The red dot™ software platform is a software workflow tool designed to aid the clinical assessment of adult Chest X-Ray cases with features suggestive of Pneumothorax in the medical care environment. red dot™ analyzes cases using an artificial intelligence algorithm to identify suspected findings. It makes case-level output available to a PACS/workstation for worklist prioritization or triage. red dot™ is not intended to direct attention to specific portions of an image or to anomalies other than Pneumothorax. Its results are not intended to be used on a stand-alone basis for clinical decision-making nor is it intended to rule out Pneumothorax or otherwise preclude clinical assessment of X-Ray cases.

Device Description

Behold.ai red dot™ is a radiological computer-assisted triage and notification software system. The software automatically analyzes PA/AP chest x-rays and alerts the PACS/RIS workstation once images with features suggestive of pneumothorax are identified.

Through the use of the red dot™ device, a radiologist is able to review studies with features suggestive of pneumothorax earlier than in standard of care workflow.

In summary, the red dot™ device provides a passive notification through the PACS/workstation to the radiologists indicating the existence of a case that may potentially benefit from prioritization. It doesn't output an image and therefore it does not mark, highlight, or direct users' attention to a specific location on the original chest X ray.

The device aim is to aid in prioritization and triage of radiological medical images only.

The main components of the red dot™ device are described below.

Image input, validation and anonymization
After a chest x-ray has been performed, a copy of the study is received and processed by the red dot™ device. Following receipt of a study, the validation feature ensures the image is valid (i.e. has readable pixels) and the anonymization feature removes or anonymizes Personally Identifiable Information (PII) such as Patient Name, Patient Birthdate, and Patient Address.
red dot™ Image Analysis Algorithm
This component of the device is primarily comprised of the visual recognition algorithm that is responsible for detecting images with potential abnormalities. Once a study has been validated, the algorithm analyzes the frontal chest x-ray for detection of suspected findings suggestive of pneumothorax.
PACS Integration Feature
The results of a successful study analysis is provided to an integration engine in a standard JSON message containing sufficient information to allow the integration engine to notify the PACS/workstation for prioritization through the worklist interface.

AI/ML Overview

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

Acceptance Criteria and Device Performance

1. Table of Acceptance Criteria and Reported Device Performance

Metric	Acceptance Criteria (Implied)	Reported Device Performance
AUROC	> 0.95 (as stated for "prespecified performance goals")	0.975 (95% CI: [0.966 - 0.984])
Sensitivity	Lower bound of 95% CI > 80%	94.65% (95% CI: [91.46 - 96.91])
Specificity	Lower bound of 95% CI > 80%	87.95% (95% CI: [85.04 - 90.46])
Accuracy	Not explicitly stated as an acceptance criterion bound beyond the above, but reported.	90.20% (95% CI: [88.06 - 92.08])
Processing Time (red-dot™)	Substantially equivalent to predicate (Zebra HealthPNX: 22.1 seconds)	13.8 seconds (Mean, 95% CI: [13.0 - 14.5])
Notification Transit Time	Implied to be part of overall timing comparison with predicate	15.5 seconds (Average from 3 live customer sites)
Total red dot™ Performance Time	Substantially equivalent to predicate (Zebra HealthPNX: 22.1 seconds)	29.3 seconds

Note on Acceptance Criteria: The document explicitly states that the AUROC was above 0.95 and "all lower bounds of the 95% confidence intervals exceeded 80% and achieved the prespecified performance goals in the study" for the classification metrics (AUROC, Sensitivity, Specificity). For the timing, the acceptance criterion is defined as being "substantially equivalent" to the predicate.

Study Details Proving Device Meets Acceptance Criteria

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

Test Set Sample Size: 888 CXR images.
Data Provenance: Retrospective, anonymous study.
- Country of Origin: United States (n=738 cases from 2 clinical sites) and United Kingdom (n=150 cases from 2 clinical sites).

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

Number of Experts: At least two ABR certified radiologists reviewed each CXR image. A third reader was involved in the event of disagreement/discrepancy.
Qualifications of Experts: All readers were "ABR certified radiologists" and "received training related to imaging findings defining each condition per protocol prior to the review."

