard analysis was completed and risk control was implemented to mitigate identified hazards. The testing results support that all the software specifications have met the acceptance criteria of each module and interaction of processes. Rayvolve device passes all the testing and supports the claims of substantial equivalence with the predicate.
Validation activities included a usability study of Rayyolve under normal conditions for use. The study demonstrated:
- -Non-invasive usability because users' habits are unchanged
- -Comprehension of the instructions for use provided with the device
7.2 Biocompatibility testing
There are no direct or indirect patient-contacting components of Rayvolve. Therefore, patient contact information is not needed for this device.
7.3 Electrical safety and electromagnetic compatibility (EMC)
Rayvolve is a software-only device, therefore: electrical safety and EMC testing are not applicable.
7.4 Bench Testing
AZmed conducted a standalone performance assessment on 2626 radiographs for all the study types (anatomic areas of interest) and views in the indication for Use.
7.4.1 Acceptance criteria / Endpoints
Regarding the performance standalone study:
- The primary endpoint of the standalone study is to characterize the detection accuracy of Rayvolve for detecting adult patient fractures.
- -The secondary endpoint is to demonstrate Rayvolve's ability to perform across different subgroup variables. More precisely, the goal is to compute Rayvolve AUC, sensitivity, and specificity for all the potential and relevant observable subgroups such as gender, age, anatomic region, machine acquisition, machine view, as well as Rayvolve performances depending on weight-bearing and complex & uncommon cases.
{11}------------------------------------------------
Image /page/11/Picture/0 description: The image shows the logo for AZMED on the left side of the image. The right side of the image contains the text "510 k summary" and "Page 9 on 15". The logo is a blue hexagon with the letters AZ inside of it. The text is in a sans-serif font.
Regarding the MRMC study:
- The primary objective of the clinical reader study is to determine whether the diagnostic accuracy of readers aided by Rayvolve is superior to reader accuracy when unaided by Rayvolve, as determined by the AUC of the ROC curve: H0: T-test for p (no statistical difference) > 0.05; H1: T-Test for p (statistical difference) < 0.05
- -The secondary objective is to report the sensitivity and the specificity of the Rayvolveaided and unaided reads.
7.4.2 Data & Patient information
For both standalone and MRMC studies, the subgroups have been determined based on the data set composed with the following inclusion criteria:
- -De-identified radiographs
- -Frontal, LAT, Oblique and Axillary views
- Adult patient, minimum of 22 years of age -
Regarding the performance standalone study:
Here are the different subgroups/confounders evaluated:
- -Gender
- -Age
- Anatomic region -
- Machine acquisition -
- -Machine view
- Weight-bearing radiographs -
- -Complex and uncommon radiographs
Here is other information about the patients:
- The 2626 radiographs (samples) of the performance study were taken from 851 patients -
- -The number of samples is the number of radiographs, thus the study contains 2626 samples.
- Gender split: 440 female patients, 411 male patients. -
- Age split: 468 patients between 22 and 60 (mean: 38 y.o., std: 12 y.o.), 383 patients above 60 (mean: 81 y.o., std: 18 y.o.)
- -Ethnicity: no information was available regarding ethnicity.
{12}------------------------------------------------
Image /page/12/Picture/0 description: The image shows the logo for AZMED on the left side of the image. The middle of the image says "510 k summary". The right side of the image says "Page 10 on 15".
Regarding the MRMC study:
Here are the different subgroups/confounders evaluated:
- Gender
- -Age
- -Machine acquisition
Here is other information about the patients:
- The 186 cases (samples) were taken from 186 patients Regarding the MRMC study, a case is equivalent to a patient. Thus, the study contains 186 samples.
- -Gender split: 96 male patients, 90 female patients.
- -Age split: 79 patients between 22 and 60, 107 patients above 60.
- -Ethnicity: no information was available regarding ethnicity.
7.4.3 Collected images
To ensure the independence between data for both standalone and MRMC studies, and Rayvolve training data, no radiograph in the validation (bench or clinical testing) study is part of Rayvolve's training set. Rayvolve training set has been established before the collection of the standalone and MRMC studies data.
Data were acquired from 4 sites in US.
