The Overjet Caries Assist (OCA) is a radiological, automated, concurrent read, computer-assisted detection software intended to aid in the detection and segmentation of caries on bitewing radiographs. The device provides additional information for the dentist to use in their diagnosis of a tooth surface suspected of being carious. The device is not intended as a replacement for a complete dentist's review or that takes into account other relevant information from the image, patient history, and actual in vivo clinical assessment.

Device Description

Overjet Caries Assist (OCA) is a radiological automated concurrent read computer-assisted detection (CAD) software intended to aid in the detection and segmentation of caries on bitewing radiographs. The device provides additional information for the clinician to use in their diagnosis of a tooth surface suspected of being carious. 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.

OCA is a software-only device which operates in three layers - a Network Layer, a Presentation Layer, and a Decision Layer (as shown in the data flow diagram below). Images are pulled in from a clinic/dental office, and the Machine Learning model creates predictions in the Decision Layer and results are pushed to the dashboard, which are in the Presentation Layer.

The Machine Learning System within the Decision Layer processes bitewing radiographs and annotates suspected carious lesions. It is comprised of four modules:

Image Classifier The model evaluates the incoming radiograph and predicts the ● image type between Bitewing and Periapical Radiograph. This classification is used to support the data flow of the incoming radiograph. As part of the classification of the image type any non-radiographs are classified as "junk" and not processed. These include patient charting information, or other non-bitewing or periapical radiographs. OCA shares classifier and Tooth Number modules with the Overjet Dental Assist product cleared under K210187.
. Tooth Number Assignment module - This module analyzes the processed image and determines what tooth numbers are present and provides a pixel wise segmentation mask for each tooth number.
Caries module - This module outputs a pixel wise segmentation mask of all carious lesions using an ensemble of 3 U-Net based models. The shape and location of every carious lesion is contained in this mask as the carious lesions' predictions.
Post Processing The overlap of tooth masks from the Tooth Number . Assignment Module and carious lesions from the Caries Module is used to assign specific carious lesions to a specific tooth. The Image Post Processor module annotates the original radiograph with the carious lesions' predictions.

AI/ML Overview

Acceptance Criteria and Device Performance for Overjet Caries Assist

The Overjet Caries Assist (OCA) is a radiological, automated, concurrent read, computer-assisted detection software intended to aid in the detection and segmentation of caries on bitewing radiographs. The device's performance was evaluated through standalone testing of the AI algorithm and a clinical reader improvement study.

1. Table of Acceptance Criteria and Reported Device Performance

Measure	Acceptance Criteria (Predicate Device Performance)	Reported Device Performance (Overjet Caries Assist)
Reader Improvement Study
Increase in dentist's sensitivity with AI assistance	Approximately 20% increase in sensitivity for the predicate device. For OCA, a greater than 15% increase in dentist's sensitivity was established as acceptance criteria.	Overall reader sensitivity improved from 57.9% to 76.2% (an increase of 18.3 percentage points, satisfying the >15% criterion). - Primary caries: 60.5% to 79.4% (18.9 pp improvement). - Secondary caries: 49.8% to 63.0% (13.2 pp improvement).
Specificity with AI assistance	Not explicitly defined as an improvement criterion for the predicate, but overall specificity is a key measure.	Overall reader specificity decreased slightly from 99.3% to 98.4% (a decrease of less than 1%), deemed acceptable by the applicant as the benefit in sensitivity outweighs this slight decrease.
AFROC Score (Assisted)	The predicate did not explicitly state an AFROC criterion, but improving diagnostic accuracy is implicit.	AUC increased from 0.593 (unassisted) to 0.649 (assisted), for an increase of 0.057 (statistically significant, p < 0.001).
Standalone Performance (AI Algorithm Only)
Standalone Sensitivity	Not directly comparable to predicate's standalone AI performance, as the predicate's description focuses on human improvement.	Overall standalone sensitivity: 72.0% (95% CI: 62.9%, 81.1%) - Primary caries: 74.4% (95% CI: 64.4%, 84.4%) - Secondary caries: 62.5% (95% CI: 46.6%, 78.4%)
Standalone Specificity	Not directly comparable to predicate's standalone AI performance.	Overall standalone specificity: 98.1% (95% CI: 97.7%, 98.5%)
Lesion Segmentation (Dice Score)	Not explicitly provided for the predicate device.	Mean Dice score for true positives: - Primary caries: 0.69 (0.66, 0.72) - Secondary caries: 0.75 (0.71, 0.79)

2. Sample Size and Data Provenance for the Test Set

Sample Size for Test Set: 352 bitewing radiographs (104 containing caries / 248 without caries).
Data Provenance: Not explicitly stated in the provided text (e.g., country of origin). However, given the context of U.S. FDA clearance and the use of US-licensed dentists, it is likely that the data is either from the US or representative of populations seen in the US. The type of data is retrospective, as existing radiographs were used.