4. Adjudication Method for the Test Set

Adjudication Method: "The ground truth was determined by two readers with a third reader in the event of disagreement/discrepancy." Ground truth for a condition was defined as agreement between two readers. This is a 2+1 adjudication method.

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

MRMC Study: No, a multi-reader multi-case (MRMC) comparative effectiveness study was not reported as having been done to directly compare human readers with and without AI assistance. The study described is a standalone performance validation of the AI algorithm against a consensus ground truth.

6. If a Standalone Performance Study Was Done

Standalone Study: Yes, a standalone (algorithm only without human-in-the-loop performance) study was explicitly done. The reported metrics (AUROC, Accuracy, Sensitivity, Specificity) are for the red dot™ algorithm's performance in detecting pneumothorax.

7. The Type of Ground Truth Used

Type of Ground Truth: Expert consensus. Specifically, "agreement between two readers" from ABR certified radiologists, with a third radiologist for discrepancy resolution.

8. The Sample Size for the Training Set

Training Set Sample Size: The document does not specify the sample size for the training set. It only describes the test set.

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

Training Set Ground Truth Establishment: The document does not provide details on how the ground truth for the training set was established. It only focuses on the data used for the performance evaluation (test set).

Summary

{0}------------------------------------------------

February 28, 2020

Image /page/0/Picture/1 description: The image shows the logos of the Department of Health and Human Services and the Food and Drug Administration (FDA). The Department of Health and Human Services logo is on the left, and the FDA logo is on the right. The FDA logo includes the letters "FDA" in a blue square, followed by the words "U.S. FOOD & DRUG ADMINISTRATION" in blue text. The logos are simple and professional, and they are easily recognizable.

Behold.AI Technologies Limited % John J. Smith, M.D., J.D. Regulatory Counsel Hogan Lovells US LPP 555 13th Street. NW WASHINGTON DC 20004

Re: K191556

Trade/Device Name: red dot™ Device Regulation Number: 21 CFR 892.2080 Regulation Name: Radiological computer aided triage and notification software Regulatory Class: Class II Product Code: QFM Dated: January 29, 2020 Received: January 29, 2020

Dear Dr. Smith:

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.

For

Thalia T. Mills, Ph.D. Director Division of Radiological Health OHT7: Office of In Vitro Diagnostics and Radiological Health Office of Product Evaluation and Quality Center for Devices and Radiological Health

Enclosure

{2}------------------------------------------------

510(k) Number (if known)

K191556

Device Name

red dot™ device

Indications for Use

Type of Use (Select one or both, as applicable)

区 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 the text "behold.ai – 510(k) Summary". The word "behold.ai" is in teal, while the rest of the text is in black. The text appears to be the title or heading of a document or presentation.

Image /page/3/Picture/1 description: The image is a logo of an eye. The eye has a teal iris with a dark blue pupil. The iris is made up of many small, rectangular shapes. The pupil is white and is located in the center of the eye.

Behold.ai red dot™

510(k) SUMMARY Behold.ai Technologies Limited's red dot™

K191556

Submitter

Behold.ai Technologies Limited 91-97 Bohemia Road, St. Leonards-On-Sea, United Kingdom, TN37 6RJ Phone: (+44) 7734 048778

Contact Person: Simon Rasalingham

Date Prepared: January 29, 2020

Name of Device: Behold.ai red dot™ Device Classification Name: Radiological Computer-Assisted Prioritization Software for Lesions Regulatory Class: Class II (Radiology) Product Code: QFM

Predicate Device

The red dot™ device is substantially equivalent to the following device:

Manufacturer Name	Zebra Medical Vision Ltd.
Devices Trade Name	HealthPNX
510(k) Number	K190362

A. Device Description

Through the use of the red dot™ device, a radiologist is able to review studies with features suggestive of pneumothorax earlier than in standard of care workflow.

Information Classification: BEHOLD.AI PUBLIC

Document Version 2.1

{4}------------------------------------------------

Image /page/4/Picture/1 description: The image is a stylized graphic of an eye. The iris is teal and has a ring of teal rectangles around the pupil. The pupil is black with a white glare spot in the upper left.