7.4.4 Results
AZmed conducted a standalone performance assessment on 2626 radiographs for all the study types (anatomic areas of interest) and views in the Indications for Use. The results of standalone testing demonstrated that Rayvolve detects fractures of the musculoskeletal system radiographs with high sensitivity (0.98763, 95% Wilson's Confidence Interval (CI): 0.97559; 0.99421), high specificity (0.88558; 95% Wilson's CI: 0.87119; 0.89882) and high Area Under The Curve (AUC) of the Receiver Operating Characteristic (ROC) (0.98607; 95% Bootstrap CI: 0.98104; 0.99058). Additionally, the results demonstrated that Rayvolve performs with high accuracy across study types (anatomic areas of interest, views, patient age and sex and machine) and across potential confounders such as different x-ray manufacturers.
The results of the standalone testing demonstrated that Rayvolve detects fractures of the musculoskeletal system radiographs with high AUC across the following subgroups:
{13}------------------------------------------------
| Image: AZMED logo | 510 k summary | Page 11 on 15 |
|---|---|---|
| ------------------- | --------------- | --------------- |
| AUC (Bootstrapped CI) | |
|---|---|
| All (2626) | 0.98607(0.98104; 0.99058) |
| Anatomic Area (Nb of images) | AUC (Bootstrapped CI) |
|---|---|
| Ankle (232) | 0.99137 (0.98374; 0.99727) |
| Clavicle (171) | 0.97806 (0.94626; 0.99761) |
| Elbow (192) | 0.9964 (0.99059; 1.0) |
| Forearm (157) | 0.9953 (0.98909; 0.99937) |
| Humerus (181) | 0.9955 (0.98960; 0.99943) |
| Hip (198) | 0.95821 (0.93239; 0.98014) |
| Knee (239) | 0.97742 (0.95084; 0.99592) |
| Pelvis (149) | 0.97676 (0.95241; 0.99638) |
| Shoulder (150) | 0.97814 (0.94147; 0.99958) |
| Tibia/Fibula (232) | 0.98285 (0.95925; 0.9978) |
| Wrist (225) | 0.99567 (0.99126; 0.99897) |
| Hand (252) | 0.99552 (0.99074; 0.99898) |
| Foot (248) | 0.99162 (0.98238; 0.99823) |
| Gender (Nb of images) | AUC (Bootstrapped CI) |
|---|---|
| Male (1306) | 0.98822 (0.98186; 0.99409) |
| Female (1320) | 0.98395 (0.97589; 0.99101) |
View (Nb of images) AUC (Bootstrapped CI)
{14}------------------------------------------------
| AZMED | 510 k summary | Page 12 on 15 |
|---|---|---|
| ------- | --------------- | --------------- |
| Frontal (1279) | 0.97872 (0.96845; 0.98805) |
|---|---|
| Lateral (1033) | 0.99218 (0.98903; 0.99477) |
| Oblique (268) | 0.9958 (0.98979; 0.99977) |
| Axillary (46) | 0.99675 (0.98734; 1.0) |
| Age (Nb of images) | AUC (Bootstrapped CI) |
| 22-60 (1403) | 0.99049 (0.98359; 0.99598) |
| Machine (Nb of images) | AUC (Bootstrapped CI) |
|---|---|
| GEHC Discovery XR 656 (1234) | 0.98482 (0.97920; 0.99264) |
| Philips DigitalDiagnost (840) | 0.98635 (0.97657; 0.99345) |
| Carestream Health DRX-1 (552) | 0.98842 (0.97754; 0.99689) |
0.98102 (0.97487; 0.98941)
Abbreviations: AUC=Area under the Curve; CI=confidence interval.
60 (1223)
Four additional studies (comprising a total of 3574 cases) were used to demonstrate the generalizability of Rayvolve across multiple imaging devices. It was demonstrated that Rayvolve had reached similar and stable performance across different medical imaging acquisition device providers (Siemens Healthineers, Philips Healthcare, Carestream, GE Healthcare, Fujifilm, MinXray, Hologic, Shimadzu, Agfa, Duet, Primax, Kodak, Samsung and Thales).
