AI Contouring (VA10A)

K261306 · Varian Medical Systems · QKB · Jul 10, 2026 · Radiology

Device Facts

• Record ID: K261306
• Device Name: AI Contouring (VA10A)
• Applicant: Varian Medical Systems
• Product Code: QKB · Radiology
• Decision Date: Jul 10, 2026
• Decision: SESE
• Submission Type: Traditional
• Regulation: 21 CFR 892.2050
• Device Class: Class 2
• Attributes: AI/ML, Software as a Medical Device, Real-World Evidence

Real-World Evidence

Submission	Device	Sponsor	RWD Sources	RWE Use Summary	Key Tags
K261306 · Jul 10, 2026	AI Contouring (VA10A)	Varian Medical Systems	Retrospective clinical CT and MR image datasets from multiple global sites (North/South America, Asia, Australia, Europe)	Retrospective clinical image data was used to train and validate deep-learning-based autocontouring models for radiation therapy treatment planning.	Retrospective clinical data; Deep learning training; Algorithm validation; Multi-site clinical images

Clinical Evidence

Study Design	Population	Comparator	Key Endpoints
Algorithm Evaluation (CT and MR); Retrospective performance evaluation	Patients undergoing CT or MR imaging for radiation therapy planning; Sample Size: 469 CT subjects; 294 MR subjects (153 Pelvis, 81 Brain OAR, 60 Brain Metastases); Number of Sites: Multiple sites across North/South America, Asia, Australia, and Europe	syngo.via RTiS VC10 (Reference Device)	Dice similarity coefficient, Average Surface Distance (ASSD), Lesionwise Sensitivity

AI Performance

Output	Algorithm	Acceptance	Observed	Dev DS	Dev Readers	Test DS	Test Readers
CT Autocontouring	Anatomical landmark detection and DI2IN segmentation	Dice score non-inferiority to reference; ASSD score non-inferiority to reference; Average user evaluation >= 3	Dice and ASSD results provided in Tables 7-10	Training data: large, diverse datasets from multiple regions (North/South America, Asia, Australia, Europe) and multiple scanner vendors.	—	Test cohort: 469 subjects (413 from predicate, 56 new).	—
MR Brain Metastases Contouring	Deep learning-based segmentation model	Lesionwise DICE non-inferiority to reference; Lesionwise Sensitivity non-inferiority to reference	Lesionwise DICE: 0.74; Lesionwise Sensitivity: 92.5%	Training data: intraparenchymal metastases >= 2mm, not adjacent to surgical bed, acquired pre-RT.	—	Test cohort: 60 subjects.	—
MR Brain OAR Contouring	Deep learning-based segmentation model	Dice score non-inferiority to reference; ASSD score non-inferiority to reference; Average user evaluation >= 3	Dice and ASSD results provided in Tables 13-14	Training data: contrast-enhanced MR images, 1.5T and 3T field strengths.	—	Test cohort: 81 subjects.	—
MR Pelvis OAR Contouring	Deep learning-based segmentation model	Dice score non-inferiority to reference; ASSD score non-inferiority to reference; Average user evaluation >= 3	Dice and ASSD results provided in Tables 13-14	Training data: noncontrast enhanced male pelvis images, 1.5T and 3T field strengths, excluding prostatectomy cases.	—	Test cohort: 153 subjects.	—

Indications for Use

AI Contouring is a post-processing software intended to automatically contour CT and MR structures, including known (diagnosed) brain metastases, using deep-learning-based algorithms. Contours that are generated by AI Contouring may be used as input for clinical workflows for radiation therapy treatment planning. AI Contouring must be used in conjunction with appropriate software such as Treatment Planning Systems and Interactive Contouring applications, to review, edit, and accept contours generated by AI Contouring. The outputs of AI Contouring are intended to be used by qualified and trained medical professionals.

Device Story

AI Contouring (VA10A) is a post-processing software for automatic segmentation of anatomical structures and pathologies on CT and MR images; supports radiation therapy treatment planning. Input: voxel stream from partner medical devices (e.g., Eclipse, Velocity). Operation: utilizes deep learning-based segmentation models (DI2IN for CT; four independent models for MR) to compute structure sets based on configured templates; supports margin, boolean, and cropping operations. Output: voxel stream of contoured structures returned to partner device. Used in professional healthcare facilities by qualified clinicians. Clinicians review, edit, and accept contours via treatment planning systems. Benefits: provides consistent, editable contours; streamlines radiotherapy planning workflow.

Clinical Evidence

Bench-only testing. Performance evaluated on 469 CT subjects and 294 MR subjects (153 pelvis OAR, 81 brain OAR, 60 brain metastases). Metrics included Dice similarity coefficient and Average Symmetric Surface Distance (ASSD). Statistical non-inferiority demonstrated against reference device syngo.via RTiS VC10. Subgroup analyses performed across manufacturers, slice thicknesses, and field strengths (1.5T/3T) showed no notable variation. No clinical data.

Technological Characteristics

Deep learning-based software; modular architecture (CT Autocontouring Module, MR Autocontouring Module). Uses DI2IN and anatomical landmark detection. Connectivity: integrated into Varian platforms (Eclipse, Velocity). Standards: ISO 14971, ISO 13485, IEC 62304, IEC 62366-1, IEC 82304-1. Software lifecycle processes compliant with FDA guidance.

Indications for Use

Indicated for automatic contouring of CT and MR anatomical structures and known (diagnosed) brain metastases in adult patients for radiation therapy treatment planning. Must be used by qualified, trained medical professionals in conjunction with treatment planning systems for review, editing, and acceptance of contours.

Regulatory Classification

Identification

A medical image management and processing system is a device that provides one or more capabilities relating to the review and digital processing of medical images for the purposes of interpretation by a trained practitioner of disease detection, diagnosis, or patient management. The software components may provide advanced or complex image processing functions for image manipulation, enhancement, or quantification that are intended for use in the interpretation and analysis of medical images. Advanced image manipulation functions may include image segmentation, multimodality image registration, or 3D visualization. Complex quantitative functions may include semi-automated measurements or time-series measurements.

Special Controls

*Classification.* Class II (special controls; voluntary standards—Digital Imaging and Communications in Medicine (DICOM) Std., Joint Photographic Experts Group (JPEG) Std., Society of Motion Picture and Television Engineers (SMPTE) Test Pattern).