Behold.ai red dot™

The device aim is to aid in prioritization and triage of radiological medical images only.

B. Technical Components

The main components of the red dot™ device are described below.

1. Image input, validation and anonymization
  After a chest x-ray has been performed, a copy of the study is received and processed by the red dot™ device. Following receipt of a study, the validation feature ensures the image is valid (i.e. has readable pixels) and the anonymization feature removes or anonymizes Personally Identifiable Information (PII) such as Patient Name, Patient Birthdate, and Patient Address.
1. red dot™ Image Analysis Algorithm
  This component of the device is primarily comprised of the visual recognition algorithm that is responsible for detecting images with potential abnormalities. Once a study has been validated, the algorithm analyzes the frontal chest x-ray for detection of suspected findings suggestive of pneumothorax.
1. PACS Integration Feature
  The results of a successful study analysis is provided to an integration engine in a standard JSON message containing sufficient information to allow the integration engine to notify the PACS/workstation for prioritization through the worklist interface.

C. Intended Use / Indications for Use

D. Summary of Technological Characteristics

Behold.ai red-dot™ tool uses Artificial Intelligence (Al) and pattern recognition technology to analyze chest X-rays. The Behold.ai red-dot™ tool notifies a clinician of the presence of Pneumothorax in a radiological image. The technological characteristics of the red dot™ device and the HealthPNX predicate are compared below:

Information Classification: BEHOLD.AI PUBLIC

Document Version 2.1

Information within this document must not be disclosed unless with prior written consent.

{5}------------------------------------------------

Image /page/5/Picture/1 description: The image is a stylized graphic of an eye. The iris is teal and has a ring of dark teal rectangles around the pupil. The pupil is black with a white glare spot in the upper left.

Behold.ai red dot™

TechnologicalCharacteristics	Subject Device:Behold.ai red dot™	Predicate Device:Zebra HealthPNXK190362	If different,Impact on Safetyand or Efficacy
Indication for use	The red dot™ software platformis a software workflow tooldesigned to aid the clinicalassessment of adult Chest X-Ray cases with features suggestive ofPneumothorax in the medicalcare environment. red dot™analyzes cases using an artificialintelligence algorithm to identifysuspected findings. It makescase-level output available to aPACS/workstation for worklistprioritization or triage. red dot™is not intended to direct attentionto specific portions of an imageor to anomalies other thanPneumothorax. Its results are notintended to be used on a stand-alone basis for clinical decision-making nor is it intended to ruleout Pneumothorax or otherwisepreclude clinical assessment ofX-Ray cases.	The Zebra Pneumothorax deviceis a software workflow tooldesigned to aid the clinicalassessment of adult Chest X-Ray cases with features suggestive ofPneumothorax in the medical careenvironment. HealthPNXanalyzes cases using an artificialintelligence algorithm to identifysuspected findings. It makescase-level output available to aPACS/workstation for worklistprioritization or triage. HealthPNXis not intended to direct attentionto specific portions of an image orto anomalies other thanPneumothorax. Its results are notintended to be used on a stand-alone basis for clinical decision-making nor is it intended to ruleout Pneumothorax or otherwisepreclude clinical assessment of X-Ray cases	Same
Notification-only,parallel workflowtool	Yes	Yes	Same
Intended user	Hospital networks and trainedclinicians	Hospital networks and trainedclinicians	Same
Radiological imagesformat	DICOM	DICOM	Same
Identify patientswith a pre-specifiedclinical condition	Yes	Yes	Same
Clinical condition	Pneumothorax	Pneumothorax	Same
Alert to finding	Passive notification flagged forreview	Passive notification flagged forreview	Same
TechnologicalCharacteristics	Subject Device:Behold.ai red dot™	Predicate Device:Zebra HealthPNXK190362	If different,Impact on Safetyand or Efficacy
Independent ofstandard of careworkflow	Yes; No cases are removed fromworklist	Yes; No cases are removed fromworklist	Same
Modality	X-Ray	X-Ray	Same
Body Part	Chest	Chest	Same
ArtificialIntelligencealgorithm	Yes	Yes	Same
Limited to analysisof imaging data	Yes	Yes	Same
Aids promptidentification ofcases with indicatedfindings	Yes	Yes	Same
Where results arereceived	PACS / RIS / EPR / Workstation	PACS / Workstation	No significantdifference

Information Classification: BEHOLD.AI PUBLIC

Document Version 2.1

Information within this document must not be disclosed unless with prior written consent.