3. Number of Experts and Qualifications for Ground Truth

Number of Experts: Three US-licensed dentists for initial consensus, and one Dental Radiologist for adjudication of non-consensus labels.
Qualifications of Experts: All experts were US-licensed dentists. The adjudicating expert was specifically a Dental Radiologist. No further details on years of experience were provided.

4. Adjudication Method for the Test Set

The adjudication method used was a "3+1" approach. Ground truth was initially established by the consensus labels of three US-licensed dentists. Any non-consensus labels were then adjudicated by a Dental Radiologist.

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

Yes, an MRMC comparative effectiveness study was conducted.

Effect Size of Human Readers Improvement with AI vs. without AI assistance:
- Sensitivity: Overall reader sensitivity improved by 18.3 percentage points (from 57.9% unassisted to 76.2% assisted).
  - For primary caries, sensitivity improved by 18.9 percentage points (60.5% unassisted to 79.4% assisted).
  - For secondary caries, sensitivity improved by 13.2 percentage points (49.8% unassisted to 63.0% assisted).
- Specificity: Overall reader specificity decreased slightly by 0.9 percentage points (from 99.3% unassisted to 98.4% assisted).
- AFROC AUC: The average AUC for all readers increased by 0.057 (from 0.593 unassisted to 0.649 assisted). This increase was statistically significant (p < 0.001).
- Average Dice Scores for Segmentation:
  - Primary caries: Mean Dice scores improved from 0.67 unassisted to 0.69 assisted.
  - Secondary caries: Mean Dice scores improved from 0.65 unassisted to 0.74 assisted. (Note: These segmentation improvements were not statistically significant).

6. Standalone (Algorithm Only) Performance

Yes, standalone performance (algorithm only without human-in-the-loop) was conducted.

Overall standalone sensitivity: 72.0% (95% CI: 62.9%, 81.1%)
Overall standalone specificity: 98.1% (95% CI: 97.7%, 98.5%)
Lesion Segmentation (Dice Score):
- Primary caries: Mean Dice score of 0.69
- Secondary caries: Mean Dice score of 0.75

7. Type of Ground Truth Used

The ground truth used was expert consensus complemented by expert adjudication. Specifically, a consensus of three US-licensed dentists, with non-consensus cases adjudicated by a Dental Radiologist.

8. Sample Size for the Training Set

The sample size for the training set is not provided in the excerpt. The document only mentions "training data" in the context of the algorithm's capability to learn during its operation, but not a specific size for its initial training.

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

The method for establishing ground truth for the training set is not explicitly detailed in the provided text. It generally states that the algorithm "has been trained," but does not provide information on how the ground truth for that training was established.

Summary

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

May 10, 2022

Image /page/0/Picture/1 description: The image shows the logo of the U.S. Food and 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.

Overjet, Inc. % Adam Odeh Regulatory Contact 560 Harrison Ave, Unit 403 BOSTON MA 02118

Re: K212519

Trade/Device Name: Overjet Caries Assist Regulation Number: 21 CFR 892.2070 Regulation Name: Medical Image Analyzer Regulatory Class: Class II Product Code: MYN Dated: April 6, 2022 Received: April 7, 2022

Dear Mr. Adam Odeh:

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 medical devices and radiation-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,

Laurel Burk, Ph.D. Assistant Director Diagnostic X-ray Systems Team DHT 8B: Division of Radiological Imaging 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) K212519

Device Name Overjet Caries Assist

Indications for Use (Describe)

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

Prescription Use (Part 21 CFR 801 Subpart D)	Over The Counter Use (21 CFR 801)
-----------------------------------------------------------------------------------------------------------------------------	--------------------------------------------------------------------------------------------------

|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}------------------------------------------------

Section 4: 510(k) Summary (K212519)

This summary of 510(k) information is being submitted in accordance with the requirements of 21CFR Part 807.92

1. Date Prepared: October 19, 2021
1. Applicant

Overjet, Inc. 560 Harrison Ave Unit 403 Boston, MA 02118 Contact Person: Adam N. Odeh Email: adam.odeh@overjet.ai

1. Trade Name Overjet Caries Assist
Common Name 4. Medical Imaging Analyzer
1. Classification 21 CFR 892.2070. Product code MYN. Class 2. Radiology
Device Description 6.