Predicate Devices

• AI Rad Companion Organs RT (K242745)

Reference Devices

• syngo.via RT Image Suite (VC10) (K252304)

Submission Summary (Full Text)

{0}

**FDA** U.S. FOOD & DRUG
ADMINISTRATION

July 10, 2026

Varian Medical Systems
Lynn Allman
Senior Director, Regulatory Affairs
3100 Hansen Way
Palo Alto, California 94304

Re: K261306

Trade/Device Name: AI Contouring (VA10A)

Regulation Number: 21 CFR 892.2050

Regulation Name: Medical Image Management And Processing System

Regulatory Class: Class II

Product Code: QKB

Dated: April 20, 2026

Received: April 20, 2026

Dear Lynn Allman:

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 (the 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 available 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.

U.S. Food & Drug Administration
10903 New Hampshire Avenue
Silver Spring, MD 20993
www.fda.gov

{1}

K261306 - Lynn Allman

Page 2

Additional information about changes that may require a new premarket notification are provided in the FDA guidance documents entitled "Deciding When to Submit a 510(k) for a Change to an Existing Device" (https://www.fda.gov/media/99812/download) and "Deciding When to Submit a 510(k) for a Software Change to an Existing Device" (https://www.fda.gov/media/99785/download).

Your device is also subject to, among other requirements, the Quality Management System Regulation (QMSR) (21 CFR Part 820), which includes, but is not limited to, ISO 13485 clause 7.3 (Design controls), ISO 13485 clause 8.3 (Nonconforming product), ISO 13485 clause 8.5.2 (Corrective action), and ISO 13485 clause 8.5.3 (Preventative action). Please note that regardless of whether a change requires premarket review, the QMSR requires device manufacturers to review and approve changes to device design and production (ISO 13485 clause 7.3 and ISO 13485 clause 7.5) and document changes and approvals in the Medical Device File (ISO 13485 clause 4.2.3).

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 (reporting of medical device-related adverse events) (21 CFR Part 803) for devices or postmarketing safety reporting (21 CFR Part 4, Subpart B) for combination products (see https://www.fda.gov/combination-products/guidance-regulatory-information/postmarketing-safety-reporting-combination-products); good manufacturing practice requirements as set forth in the Quality Management System Regulation (QMSR) (21 CFR Part 820) for devices or current good manufacturing practices (21 CFR Part 4, Subpart A) for combination products; and, if applicable, the electronic product radiation control provisions (Sections 531-542 of the Act); 21 CFR Parts 1000-1050.

All medical devices, including Class I and unclassified devices and combination product device constituent parts are required to be in compliance with the final Unique Device Identification System rule ("UDI Rule"). The UDI Rule requires, among other things, that a device bear a unique device identifier (UDI) on its label and package (21 CFR 801.20(a)) unless an exception or alternative applies (21 CFR 801.20(b)) and that the dates on the device label be formatted in accordance with 21 CFR 801.18. The UDI Rule (21 CFR 830.300(a) and 830.320(b)) also requires that certain information be submitted to the Global Unique Device Identification Database (GUDID) (21 CFR Part 830 Subpart E). For additional information on these requirements, please see the UDI System webpage at https://www.fda.gov/medical-devices/device-advice-comprehensive-regulatory-assistance/unique-device-identification-system-udi-system.

Also, please note the regulation entitled, "Misbranding by reference to premarket notification" (21 CFR 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-devices/medical-device-safety/medical-device-reporting-mdr-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/medical-devices/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-devices/device-advice-comprehensive-regulatory-

{2}

K261306 - Lynn Allman

Page 3

assistance/contact-us-division-industry-and-consumer-education-dice) for more information or contact DICE by email ([e-mail]) or phone [phone] or 301-796-7100).

Sincerely,

Lora D. Weidner, Ph.D.
Assistant Director
Radiation Therapy Team
DHT8C: Division of Radiological
Imaging and Radiation Therapy Devices
OHT8: Office of Radiological Health
Office of Product Evaluation and Quality
Center for Devices and Radiological Health

Enclosure

{3}

| Indications for Use | | |
| --- | --- | --- |
| Please type in the marketing application/submission number, if it is known. This textbox will be left blank for original applications/submissions. | K261306 | ? |
| Please provide the device trade name(s). | | ? |
| AI Contouring (VA10A) | | |
| Please provide your Indications for Use below. | | ? |
| AI Contouring is a post-processing software intended to automatically contour CT and MR structures, including known (diagnosed) brain metastases, using deep-learning-based algorithms. Contours that are generated by AI Contouring may be used as input for clinical workflows for radiation therapy treatment planning. AI Contouring must be used in conjunction with appropriate software such as Treatment Planning Systems and Interactive Contouring applications, to review, edit, and accept contours generated by AI Contouring. The outputs of AI Contouring are intended to be used by qualified and trained medical professionals. | | |
| Please select the types of uses (select one or both, as applicable). | ☑ Prescription Use (21 CFR 801 Subpart D) ☐ Over-The-Counter Use (21 CFR 801 Subpart C) | ? |

{4}

K261306

varian

A Siemens Healthineers Company

# Premarket Notification - 510(k) Summary

Traditional 510(k) Submission for AI Contouring (VA10A)

# I. Submitter's Name

Varian Medical Systems

3100 Hansen Way

Palo Alto, CA 94304

Contact Name: Lynn, Allman, PhD., Senior Director Regulatory Affairs

Phone: [phone]

E-mail: [e-mail]

Date Prepared: April 20, 2026

# II. Device Information

Proprietary Name: AI Contouring (VA10A)

Classification Name: Medical image management and processing system

Regulation Number: §892.2050

Product Code: QKB

# III. Predicate Device

AI Rad Companion Organs RT (K242745)

Reference Device: syngo.via RT Image Suite (VC10) (K252304)

# IV. Device Description

The AI Contouring VA10A software is intended for use in the automatic segmentation (autocontouring) of anatomical structures and pathologies on CT and MR images to support radiation therapy treatment planning. It is designed to assist clinicians by providing consistent, editable contours of organs at risk (OARs) and target volumes, streamlining the radiotherapy planning workflow.

The AI Contouring VA10A system consists of the following components:

- A software engine integrated into compatible Varian medical device platforms (e.g., Eclipse, Velocity)
- Version-controlled AI Contouring software modules for CT and MR image segmentation
- Four independent deep learning-based segmentation models for MR images
- Multiple deep image-to-image network (DI2IN) models for CT-based segmentation

The primary purpose of the system is to receive a data-stream from partner medical devices (Eclipse Treatment Planning System and Velocity) consisting of a voxel stream generated out of medical image files, compute structure sets representing anatomical structures that are specified by the configured template and return those results to the partner medical device as a voxel stream. The application includes capabilities and functionalities to support:

- Auto contouring of anatomical structures and known (diagnosed) brain metastases
- Template configuration

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 1 of 24

{5}

varian

A Siemens Healthineers Company

• Expansion of auto-contoured structures including

o Margin
o Boolean
o Cropping operations

The AI Contouring is a modular, deep learning-based system that includes:

• CT Autocontouring Module: Uses anatomical landmark detection and DI2IN segmentation to identify and contour structures in CT images.
• MR Autocontouring Module: Includes four independent models for:

• Brain Metastases (T1wPost MR)
• Brain Organs at Risk (OARs)
• Pelvis OARs (T1-weighted)
• Pelvis OARs (T2-weighted)

Each model is trained on large, diverse datasets and optimized for high accuracy and generalizability across imaging modalities and patient demographics. The subject device additionally supports contouring known (diagnosed) brain metastasis.