Additionally, the performance of Rayvolve was validated for potential confounders including weight-bearing and non-weight bearing bone fractures, complex and uncommon fractures different X-ray Machine providers.
| Particular groups (Nb of images) | AUC (Bootstrapped CI) |
|---|---|
| Complex & uncommon (547) | 0.96102 (0.95223; 0.99615) |
{15}------------------------------------------------
| 17 0 0 0 7 MED 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 | 510 k summary | Page 13 on 15 |
|---|
| Non complex & uncommon (2079) | 0.99607 (0.98862; 0.99701) |
|---|---|
| weight-bearing (1298) | 0.98059 (0.96162; 0.99458) |
| Non-weight-bearing (1328) | 0.99143 (0.97916; 0.99912) |
7.5 Animal testing
Not applicable. Animal studies are not necessary to establish the substantial equivalence of this device.
7.6 Clinical data
AZmed conducted a fully-crossed multiple readers, multiple case (MRMC) retrospective reader study to determine the impact of Rayvolve on reader performance in diagnosing fractures.
The primary objective of the study was to determine whether the diagnostic accuracy of readers aided by Rayvolve ("Rayvolve-aided") is superior to the diagnostic accuracy of readers unaided by Rayvolve ("Rayvolve-unaided") as determined by the AUC of the Receiver Operating Characteristic (ROC) Curve. The secondary objective is to report the sensitivity and the specificity of the Rayvolve-aided and unaided reads.
24 clinical readers each evaluated 186 cases in Rayvolve's indication for use under both Rayvolve-aided and Rayvolve-unaided conditions. Each case had been previously evaluated by a panel of three US board-certified MSK radiologists to provide ground truth binary labeling indicating the presence or absence of fracture and the localization information for fractures. The MRMC study consisted of two independent reading sessions separated by a washout period of at least one month to avoid memory bias.
For each case, each reader was required to provide a binary determination of the presence or absence of a fracture and also to draw a bounding box around each fracture on the image to determine the localization of each fracture.
In addition to this binary decision of the readers regarding the presence or absence of fracture, each reader should also provide a report score with an ordinal value.
This report score has been collected for every case and for every reader with and without aid of Rayvolve device. The report score has been used for ROC data.
{16}------------------------------------------------
Image /page/16/Picture/0 description: The image shows the title of a document. The left side of the image shows the logo for AZmed. The middle of the image shows the text "510 k summary". The right side of the image shows the text "Page 14 on 15".
The results of the study found that the diagnostic accuracy of readers in the intended use population is superior when aided by Rayvolve than when unaided by Rayvolve, as measured at the task of fracture detection using the AUC of the ROC curve as calculated by the DBM modeling approach.
Image /page/16/Figure/2 description: The image is a plot comparing the ROC curves of unaided and aided readers in a clinical study. The x-axis represents 1-Specificity, ranging from 0.0 to 1.0, while the y-axis represents Sensitivity, also ranging from 0.0 to 1.0. The plot shows two curves: one for unaided readers, represented by a dashed red line, and another for aided readers, represented by a solid blue line. The aided readers curve is generally higher than the unaided readers curve, indicating better performance.
Clinical Reader Study Results Rayvolve-Aided vs Rayvolve-Unaided ROC Curves
In particular, the study results demonstrated:
- Reader AUC was significantly improved from 0.84602 to 0.89327, a difference of -0.04725 (95% CI: 0.03376; 0.061542), across the 186 cases within Rayvolve's Indications for Use, spanning 13 study types (anatomic areas of interest) (p=0.0041).
- -Reader sensitivity was significantly improved from 0.86561 (95% Wilson's CI: 0.84859, 0.88099) to 0.9554 (95% Wilson's CI: 0.94453, 0.96422)
- Reader specificity was improved from 0.82645 (95% Wilson's CI: 0.81187, 0.84012) to 0.83116 (95% Wilson's CI: 0.81673, 0.84467)
{17}------------------------------------------------
Image /page/17/Picture/0 description: The image shows a document page with the AZmed logo on the left. The text "510 k summary" is in the center of the page. The text "Page 15 on 15" is on the right side of the page, indicating it is the last page of the document.