The Machine Learning System within the Decision Layer processes bitewing radiographs and annotates suspected carious lesions. It is comprised of four modules:

Image Classifier The model evaluates the incoming radiograph and predicts the ● image type between Bitewing and Periapical Radiograph. This classification is

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

used to support the data flow of the incoming radiograph. As part of the classification of the image type any non-radiographs are classified as "junk" and not processed. These include patient charting information, or other non-bitewing or periapical radiographs. OCA shares classifier and Tooth Number modules with the Overjet Dental Assist product cleared under K210187.

. Tooth Number Assignment module - This module analyzes the processed image and determines what tooth numbers are present and provides a pixel wise segmentation mask for each tooth number.
Caries module - This module outputs a pixel wise segmentation mask of all carious lesions using an ensemble of 3 U-Net based models. The shape and location of every carious lesion is contained in this mask as the carious lesions' predictions.
Post Processing The overlap of tooth masks from the Tooth Number . Assignment Module and carious lesions from the Caries Module is used to assign specific carious lesions to a specific tooth. The Image Post Processor module annotates the original radiograph with the carious lesions' predictions.
1. Indications for Use

The Overjet Caries Assist (OCA) is a radiological, automated, concurrent read, computer-assisted detection software intended to aid in the detection and segmentation of caries on bitewing radiographs. The device provides additional information for the dentist to use in their diagnosis of a tooth surface suspected of being carious. The device is not intended as a replacement for a complete dentist's review or their clinical judgment that takes into account other relevant information from the image, patient history, and actual in vivo clinical assessment.

1. Intended Patient Population:
  The intended patient population of the device is patients that have permanent dentition, and are at least 18 years of age.
1. Warning and Limitations:
- The safety and effectiveness of the system has not been established in patients with primary or mixed dentition.
- The device should only be used by trained dentists.
- Overjet Caries Assist system assists only in potential caries detection, not interpretation or diagnosis. It should not be relied on as the sole decision-making tool for diagnosis or treatment.
- Only images from a supported dental radiograph system as defined in the manual can be ●

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

used.

. The Overjet Dental Assist system is not intended for images smaller than 500 x 500 resolution. Overjet cannot guarantee the accuracy of results when Overjet Caries Assist is used on images of lower resolution.
The product is not 100% sensitive, and some caries will not be detected. This can delay ● necessary treatment.
Gross decay detection is not supported by Overjet Caries Assist, as the algorithm has been trained to detect caries within the tooth structure, rather than to detect a lack of tooth structure as is often observed with severe decay.
Endodontic access may be mistaken by the product as caries, due to the similarities on the radiographs.
Overiet Caries Assist cannot detect carious lesions that are not visible to a dentist on bitewing radiographs (e.g., obstructed by radiopaque restorations).
. The product has the potential for false positive or false negative outputs. This could result in unnecessary treatment on rare occasions. Final clinical determination is the responsibility of the treating clinician with the assessment of the actual patient's dentition and multiple clinical indicators of treatment unrelated to the prediction.
The dentist should use all appropriate clinical information to render a final clinical . opinion, with radiographic interpretation being one component of the determination process.
The dentist is responsible for reviewing the segmentation accuracy prior to making ● diagnostic decisions, and for manually adjusting segmentation when deemed necessary.