# V. Indications for Use

AI Contouring is a post-processing software intended to automatically contour CT and MR structures, including known (diagnosed) brain metastases, using deep-learning-based algorithms.

Contours that are generated by AI Contouring may be used as input for clinical workflows for radiation therapy treatment planning. AI Contouring must be used in conjunction with appropriate software such as Treatment Planning Systems and Interactive Contouring applications, to review, edit, and accept contours generated by AI Contouring.

The outputs of AI Contouring are intended to be used by qualified and trained medical professionals.

# VI. Comparison of Technological Characteristics with the Predicate Device

Table 1: Comparison of Technological Characteristics with the Predicate Device

| Feature | Predicate Device: AI-Rad Companion Organs RT (K242745) | Subject Device: AI Contouring VA10A | Comparison |
| --- | --- | --- | --- |
| Intended Use | AI-Rad Companion Organs RT is a post-processing software intended to automatically contour DICOM CT and MR pre-defined structures using deep learning algorithms. | AI Contouring is a post-processing software intended to automatically contour CT and MR structures using deep-learning-based algorithms. | The subject device and predicate device share a similar intended use. Both devices are intended for automated contouring of CT and MR structures for radiation therapy treatment planning. The subject device additionally supports contouring known (diagnosed) brain metastasis. The subject device additionally supports |
| Indications for Use | AI-Rad Companion Organs RT is a post-processing software intended to automatically contour DICOM CT and MR pre-defined structures using deep-learning-based algorithms. Contours that are generated by AI-Rad Companion Organs RT may be used as input for clinical workflows including external beam radiation therapy treatment planning. AI-Rad Companion Organs RT must be used in conjunction with appropriate software such as Treatment Planning Systems and Interactive | AI Contouring is a post-processing software intended to automatically contour CT and MR structures, including known (diagnosed) brain metastases, using deep-learning-based algorithms. Contours that are generated by AI Contouring may be used as input for clinical workflows for radiation therapy treatment planning. AI Contouring must be used in conjunction with appropriate software such as Treatment Planning Systems and Interactive Contouring | |

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 2 of 24

{6}

varian

A Siemens Healthineers Company

| | Contouring applications, to review, edit, and accept contours generated by AI-Rad Companion Organs RT. The outputs of AI-Rad Companion Organs RT are intended to be used by trained medical professionals. The software is not intended to automatically detect or contour lesions. | applications, to review, edit, and accept contours generated by AI Contouring. The outputs of AI Contouring are intended to be used by qualified and trained medical professionals. | contouring known (diagnosed) brain metastasis |
| --- | --- | --- | --- |
| Contraindications for Use | The intended patient population is not subject to any restrictions. However, automated contouring works best with adult patients. | The intended patient population is not subject to any restrictions. However, automated contouring works best with adult patients. | Same as predicate |
| Environment | Professional healthcare facilities | Professional healthcare facilities. The device is designed to be used in conjunction with a partner medical device on a workstation or a user provided computational environment to perform automated contouring in the radiation therapy context. The partner medical device provides DICOM medical image viewing and annotation capabilities, while AI contouring only processes images to create contours. | Substantially equivalent |
| Type of Users | Qualified healthcare professionals | Qualified healthcare professionals | Same as predicate |

The modified device, referred to as the “subject device” throughout this summary, is release version VA10A of the AI Contouring.

At a high level, both the predicate device and the subject device are based on the same characteristics:

- Both the subject device and the predicate are software tools used to automatically contour CT and MR structures
- They are computer-based software devices used by trained medical professionals with treatment planning systems
- The outputted contours by both devices are used as input for clinical workflows for radiation therapy treatment planning

AI Contouring contains a subset of features from the predicate and reference device. Below compares the Advanced Contouring feature against the predicate and reference device. The reference device, syngo.via RTiS VC10, is used to establish methods and test criteria to help establish substantial equivalence for AI Contouring. syngo.via RTiS VC10 is not used as a predicate device but serves to establish the acceptability of analytical and statistical methodologies, including segmentation accuracy metrics (e.g., Dice and ASSD) and non-inferiority testing frameworks. Given that AI Contouring employs the same underlying deep learning technology and derives from comparable datasets and validated algorithms, the use of syngo.via RTiS VC10 provides a scientifically justified benchmark for performance evaluation.

Table 2: Comparing Advanced Contouring between the subject and predicate/reference device

| Feature | Similarities | Differences in AI Contouring | Impact assessment |
| --- | --- | --- | --- |

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 3 of 24

{7}

varian

A Siemens Healthineers Company

| Advanced Contouring | The underlying deep learning technology for this extension is unchanged. It reuses the same technology. | AI Contouring does not include manual or adaptive contouring tools, but auto-contouring tools are substantially equivalent to AI Rad Companion and a subset of tools present in Syngo.Via RT Image Suite. | While AI Contouring is a distinct medical device from the predicate and reference devices, it re-uses the algorithms and technology from the predicate and reference devices with no changes. |
| --- | --- | --- | --- |

### VII. Summary of Performance Testing (Non-Clinical Testing)

The following performance data was provided in support of the substantial equivalence determination.

#### Software Verification and Validation Testing:

Software verification and validation was conducted, and enhanced documentation was provided as recommended by FDA's Guidance for Industry and FDA Staff, "Guidance for the Content of Premarket Submissions for Software Contained in Medical Devices."

Human Factors and Usability testing was conducted and documentation was provided as recommended in FDA's guidance document "Applying Human Factors and Usability Engineering to Medical Devices (Feb 2016)."

Cybersecurity and Interoperability requirements were assessed per FDA guidance's "Cybersecurity in Medical Devices: Quality System Considerations and Content of Premarket Submissions (Feb 2026)", "Postmarket Management of Cybersecurity in Medical Devices (Jan 2016)", "Design Considerations and Premarket Submission Recommendations for Interoperable Medical Devices (Jan 2016)".