{6}------------------------------------------------

Image /page/6/Picture/1 description: The image is a stylized graphic of an eye. The eye has a teal iris with a dark teal pupil. The iris is surrounded by a teal ring with evenly spaced dark teal rectangles. The rectangles are arranged in a circular pattern around the iris.

Behold.ai red dot™

E. Performance Data

A multi-center IRB approved HIPAA compliant retrospective anonymized study of adult patient CXR images was performed. CXR images positive and negative for pneumothorax were collected consecutively based on the inclusion and exclusion criteria as per protocol.

The final standalone dataset was composed of 888 CXR images from 2 clinical sites from the United States (n=738) and 2 clinical sites from the United Kingdom (n=150). The final standalone dataset included pneumothorax positive (n=299) and negative cases (n=589). Both groups contained confounding imaging features.

Each CXR image was reviewed by at least two ABR certified radiologists (readers received training related to imaging findings defining each condition per protocol prior to the review. All films were stripped of all identifiers and clinical information. Readers were blinded to each other's reads and to the red dot™ software output. The ground truth was determined by two readers with a third reader in the event of disagreement/discrepancy. Ground truth for a condition was defined as agreement between two readers. The breakdown for the final ground truthed standalone dataset for each site was the following:

Information Classification: BEHOLD.AI PUBLIC

Document Version 2.1

{7}------------------------------------------------

Image /page/7/Picture/1 description: The image is a stylized graphic of an eye. The eye has a dark teal iris with a white pupil. The outer ring of the eye is a lighter teal color and is segmented into many small, rectangular shapes, giving it a textured appearance.

Behold.ai red dot™

Disease Class	TotalNumberof Cases	Site 1Cases	Site 2Cases	Site 3Cases	Site 4Cases
PNX, Non-normalControl, Normal Control	888	419	319	77	73
PNX	299	177	95	14	13
Non-normal Control	167	50	50	36	31
Normal Control	422	192	174	27	29

The AUROC, accuracy, sensitivity and specificity for the detection of pneumothorax, along with the associated 95% confidence intervals are presented below. As shown, the AUROC is above 0.95 and all lower bounds of the 95% confidence intervals exceeded 80% and achieved the prespecified performance goals in the study.

Pneumothorax (N=888)

. AUROC: 0.975 with 95% confidence intervals [0.966 - 0.984]
Accuracy: 90.20% (801/888) with 95% confidence interval: [88.06 92.08] .
. Sensitivity: 94.65% (283/299) with 95% confidence interval: [91.46 - 96.91]
Specificity: 87.95% (518/589) with 95% confidence interval: [85.04 - 90.46]

Additional Endpoints and Analysis

Clinical data regarding the time saving benefit resulting from use of the device as compared to the standard of care is provided.

Definitions

Processing time: the time taken to process the x-ray image and issue a red dot™ result
Notification transit time: the time taken for the notification to reach the PACS/RIS/EPR worklist after being issued by the red dot™ device
red dot™ performance time: this is equal to the sum of the processing time and . the notification transit time

As described in the definitions list above, the red-dot™ processing time includes the time to receive the DICOM image, upload it to the model inference server, analyze the image and issue the result (PNX notification or no PNX notification) to be sent to the hospital system (RIS or PACS Worklist).

The following table summarizes the red-dot™ processing time in seconds for 888 unique X-rays in the aggregate data.

Information Classification: BEHOLD.AI PUBLIC

Document Version 2.1

{8}------------------------------------------------

Image /page/8/Picture/1 description: The image is a stylized graphic of an eye. The eye has a teal iris with a dark teal pupil. The iris is surrounded by a teal ring with evenly spaced dark teal rectangles. The pupil is white.