10. Predicate Device

Device - Logicon Caries Detector Manufacturer - Carestream Dental PMA - P980025 (down-classified to Class II under 85 FR 3548, Jan. 22, 2020)

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

9. Substantial Equivalence

Device	CarestreamLogiconCaries Detector	Overjet CariesAssist(proposed)
510k	P980025	K212519
Regulation No./ Description	CFR 892.2070Medical image analyzer	CFR 892.2070Medical image analyzer
Indications	The Logicon Caries Detector is asoftware device that is an aid in thediagnosis of caries that havepenetrated into the dentin, onun-restored proximal surfaces ofsecondary dentition through thestatistical analysis of digitalintra-oral radiographic imagery. Thedevice provides additionalinformation for the clinician to usein his/her diagnosis of a toothsurface suspected of being carious.It is designed to work in conjunctionwith an existing Carestream dentalRVG digital X-ray radiographicsystem with dental imagingsoftware (dis) for Windows XP orhigher.	The Overjet Caries Assist (OCA) is aradiological, automated, concurrentread, computer-assisted detectionsoftware intended to aid in thedetection and segmentation of carieson bitewing radiographs. The deviceprovides additional information forthe dentist to use in their diagnosis ofa tooth surface suspected of beingcarious. The device is not intended asa replacement for a complete dentist'sreview or their clinical judgment thattakes into account other relevantinformation from the image, patienthistory, and actual in vivo clinicalassessment.
End User	Dentist	Dentist
Patient Population	Patients requiring dental services,all sexes, no age restriction	Patients requiring dental services,all sexes, at least 18 years of age,and with permanent dentition.
Platform	Windows PC	Web - Edge, Chrome, Firefox
OS	Microsoft Window 7, 8, 10	Any
User Interface	Mouse, Keyboard	Mouse, Keyboard, Trackpad
Image InputSources	Images can be scanned, loadedfrom connected Carestream imagesolutions	Images imported from theradiographic device, or from thepractice management system, fromCarestream or Schick sensors
Image format	Carestream	jpg, png, eop, jif, dicom
ProcessingArchitecture	The software provides graphicalrepresentation of the density changein a tooth, by looking for a patternof density dips starting at the toothsurface, penetrating the enamel andgoing into the dentin. Enamel isrepresented by 10 green lines anddentin by 5 blue lines. If a patternsuggestive of caries exists, the dipsare highlighted with red dots towarn the dentist.	Three layers:- The Network layer works with thepractice PACS or EMR to transmitthe image and meta-data to Overjet.- The decision layer processes theimage to ensure it is the correct datatype, and then annotates it via thealgorithm- The presentation layer displays theannotated image in a non-diagnosticviewer. The dentist can filter,display, hide, create and edit theannotations presented.
Data Source	Bitewing radiographs acquiredfrom Carestream dental RVGdigital X-ray radiographicsystem	Digital files of Bitewing radiographswhose longer edge is greater than 500pixel resolution
Output	• Outline of suspected region• Tooth Density• Lesion (caries) probability	Caries detection and segmentationon radiograph resulting in outline ofsuspected caries
PerformanceTesting	Increase in dentist's sensitivity ofapproximately 20%	Increase in dentist's sensitivity ofgreater than 15%
Level of Concern	Moderate	Moderate

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

Overjet Caries Assist is determined to be substantially equivalent to the Carestream Logicon Caries Detector approved in P980025 and later down-classified to Class II under 85 FR 3548, Jan. 22, 2020. Both systems are software intended to support dental professionals in their diagnosis and treatment planning for caries.

Both software systems automatically annotate suspected carious lesions for the dentist to review. The Logicon software displays suspected carious lesions as dips in annotated lines within the radiograph, green lines for the enamel and blue lines for the dentin. The Overjet Caries Assist presents suspected carious lesions as segmented polygons outlining the prediction. Both systems allow users to visualize the radiograph with the annotations, add their own annotations, and use the information as part of their diagnostic process.

Other similarities include both systems have no direct contact with the patient, both systems evaluate oral cavity radiographs, both systems utilize standard image types, and both systems connect to practice management systems. Logicon compatibility is limited to Carestream products

Some differences between the systems include the location of the software, the user interface, and the availability of additional features. A primary difference is the Carestream

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

Logicon is a local software while Overjet Caries Assist is a cloud native application. While Logicon and Overjet Caries Assist have different user interfaces, both are accessed by computer and are intended for dental professionals to review annotations on dental radiographs. Overjet does not believe that the differences raise a concern of substantial equivalence and these differences do not interfere with the ability of the Overjet software to achieve its intended use.

10. Performance Testing

Overjet has conducted performance testing according to FDA's "Guidance for Industry and Food and Drug Administration Staff Computer-Assisted Detection Devices Applied to Radiology Images and Radiology Device Data - Premarket Notification [510(k)] Submissions Document" issued on: July 3, 2012, as part of the development process of the caries model. Performance testing included standalone testing and a clinical reader evaluation. All testing demonstrated that the Overjet Caries Assist software met prespecified requirements.