Testing of AI Contouring followed a multi-level test approach, through unit, integration, and system (end-to-end) level testing. Software design verification and design validation testing was performed according to the FDA Quality System Regulation (21 CFR §820), ISO 13485 Quality Management System Standard, ISO 14971 Risk Management Standard, and IEC 62304 Software Life Cycle Process standard. Test results demonstrate conformance to applicable requirements specifications and assure hazard safeguards function properly. Software design verification and design validation testing was conducted, and documentation is provided as recommended by FDA's Guidance for Industry and FDA Staff, "Content of Premarket Submissions for Device Software Functions" (June 2023).

Test results demonstrate conformance to applicable requirements and specifications. No animal studies have been included in this pre-market submission.

### VIII. Summary of Clinical Validation

#### Mitigating Bias and Improving Generalizability

For training and validation of the algorithm, data was considered from different regions, such as North and South America, Asia, Australia, and Europe. This data provided a large variability in age, gender, geographic origin, and so forth. In addition, RT data acquired with different hardware and image settings was considered, for example, multiple CT or MR scanner vendors, and different slice thickness. The characteristics (for example, size and diversity) of both the data sets were chosen to achieve a reliable model that is robust against potential bias.

#### Imaging Equipment

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 4 of 24

{8}

varian

A Siemens Healthineers Company

For CT imaging data, AI Contouring has been tested and validated on multi-vendor data sets, including Siemens Healthineers, GE, and Philips scanners.

For MR Pelvic OAR (organ at risk), AI Contouring has been tested and validated on data sets from Siemens Healthineers, Philips, and GE scanners for T1w image series.

For MR Brain OAR and MR Brain Metastases, AI Contouring has been tested on data sets from Siemens Healthineers and GE scanners.

### Training Data

The performance of the AI Contouring algorithm has been evaluated for several clinically relevant subgroups. However, as the number of available patients was very limited, no analysis was conducted for the subgroup of CT images with/without contrast medium; and the subgroup of MR brain metastases with different field strengths.

CT AI Contouring: The composition of the CT training data (in cases where patient age was available) predominantly consisted of patients in the age group of 50–70 and those older than 70 years. While this is appropriate to the intended patient population for radiotherapy, there was a relatively small proportion of testing and training data representing patients below the age of 50. This indicates that a more careful review for this age group may be necessary.

CT AI Contouring on photon counting CT or multi-energy CT: Monoenergetic scans or virtual monoenergetic images (VMI) are recommended. Performance was evaluated using patient cases acquired with the NAEOTOM Alpha class using varying energies such as virtual monoenergetic images (VMIs) and convolution kernels (soft to sharp). The VMIs were evaluated between 50 keV and 100 keV, which cover the image contrasts achievable with a conventional X-ray spectrum between 80 kVp and 140 kVp.

MR AI Contouring: The training data for MR pelvis AI Contouring comprised noncontrast enhanced images of the male pelvis, with magnetic field strengths of 1.5 T and 3 T, and excluded prostatectomy cases. The training data for MR brain AI Contouring comprised contrast-enhanced MR images, with magnetic field strengths of 1.5 T and 3 T. The training data for the MR brain metastases algorithm comprised of intraparenchymal metastases of at least 2 mm in each dimension and not adjacent to a surgical bed, with images acquired preradiation therapy.

MR AI Contouring for brain metastases: Software is not intended to be used for diagnostic purposes. In case of additionally found metastases (compared to the diagnostic information), the physician/radiologist shall perform a review.

### Subgroup Analysis

#### CT Imaging

Subgroup analysis was performed to evaluate the performance in terms of average Dice according to the following subgroups:

- Manufacturer of the planning CT system: Siemens, GE, Philips
- Slice thickness: <= 1 mm, (1-2] mm, (2-3] mm, > 3 mm
- Gender: Male or Female

No notable variation between the subgroups was detected.

#### MR Imaging

Subgroup analysis was performed to evaluate the performance in terms of average Dice according to

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 5 of 24

{9}

varian

A Siemens Healthineers Company

the following subgroups:

• Manufacturer of the MR system:

• Brain OAR and Metastases: Siemens, GE
• Pelvis OAR: Siemens, GE, Philips

• Magnetic Field Strength of the MR System: 1.5 Tesla or 3.0 Tesla

• Gender: Male or Female.

No notable variation between the subgroups was detected.

# Algorithm Evaluation

The CT AI Contouring algorithm was tested on 469 subjects. The MR AI Contouring algorithm was tested on 153 subjects for pelvis OARs, 81 subjects for brain OARs, and 60 subjects for brain metastases. The data of the test cohort was randomly selected and was completely independent from the training and validation data.

A quantitative evaluation was performed in the form of fully automated bench tests that ensured the quantitative quality of the algorithm results by comparing them with a manually annotated ground truth.

Table 3: Distribution of test data across subgroups for CT auto-contouring

| Subgroup | # Test data sets |
| --- | --- |
| Data Source | Europe: 107, US: 142, Canada: 16, South America: 83, Australia: 29, Asia: 33, Unknown: 13 |
| Body Region | Head&Neck: 113, Thorax&Abdomen: 254, Pelvis: 92 |
| Gender | Male: 200, female: 218, Unknown: 51 |
| Age | <=30: 1 [30-50]: 6 [50;70]: 46 >70: 21 Unknown: 395 |
| Slice thickness (in mm) | <=1:20 {1,2}: 220 {2,3}: 209 >3: 20 |
| Manufacturer | Siemens: 148, GE: 90, Philips: 147, unknown/others: 81 |

Table 4: Distribution of test data across subgroups for MR auto-contouring of brain metastasis

| Subgroup | # Test data sets |
| --- | --- |
| Data Source | US: 11, Europe: 49, Unknown: 19 |
| Body Region | Intraparenchymal Brain |
| Sequence | MR T1W Post-contrast |
| Gender | Male: 24, Female: 17, Unknown: 19 |
| Age | [<30Y]: 3 [30Y – 50Y]: 16 [50Y – 60Y]: 10 [60Y – 70Y]: 17 [>70Y]: 14 |
| Slice Thickness | <1mm: 1 1mm: 40 1.2 mm: 15 |

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 6 of 24

{10}

varian

A Siemens Healthineers Company

| | 2.0mm: 3 2.2 mm: 1 |
| --- | --- |
| Field Strength | 1.5T: 48, 3.0T: 12 |
| Manufacturer | Siemens: 33, GE: 27 |