Behold.ai red dot™

888 Cases
ProcessingTime (seconds)	Mean	StandardDev	95% LowerConfidenceLimit	95% UpperConfidenceLimit	Median
red-dot™processing time	13.8	10.9	13.0	14.5	8.56

The notification transit time is the time taken for the red dot™ notification to travel from the red dot™ device to the point at which the result is displayed in the destination RIS or PACS Worklist. Using data (N=300) established from 3 live customer sites in the UK (100 cases sampled from 3 sites), the average notification transit time in a typical scenario is 15.5 seconds.

When summing the processing time to the notification transit time, the total red dot™ performance time is calculated as 29.3 seconds. This is a timing performance that is substantially equivalent to the Zebra predicate device (K190362. 22.1 seconds). Furthermore, the predicate demonstrated the benefit of triage compared to the standard of care for pneumothorax, meeting the QFM product code requirements.

The red dot™ software device reaching equivalent classification performance (> 0.95 AUC) as well as timing performance to the Zebra predicate device supports that the red dot™ software device can provide equivalent benefit for effective triage as demonstrated by the Zebra predicate device time savings study.

F. Conclusion

Behold.ai red dot™ is as safe and effective as the HealthPNX predicate device. The red dot™ device and the HealthPNX predicate device are both software-only devices intended to aid in triage of radiological images, independent of standard of care workflow. The labeling of both devices are limited to the categorization of exams and are not to be used in-lieu of full patient evaluation or relied upon to make or confirm diagnosis.

The subject and predicate devices operate in parallel to the standard of care workflow in the sense that they do not change the original image, do not provide any marking, and do not remove cases from the standard of care. The minor differences between the subject device and the predicate raise no new issues of safety or effectiveness. In addition, performance testing demonstrates that the red dot™ performs as intended. The red dot™ device is therefore substantially equivalent to the HealthPNX predicate.

Information Classification: BEHOLD.AI PUBLIC

Regulation Number and Section

§ 892.2080 Radiological computer aided triage and notification software.

(a)
Identification. Radiological computer aided triage and notification software is an image processing prescription device intended to aid in prioritization and triage of radiological medical images. The device notifies a designated list of clinicians of the availability of time sensitive radiological medical images for review based on computer aided image analysis of those images performed by the device. The device does not mark, highlight, or direct users' attention to a specific location in the original image. The device does not remove cases from a reading queue. The device operates in parallel with the standard of care, which remains the default option for all cases.(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 notification and triage algorithms and all underlying image analysis algorithms including, but not limited to, a detailed description of the algorithm inputs and outputs, each major component or block, how the algorithm 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 effective triage (
e.g., improved time to review of prioritized images for pre-specified clinicians).(iii) Results from performance testing that demonstrate that the device will provide effective triage. The performance assessment must be based on an appropriate measure to estimate the clinical effectiveness. The test dataset must contain sufficient numbers of cases from important cohorts (
e.g., subsets defined by clinically relevant confounders, effect modifiers, associated diseases, and subsets defined by image acquisition characteristics) such that the performance estimates and confidence intervals for these individual subsets can be characterized with the device for the intended use population and imaging equipment.(iv) Stand-alone performance testing protocols and results of the device.
(v) Appropriate software documentation (
e.g., device hazard analysis; software requirements specification document; software design specification document; traceability analysis; description of verification and validation activities including system level test protocol, pass/fail criteria, and results).(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 intended user and user training that addresses appropriate use protocols for the device;
(iii) Discussion of 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 for certain subpopulations), as applicable;(iv) A detailed description of compatible imaging hardware, imaging protocols, and requirements for input images;
(v) Device operating instructions; and
(vi) A detailed summary of the performance testing, including: test methods, dataset characteristics, triage effectiveness (
e.g., improved time to review of prioritized images for pre-specified clinicians), diagnostic accuracy of algorithms informing triage decision, and results with associated statistical uncertainty (e.g., confidence intervals), including a summary of subanalyses on case distributions stratified by relevant confounders, such as lesion and organ characteristics, disease stages, and imaging equipment.