Standalone Testing

Standalone performance of the Overjet AI algorithm for the 352 images is shown in the tables below. Sensitivity and specificity were summarized based on surfaces, and 95% CIs were provided based on treating the subject as the basis of a cluster. A total of 7,129 surfaces were available and included in the analysis.

Sensitivity

Overall standalone sensitivity was 72.0% (62.9%, 81.1%). When broken down further. sensitivity was 74.4% (64.4%, 84.4%) for primary caries, and 62.5% (46.6%, 78.4%) for secondary caries.

Specificity

Overall standalone specificity was 98.1% (97.7%, 98.5%).

Subgroup Analyses:

Subgroup analyses were also performed for age, gender, and sensor, as well as by primary vs secondary classification, and associated restoration (for secondary caries).

{9}------------------------------------------------

		Ground Truth
Assessment	OverjetAI	CariesPresent	No CariesPresent	Total (%)
Observed Counts	Caries PresentNo CariesTotal (%)	175 (2.5%)68 (1.0%)243 (3.4%)	131 (1.8%)6755 (94.8%)6886 (96.6%)	306 (4.3%)6823 (95.7%)7129
DiagnosticStatistic	MeasureSensitivitySpecificity	Estimate72.0% (175/243)98.1% (6755/6886)	95% CI62.9%, 81.1%97.7%, 98.5%

Standalone Performance of Overjet AI Algorithm based on Surfaces

· Confidence interval adjusted for multiple images and caries per subject based on clustered binary data methodology.

Lesion Segmentation:

Dice coefficient analysis was performed to compare pixel-level metrics of each carious lesion with the lesion tracing provided by ground truthers. Dice scores were calculated only for true positives.

For 66 images containing primary caries, the mean Dice score was 0.69 (0.66, 0.72) with a standard deviation of 0.122). For 30 images containing secondary caries, the mean Dice score was 0.75 (0.71, 0.79) with a standard deviation of 0.112.

Thus, it is unlikely that the automatic lesion segmentation presents any new risks, and any such potential risks are mitigated by the fact that Overjet Caries Assist allows the dentist to modify the margins of any segmented lesions.

Clinical Evaluation - Reader Improvement

Overjet evaluated the Overjet Caries Assist in a multi reader fully crossed reader improvement study. 13 US licensed dentists were asked to evaluate 352 radiographs (104 containing caries / 248 without caries). Ground truth was established by the consensus labels of three US licensed dentists, and non-consensus labels were adjudicated by a Dental Radiologist. Half of the data set contained unannotated images, and the second half contained radiographs that had been processed through the OCA model. The radiographs were presented to the reader in alternating groups.

In Session 1, the readers were asked to outline suspected caries, and to review predictions from the OCA model. Each reader was asked to provide a rating of 1-4 for their confidence in the label (1 low confidence, 4 high confidence). A 30-day washout period was utilized to limit recollection bias. Following the washout, the readers were presented the same data set but with alternate grouping. If a reader saw a radiograph in the unpredicted state in session 1, they were presented with the Overjet Caries Assist

{10}------------------------------------------------

predictions in session 2.

The results were compared against a consensus ground truth, and the sensitivity, specificity, and alternative free response receiver operating characteristic (AFROC) was evaluated to characterize the performance of the readers with and without viewing the model annotations.

Unassisted vs Assisted Sensitivity:

Overall reader sensitivity improved from 57.9% (48.9%, 66.0%) to 76.2% (68.4%, 82.6%) unassisted vs assisted. For primary caries, reader sensitivity improved from 60.5% (49.3%, 69.2%) to 79.4% (71.0%, 85.9%). For secondary caries, reader sensitivity improved from 49.8% (39.5%, 60.6%) to 63.0% (52.0%, 74.8%).

Unassisted vs Assisted Specificity:

Overall reader specificity decreased slightly from 99.3% (99.1%, 99.5%) to 98.4% (94.5%, 98.8%) unassisted vs assisted.

Subgroup Analyses:

Subgroup analyses were performed for age, gender, and sensor, primary vs secondary classification, and associated restoration (for secondary caries).