Table 5: Distribution of test data across subgroups for MR auto-contouring of brain OAR

| Subgroup | # Test data sets |
| --- | --- |
| Data Source | USA: 26, EU: 35, Unknown: 20 |
| Body Region | Brain Organs at Risk (OARs) |
| Sequence | MR T1W Post-contrast |
| Gender | Male: 8, female: 18, Unknown: 20 |
| Age | [<30Y]: 1 [30Y – 50Y]: 17 [50Y – 60Y]: 15 [60Y – 70Y]: 18 [>70Y]: 20 Unknown: 10 |
| Slice Thickness | <1mm: 5 1mm: 55 1.1mm:3 1.2 mm: 13 1.4mm: 1 2.0mm: 3 2.2 mm: 1 |
| Field strength | 1.5T: 61, 3.0T: 20 |
| Manufacturer | Siemens: 50, GE: 31 |

Table 6: Distribution of test data across subgroups for MR auto-contouring of pelvis OAR

| Subgroup | # Test data sets |
| --- | --- |
| Data Source | US (4 sites): 78; Europe (7 sites): 75; Australia (1 site): 1 |
| Body Region | Male Pelvis Organs at Risk |
| Sequence | T2 W TSE, T1 VIBEDixon W |
| Gender | Male: 154 |
| Age | [40Y – 50Y]: 11 [50Y – 60Y]: 8 [60Y – 70Y]: 35 [70Y – 80Y]: 31 > 80Y: 2 Unknown: 66 |
| Slice Thickness | <4mm |
| Field strength | 1.5T: 51, 3T: 102 |
| Manufacturer | Siemens: 66, GE: 18, Philips: 69 |

### Study Results

#### CT Contouring

The algorithm was evaluated on 469 subjects. The description of the test data can be split into two cohorts A and B. The data of both cohorts were randomly selected and are completely independent from the training and validation data. Cohort A corresponds to the test cohort of 413 patients used in the

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 7 of 24

{11}

varian

A Siemens Healthineers Company

predicate device. Cohort B was newly introduced in the subject device and contains image data of an additional 56 patients. The reference device for the 'Reference Standard' is syngo.via RTiS VC10.

Our acceptance criteria combine the statistical tests and the user evaluation - only structures that pass two or more tests could be included in the final models:

1. Statistical non-inferiority of the Dice score compared with the reference device.
2. Statistical non-inferiority of the ASSD score compared with the reference device. Average user evaluation of 3 or higher, when measured on a four-point scale.

AI Contouring reuses the same deep learning algorithms and code base as syngo.via RTiS, with identical technological characteristics and no modifications to the underlying autocontouring functionality. As a result, the scientific methodologies and performance metrics previously applied and accepted for syngo.via RTiS remain applicable to the subject device.

Table 7: Quantitative evaluation results of DICE for new organs in subject device

| Structure | Reference Standard: syngo.via RTiS VC10 | Subject Device | | | Equivalence Shown |
| --- | --- | --- | --- | --- | --- |
| | DICE Mean±Std.Dev | DICE Mean | DICE Std.Dev | Lower 95th % Confidence Interval | |
| Lacrimal Gland Left | 0.72±0.073 | 0.7 | 0.069 | 0.69 | Yes |
| Lacrimal Gland Right | 0.699±0.123 | 0.68 | 0.112 | 0.66 | Yes |
| Pituitary Gland | 0.758±0.074 | 0.73 | 0.095 | 0.71 | Yes |
| Humeral Head Left | 0.942±0.021 | 0.94 | 0.019 | 0.94 | Yes |
| Humeral Head Right | 0.946±0.032 | 0.94 | 0.025 | 0.94 | Yes |
| N2 Station 1: Highest Mediastinal Nodes Left | 0.707±0.1 | 0.7 | 0.1 | 0.67 | Yes |
| N2 Station 1: Highest Mediastinal Nodes Right | 0.66±0.128 | 0.67 | 0.108 | 0.64 | Yes |
| N2 Station 2: Upper Paratracheal Nodes Left | 0.676±0.076 | 0.67 | 0.072 | 0.65 | Yes |
| N2 Station 2: Upper Paratracheal Nodes Right | 0.555±0.136 | 0.51 | 0.146 | 0.47 | Yes |
| N2 Station 3A: Prevascular Nodes | 0.756±0.063 | 0.75 | 0.072 | 0.73 | Yes |
| N2 Station 3P: Retrotracheal Nodes | 0.659±0.06 | 0.66 | 0.057 | 0.65 | Yes |
| N2 Station 4: Lower Paratracheal Nodes Left | 0.657±0.084 | 0.6 | 0.114 | 0.58 | Yes |
| N2 Station 4: Lower Paratracheal Nodes Right | 0.728±0.08 | 0.71 | 0.086 | 0.69 | Yes |
| N2 Station 5: Subaortic Nodes | 0.61±0.148 | 0.59 | 0.146 | 0.55 | Yes |
| N2 Station 6: Para-aortic Nodes | 0.588±0.148 | 0.58 | 0.143 | 0.54 | Yes |
| N2 Station 7: SubCarinal Nodes | 0.59±0.076 | 0.6 | 0.064 | 0.58 | Yes |
| N2 Station 8: Paraesophageal Nodes | 0.658±0.08 | 0.67 | 0.071 | 0.65 | Yes |

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 8 of 24

{12}

varian

A Siemens Healthineers Company

| N2 Station 9: Pulmonary Ligament Nodes Left | 0.383±0.211 | 0.39 | 0.169 | 0.35 | Yes |
| --- | --- | --- | --- | --- | --- |
| N2 Station 9: Pulmonary Ligament Nodes Right | 0.388±0.156 | 0.4 | 0.147 | 0.36 | Yes |
| N1 Station 10: Hilar Nodes Left | 0.529±0.097 | 0.55 | 0.095 | 0.52 | Yes |
| N1 Station 10: Hilar Nodes Right | 0.502±0.095 | 0.52 | 0.09 | 0.49 | Yes |
| CA Left Circumflex (LCX) | 0.312 | 0.28 | 0.125 | 0.25 | Yes |
| CA Right Coronary Artery (RCA) | 0.38 | 0.27 | 0.119 | 0.24 | No |
| Bowel Bag | 0.95±0.036 | 0.95 | 0.033 | 0.94 | Yes |
| Femoral Head Left | 0.952±0.011 | 0.95 | 0.011 | 0.95 | Yes |
| Femoral Head Right | 0.949±0.013 | 0.95 | 0.03 | 0.94 | Yes |
| Hip Bone Left | 0.934±0.016 | 0.94 | 0.017 | 0.94 | Yes |
| Hip Bone Right | 0.939±0.013 | 0.95 | 0.016 | 0.95 | Yes |
| Sacrum | 0.907±0.04 | 0.92 | 0.032 | 0.92 | Yes |