CONCLUSION 8
Both the proposed device (Rayvolve) and the predicate device (FX) are computer-assisted detection and diagnostic devices that accept as input radiographs in DICOM format and use machine learning techniques to identify and highlight fractures. The overall design of the software and the basic functionality that it provides to the end-user are the same. The differences in technological characteristics do not raise different questions of safety and effectiveness. The results of standalone and clinical studies demonstrate that Rayvolve performs according to the specifications and meets user needs and intended use and that Rayvolve can be found to be substantially equivalent to FX.
§ 892.2090 Radiological computer-assisted detection and diagnosis software.
(a)
Identification. A radiological computer-assisted detection and diagnostic software is an image processing device intended to aid in the detection, localization, and characterization of fracture, lesions, or other disease-specific findings on acquired medical images (e.g., radiography, magnetic resonance, computed tomography). The device detects, identifies, and characterizes findings based on features or information extracted from images, and provides information about the presence, location, and characteristics of the findings to the user. The analysis is intended to inform the primary diagnostic and patient management decisions that are made by the clinical user. The device is not intended as a replacement for a complete clinician's review or their clinical judgment that takes into account other relevant information from the image or patient history.(b)
Classification. Class II (special controls). The special controls for this device are:(1) Design verification and validation must include:
(i) A detailed description of the image analysis algorithm, including a description of the algorithm inputs and outputs, each major component or block, how the algorithm and output affects or relates to clinical practice or patient care, and any algorithm limitations.
(ii) A detailed description of pre-specified performance testing protocols and dataset(s) used to assess whether the device will provide improved assisted-read detection and diagnostic performance as intended in the indicated user population(s), and to characterize the standalone device performance for labeling. Performance testing includes standalone test(s), side-by-side comparison(s), and/or a reader study, as applicable.
(iii) Results from standalone performance testing used to characterize the independent performance of the device separate from aided user performance. The performance assessment must be based on appropriate diagnostic accuracy measures (
e.g., receiver operator characteristic plot, sensitivity, specificity, positive and negative predictive values, and diagnostic likelihood ratio). Devices with localization output must include localization accuracy testing as a component of standalone testing. The test dataset must be representative of the typical patient population with enrichment made only to ensure that the test dataset contains a sufficient number of cases from important cohorts (e.g., subsets defined by clinically relevant confounders, effect modifiers, concomitant disease, and subsets defined by image acquisition characteristics) such that the performance estimates and confidence intervals of the device for these individual subsets can be characterized for the intended use population and imaging equipment.(iv) Results from performance testing that demonstrate that the device provides improved assisted-read detection and/or diagnostic performance as intended in the indicated user population(s) when used in accordance with the instructions for use. The reader population must be comprised of the intended user population in terms of clinical training, certification, and years of experience. The performance assessment must be based on appropriate diagnostic accuracy measures (
e.g., receiver operator characteristic plot, sensitivity, specificity, positive and negative predictive values, and diagnostic likelihood ratio). Test datasets must meet the requirements described in paragraph (b)(1)(iii) of this section.(v) Appropriate software documentation, including device hazard analysis, software requirements specification document, software design specification document, traceability analysis, system level test protocol, pass/fail criteria, testing results, and cybersecurity measures.
(2) Labeling must include the following:
(i) A detailed description of the patient population for which the device is indicated for use.
(ii) A detailed description of the device instructions for use, including the intended reading protocol and how the user should interpret the device output.
(iii) A detailed description of the intended user, and any user training materials or programs that address appropriate reading protocols for the device, to ensure that the end user is fully aware of how to interpret and apply the device output.
(iv) A detailed description of the device inputs and outputs.
(v) A detailed description of compatible imaging hardware and imaging protocols.
(vi) Warnings, precautions, and limitations must include situations in which the device may fail or may not operate at its expected performance level (
e.g., poor image quality or for certain subpopulations), as applicable.(vii) A detailed summary of the performance testing, including test methods, dataset characteristics, results, and a summary of sub-analyses on case distributions stratified by relevant confounders, such as anatomical characteristics, patient demographics and medical history, user experience, and imaging equipment.