Unassisted vs Assisted Dice Scores:

As with standalone testing, Dice scores were calculated in comparison to ground truth for readers with and without Overjet Caries Assist.

For primary caries, mean Dice scores improved from 0.67 unassisted (standard deviation 0.012) to 0.69 assisted (SD 0.011). For secondary caries, mean Dice scores improved from 0.65 unassisted (SD 0.017) to 0.74 assisted (SD 0.017). These data suggest improved segmentation when using Overjet Caries Assist, though results were not statistically significant.

AFROC Scores:

Readers provided confidence scores for any detected caries, which were used to calculate AUC for weighted AFROC scores. For the average of all readers, AUC increased from 0.593 (0.686, 0.743) to 0.649 (0.744, 0.820), for an increase in AUC of 0.057 (0.039, 0.098) unassisted to assisted. This increase was statistically significant, with an overall p-value less than 0.001.

Summary

{11}------------------------------------------------

Increases in overall AFROC numbers clearly demonstrate improvement in caries detection by dentists when aided by Overjet Caries Assist (0.057 increase). This aligns with the observed increase in sensitivity for both primary and secondary caries (18.9% improvement for primary; 13.2% improvement for secondary caries). When considered alongside the observed decrease in overall specificity, which was less than 1%, it is clear that Overjet Caries Assist demonstrates a clear benefit for caries detection.

1. General Safety and Effectiveness Concerns
  The device labeling contains instructions for use and any necessary cautions and warnings to provide for safe and effective use of this device. Risk management was conducted according to ISO 14971 which ensured, via a risk analysis, the identification and mitigation of potential hazards. Any potential hazards were controlled via software development and design, verification, and validation testing. In addition, general and special controls of the FD&C Act established for Radiological Computer Assisted Detection and Diagnosis Software are in place to further mitigate any safety and or effectiveness risks.
1. Assessment of Non-Clinical Performance Data
  Overjet Caries Assist has been verified and validated according to Overjet's design control processes. All supporting documentation has been included in this 510(k) Premarket Notification. Verification activity included unit, integration, and system level testing. Validation testing included performing a pivotal reader study to compare the clinical performance of dentists using CAD detections from the Overjet Caries Assist software when applied to dental radiographs to that of dentists not using Overjet Caries Assist.

13. Conclusion

Overjet Caries Assist is substantially equivalent to the predicate device, Carestream Logicon Caries Detector. Any differences do not raise any concerns about the safety or efficacy of the device.

Regulation Number and Section

§ 892.2070 Medical image analyzer.

(a)
Identification. Medical image analyzers, including computer-assisted/aided detection (CADe) devices for mammography breast cancer, ultrasound breast lesions, radiograph lung nodules, and radiograph dental caries detection, is a prescription device that is intended to identify, mark, highlight, or in any other manner direct the clinicians' attention to portions of a radiology image that may reveal abnormalities during interpretation of patient radiology images by the clinicians. This device incorporates pattern recognition and data analysis capabilities and operates on previously acquired medical images. This device is not intended to replace the review by a qualified radiologist, and is not intended to be used for triage, or to recommend diagnosis.(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 algorithms including a description of the algorithm inputs and outputs, each major component or block, and algorithm limitations.
(ii) A detailed description of pre-specified performance testing methods and dataset(s) used to assess whether the device will improve reader performance as intended and to characterize the standalone device performance. Performance testing includes one or more standalone tests, side-by-side comparisons, or a reader study, as applicable.
(iii) Results from performance testing that demonstrate that the device improves reader performance in the intended use population when used in accordance with the instructions for use. The performance assessment must be based on appropriate diagnostic accuracy measures (
e.g., receiver operator characteristic plot, sensitivity, specificity, predictive value, and diagnostic likelihood ratio). The test dataset must contain a sufficient number of cases from important cohorts (e.g., subsets defined by clinically relevant confounders, effect modifiers, concomitant diseases, 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) 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; and cybersecurity).(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 reading protocol.
(iii) A detailed description of the intended user and user training that addresses appropriate reading protocols for the device.
(iv) A detailed description of the device inputs and outputs.
(v) A detailed description of compatible imaging hardware and imaging protocols.
(vi) 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 or for certain subpopulations), as applicable.(vii) Device operating instructions.
(viii) 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 lesion and organ characteristics, disease stages, and imaging equipment.