Table 8: Quantitative evaluation results of ASSD for new organs in subject device

| Structure | Reference Standard: syngo.via RTiS VC10 | Subject Device | | | Equivalence Shown |
| --- | --- | --- | --- | --- | --- |
| | ASSD Mean±Std.Dev | ASSD Mean | ASSD Std.Dev | Upper 95th % Confidence Interval | |
| Lacrimal Gland Left | 0.8 ± 0.3 | 0.85 | 0.255 | 0.92 | Yes |
| Lacrimal Gland Right | 0.9 ± 0.7 | 1 | 0.612 | 1.15 | Yes |
| Pituitary Gland | 0.7 ± 0.3 | 0.76 | 0.39 | 0.85 | Yes |
| Humeral Head Left | 0.7 ± 0.3 | 0.7 | 0.319 | 0.78 | Yes |
| Humeral Head Right | 0.7 ± 0.5 | 0.74 | 0.418 | 0.84 | Yes |
| N2 Station 1: Highest Mediastinal Nodes Left | 2.4 ± 1.2 | 2.34 | 1.013 | 2.6 | Yes |
| N2 Station 1: Highest Mediastinal Nodes Right | 2.4 ± 0.9 | 2.36 | 0.784 | 2.56 | Yes |
| N2 Station 2: Upper Paratracheal Nodes Left | 1.4 ± 0.4 | 1.43 | 0.445 | 1.54 | Yes |
| N2 Station 2: Upper Paratracheal Nodes Right | 2.3 ± 1.5 | 2.58 | 1.541 | 2.98 | Yes |
| N2 Station 3A: Prevascular Nodes | 1.6 ±0.4 | 1.58 | 0.393 | 1.68 | Yes |
| N2 Station 3P: Retrotracheal Nodes | 1.6 ± 0.6 | 1.53 | 0.575 | 1.68 | Yes |
| N2 Station 4: Lower Paratracheal Nodes Left | 1.4 ± 0.7 | 1.8 | 0.998 | 2.06 | Yes |
| N2 Station 4: Lower Paratracheal Nodes Right | 1.7 ±0.7 | 1.72 | 0.724 | 1.91 | Yes |
| N2 Station 5: Subaortic Nodes | 1.8 ±0.8 | 1.84 | 0.921 | 2.08 | Yes |

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 9 of 24

{13}

varian

A Siemens Healthineers Company

| N2 Station 6: Para-aortic Nodes | 2 ±1.1 | 1.98 | 1.089 | 2.26 | Yes |
| --- | --- | --- | --- | --- | --- |
| N2 Station 7: SubCarinal Nodes | 1.9 ±0.9 | 1.78 | 0.719 | 1.97 | Yes |
| N2 Station 8: Paraesophageal Nodes | 2 ±0.9 | 1.83 | 0.766 | 2.03 | Yes |
| N2 Station 9: Pulmonary Ligament Nodes Left | 5.3 ± 4.4 | 4.28 | 3.417 | 5.17 | Yes |
| N2 Station 9: Pulmonary Ligament Nodes Right | 3.7 ± 2 | 3.3 | 1.758 | 3.76 | Yes |
| N1 Station 10: Hilar Nodes Left | 1.2 ±0.6 | 1.19 | 0.666 | 1.37 | Yes |
| N1 Station 10: Hilar Nodes Right | 1.4 ± 0.6 | 1.27 | 0.584 | 1.42 | Yes |
| CA Left Circumflex (LCX) | 4± 2.7 | 3.81 | 1.427 | 4.17 | Yes |
| CA Right Coronary Artery (RCA) | 5.6± 5.4 | 4.58 | 2.215 | 5.15 | Yes |
| Bowel Bag | 1.9± 1.5 | 2.31 | 2.153 | 2.92 | Yes |
| Femoral Head Left | 0.5 ± 0.2 | 0.55 | 0.144 | 0.59 | Yes |
| Femoral Head Right | 0.6 ± 0.2 | 0.64 | 0.484 | 0.77 | Yes |
| Hip Bone Left | 0.6 ± 0.2 | 0.46 | 0.164 | 0.5 | Yes |
| Hip Bone Right | 0.5 ± 0.1 | 0.4 | 0.151 | 0.44 | Yes |
| Sacrum | 0.8 ± 0.6 | 0.58 | 0.417 | 0.69 | Yes |

Table 9: Robustness assessment of structures for different image resolutions.

| Structure | Reference Standard: syngo.via RTiS VC10 | | Subject Device | | | Robustness shown |
| --- | --- | --- | --- | --- | --- | --- |
| | DICE | | DICE | | | |
| | Mean | Std. Dev | Mean | Std. Dev | Lower 95th % Confidence Interval | |
| Abdominopelvic Cavity | 0.93 | 0.032 | 1.00 | 0.012 | 1.00 | Yes |
| Aorta | 0.87 | 0.032 | 0.99 | 0.019 | 0.98 | Yes |
| Atrium Left | 0.86 | 0.048 | 0.99 | 0.005 | 0.99 | Yes |
| Atrium Right | 0.79 | 0.088 | 0.99 | 0.006 | 0.99 | Yes |
| Bladder | 0.95 | 0.050 | 0.99 | 0.018 | 0.96 | Yes |
| Body | 0.99 | 0.003 | 1.00 | 0.000 | 1.00 | Yes |
| Bowel Bag | 0.95 | 0.036 | 1.00 | 0.007 | 0.99 | Yes |
| Bowel Large | 0.90 | 0.046 | 0.99 | 0.013 | 0.97 | Yes |
| Bowel Small | 0.89 | 0.044 | 0.99 | 0.014 | 0.97 | Yes |
| Brachial Plexus Left | 0.63 | 0.086 | 0.94 | 0.062 | 0.89 | Yes |
| Brachial Plexus Right | 0.66 | 0.050 | 0.93 | 0.046 | 0.88 | Yes |
| Brain | 0.98 | 0.005 | 1.00 | 0.002 | 0.99 | Yes |
| Brainstem Brouwer et al. | 0.89 | 0.023 | 0.95 | 0.122 | 0.91 | Yes |
| Brainstem DAHANCA | 0.89 | 0.023 | 0.95 | 0.083 | 0.86 | Yes |
| CA Left Anterior Descending Artery (LAD) | 0.48 | 0.097 | 0.93 | 0.067 | 0.79 | Yes |

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 10 of 24

{14}

varian

A Siemens Healthineers Company

| Chest Wall Left | 0.92 | 0.015 | 1.00 | 0.004 | 0.99 | Yes |
| --- | --- | --- | --- | --- | --- | --- |
| Chest Wall Right | 0.93 | 0.012 | 0.99 | 0.004 | 0.99 | Yes |
| Cochlea Left | 0.75 | 0.161 | 0.91 | 0.083 | 0.79 | Yes |
| Cochlea Right | 0.80 | 0.048 | 0.90 | 0.109 | 0.82 | Yes |
| Duodenum | 0.77 | 0.101 | 0.97 | 0.049 | 0.93 | Yes |
| Esophagus DAHANCA | 0.81 | 0.059 | 0.97 | 0.021 | 0.94 | Yes |
| Esophagus RTOG | 0.81 | 0.059 | 0.98 | 0.019 | 0.95 | Yes |
| Eye Globe Left | 0.89 | 0.037 | 0.98 | 0.014 | 0.96 | Yes |
| Eye Globe Right | 0.89 | 0.026 | 0.98 | 0.040 | 0.92 | Yes |
| Female Breast Left ESTRO | 0.89 | 0.046 | 0.99 | 0.012 | 0.97 | Yes |
| Female Breast Left RTOG | 0.89 | 0.046 | 0.99 | 0.014 | 0.97 | Yes |
| Female Breast Right ESTRO | 0.85 | 0.061 | 0.99 | 0.013 | 0.97 | Yes |
| Female Breast Right RTOG | 0.85 | 0.061 | 0.99 | 0.022 | 0.96 | Yes |
| Femoral Head Left | 0.95 | 0.011 | 0.99 | 0.009 | 0.99 | Yes |
| Femoral Head Right | 0.95 | 0.013 | 0.99 | 0.007 | 0.99 | Yes |
| Glottis Brouwer et al. | 0.70 | 0.100 | 0.94 | 0.028 | 0.88 | Yes |
| Glottis DAHANCA | 0.70 | 0.100 | 0.96 | 0.036 | 0.92 | Yes |
| Heart | 0.92 | 0.027 | 1.00 | 0.007 | 0.99 | Yes |
| Hip Bone Left | 0.93 | 0.016 | 0.99 | 0.004 | 0.99 | Yes |
| Hip Bone Right | 0.94 | 0.013 | 0.99 | 0.003 | 0.99 | Yes |
| Humeral Head Left | 0.94 | 0.021 | 0.99 | 0.012 | 0.99 | Yes |
| Humeral Head Right | 0.95 | 0.032 | 0.99 | 0.014 | 0.99 | Yes |
| Kidney Left | 0.93 | 0.018 | 1.00 | 0.006 | 0.99 | Yes |
| Kidney Right | 0.91 | 0.067 | 0.99 | 0.017 | 0.99 | Yes |
| Lacrimal Gland Left | 0.72 | 0.073 | 0.95 | 0.029 | 0.91 | Yes |
| Lacrimal Gland Right | 0.70 | 0.123 | 0.92 | 0.122 | 0.66 | Yes |
| Lens Left | 0.68 | 0.188 | 0.92 | 0.111 | 0.67 | Yes |
| Lens Right | 0.67 | 0.118 | 0.94 | 0.086 | 0.75 | Yes |
| Lips | 0.79 | 0.070 | 0.91 | 0.162 | 0.64 | Yes |
| Liver | 0.96 | 0.013 | 1.00 | 0.003 | 0.99 | Yes |
| LN Axilla Level I Left | 0.81 | 0.054 | 0.99 | 0.009 | 0.98 | Yes |
| LN Axilla Level I Right | 0.80 | 0.071 | 0.99 | 0.009 | 0.97 | Yes |
| LN Axilla Level II Left | 0.79 | 0.067 | 0.99 | 0.013 | 0.96 | Yes |
| LN Axilla Level II Right | 0.78 | 0.069 | 0.99 | 0.012 | 0.97 | Yes |
| LN Axilla Level III Left | 0.75 | 0.041 | 0.98 | 0.016 | 0.95 | Yes |
| LN Axilla Level III Right | 0.76 | 0.076 | 0.99 | 0.015 | 0.95 | Yes |
| LN Common Iliac Left | 0.85 | 0.052 | 0.98 | 0.041 | 0.93 | Yes |
| LN Common Iliac Right | 0.82 | 0.047 | 0.97 | 0.053 | 0.94 | Yes |
| LN External Iliac Left | 0.89 | 0.035 | 0.98 | 0.067 | 0.94 | Yes |
| LN External Iliac Right | 0.88 | 0.030 | 0.98 | 0.039 | 0.95 | Yes |
| LN Internal Iliac Left | 0.83 | 0.070 | 0.97 | 0.059 | 0.93 | Yes |
| LN Internal Iliac Right | 0.84 | 0.061 | 0.97 | 0.037 | 0.94 | Yes |
| LN Internal Mammary Left | 0.59 | 0.080 | 0.96 | 0.061 | 0.88 | Yes |
| LN Internal Mammary Right | 0.63 | 0.094 | 0.97 | 0.029 | 0.92 | Yes |
| LN Level Ia Submental Triangle | 0.64 | 0.156 | 0.92 | 0.107 | 0.82 | Yes |

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 11 of 24

{15}

varian

A Siemens Healthineers Company

| LN Level Ib Submandibular Triangle Left | 0.80 | 0.062 | 0.94 | 0.142 | 0.82 | Yes |
| --- | --- | --- | --- | --- | --- | --- |
| LN Level Ib Submandibular Triangle Right | 0.77 | 0.085 | 0.93 | 0.118 | 0.74 | Yes |
| LN Level II Upper Jugular Nodes Left | 0.82 | 0.051 | 0.95 | 0.107 | 0.89 | Yes |
| LN Level II Upper Jugular Nodes Right | 0.80 | 0.062 | 0.94 | 0.135 | 0.87 | Yes |
| LN Level III Middle Jugular Nodes Left | 0.79 | 0.067 | 0.95 | 0.053 | 0.90 | Yes |
| LN Level III Middle Jugular Nodes Right | 0.77 | 0.085 | 0.92 | 0.141 | 0.68 | Yes |
| LN Level IVa Lower Jugular Group Left | 0.71 | 0.108 | 0.93 | 0.132 | 0.88 | Yes |
| LN Level IVa Lower Jugular Group Right | 0.73 | 0.084 | 0.90 | 0.180 | 0.63 | Yes |
| LN Level IVb Medial Supraclavicular Group Left | 0.66 | 0.157 | 0.96 | 0.024 | 0.92 | Yes |
| LN Level IVb Medial Supraclavicular Group Right | 0.70 | 0.109 | 0.96 | 0.026 | 0.90 | Yes |
| LN Level IX Bucco-facial Group Left | 0.63 | 0.115 | 0.94 | 0.075 | 0.89 | Yes |
| LN Level IX Bucco-facial Group Right | 0.60 | 0.123 | 0.94 | 0.060 | 0.86 | Yes |
| LN Level V Posterior Triangle Group Left | 0.71 | 0.116 | 0.94 | 0.108 | 0.87 | Yes |
| LN Level V Posterior Triangle Group Right | 0.69 | 0.121 | 0.94 | 0.058 | 0.80 | Yes |
| LN Level Vc Lateral Supraclavicular Group Left | 0.56 | 0.154 | 0.95 | 0.047 | 0.84 | Yes |
| LN Level Vc Lateral Supraclavicular Group Right | 0.61 | 0.122 | 0.94 | 0.062 | 0.84 | Yes |
| LN Level VIa Anterior Jugular Nodes | 0.70 | 0.066 | 0.90 | 0.182 | 0.66 | Yes |
| LN Level VIb Prelaryngeal, Pretracheal, & Paratracheal Nodes | 0.66 | 0.108 | 0.94 | 0.092 | 0.90 | Yes |
| LN Level VIIa Retropharyngeal Nodes Left | 0.53 | 0.100 | 0.85 | 0.170 | 0.47 | Yes |
| LN Level VIIa Retropharyngeal Nodes Right | 0.49 | 0.148 | 0.88 | 0.143 | 0.70 | Yes |
| LN Level VIIb Retro-styloid Nodes Left | 0.73 | 0.079 | 0.92 | 0.174 | 0.55 | Yes |
| LN Level VIIb Retro-styloid Nodes Right | 0.74 | 0.094 | 0.94 | 0.099 | 0.85 | Yes |
| LN Level VIII Parotid Group Left | 0.85 | 0.037 | 0.93 | 0.153 | 0.63 | Yes |

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 12 of 24

{16}

varian

A Siemens Healthineers Company

| LN Level VIII Parotid Group Right | 0.84 | 0.042 | 0.95 | 0.091 | 0.88 | Yes |
| --- | --- | --- | --- | --- | --- | --- |
| LN Level Xa Retroauricular & Subauricular Nodes Left | 0.69 | 0.099 | 0.94 | 0.077 | 0.82 | Yes |
| LN Level Xa Retroauricular & Subauricular Nodes Right | 0.73 | 0.074 | 0.95 | 0.034 | 0.88 | Yes |
| LN Level Xb Occipital Nodes Left | 0.57 | 0.110 | 0.91 | 0.135 | 0.69 | Yes |
| LN Level Xb Occipital Nodes Right | 0.56 | 0.134 | 0.90 | 0.177 | 0.69 | Yes |
| LN Obturator Left | 0.83 | 0.026 | 0.98 | 0.036 | 0.94 | Yes |
| LN Obturator Right | 0.83 | 0.040 | 0.97 | 0.052 | 0.94 | Yes |
| LN Presacral | 0.68 | 0.140 | 0.97 | 0.031 | 0.92 | Yes |
| LN Supraclavicular Left | 0.79 | 0.083 | 0.99 | 0.014 | 0.96 | Yes |
| LN Supraclavicular Right | 0.80 | 0.067 | 0.99 | 0.016 | 0.96 | Yes |
| Lung Lobe Left Lower | 0.90 | 0.121 | 0.99 | 0.025 | 0.98 | Yes |
| Lung Lobe Left Upper | 0.94 | 0.071 | 1.00 | 0.005 | 0.99 | Yes |
| Lung Lobe Right Lower | 0.93 | 0.044 | 0.99 | 0.039 | 0.98 | Yes |
| Lung Lobe Right Middle | 0.89 | 0.077 | 0.98 | 0.078 | 0.90 | Yes |
| Lung Lobe Right Upper | 0.93 | 0.052 | 0.99 | 0.013 | 0.98 | Yes |
| Mandible | 0.88 | 0.027 | 0.93 | 0.130 | 0.69 | Yes |
| N1 Station 10: Hilar Nodes Left | 0.53 | 0.097 | 0.95 | 0.101 | 0.84 | Yes |
| N1 Station 10: Hilar Nodes Right | 0.50 | 0.095 | 0.95 | 0.098 | 0.82 | Yes |
| N2 Station 1: Highest Mediastinal Nodes Left | 0.71 | 0.100 | 0.97 | 0.062 | 0.92 | Yes |
| N2 Station 1: Highest Mediastinal Nodes Right | 0.66 | 0.128 | 0.97 | 0.067 | 0.90 | Yes |
| N2 Station 2: Upper Paratracheal Nodes Left | 0.68 | 0.076 | 0.95 | 0.121 | 0.85 | Yes |
| N2 Station 2: Upper Paratracheal Nodes Right | 0.56 | 0.136 | 0.96 | 0.067 | 0.89 | Yes |
| N2 Station 3A: Prevascular Nodes | 0.76 | 0.063 | 0.94 | 0.162 | 0.63 | Yes |
| N2 Station 3P: Retrotracheal Nodes | 0.66 | 0.060 | 0.93 | 0.156 | 0.54 | Yes |
| N2 Station 4: Lower Paratracheal Nodes Left | 0.66 | 0.084 | 0.96 | 0.074 | 0.89 | Yes |
| N2 Station 4: Lower Paratracheal Nodes Right | 0.73 | 0.080 | 0.97 | 0.088 | 0.90 | Yes |
| N2 Station 5: Subaortic Nodes | 0.61 | 0.148 | 0.97 | 0.041 | 0.91 | Yes |
| N2 Station 6: Para-aortic Nodes | 0.59 | 0.148 | 0.97 | 0.034 | 0.90 | Yes |
| N2 Station 7: SubCarinal Nodes | 0.59 | 0.076 | 0.92 | 0.145 | 0.65 | Yes |

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 13 of 24

{17}

varian

A Siemens Healthineers Company

| N2 Station 8: Paraesophageal Nodes | 0.66 | 0.080 | 0.97 | 0.070 | 0.88 | Yes |
| --- | --- | --- | --- | --- | --- | --- |
| N2 Station 9: Pulmonary Ligament Nodes Left | 0.38 | 0.211 | 0.94 | 0.094 | 0.80 | Yes |
| N2 Station 9: Pulmonary Ligament Nodes Right | 0.39 | 0.156 | 0.95 | 0.079 | 0.84 | Yes |
| Optic Chiasm | 0.37 | 0.178 | 0.82 | 0.153 | 0.61 | Yes |
| Optic Nerve Left | 0.63 | 0.118 | 0.89 | 0.123 | 0.70 | Yes |
| Optic Nerve Right | 0.63 | 0.100 | 0.91 | 0.112 | 0.74 | Yes |
| Oral Cavity | 0.89 | 0.048 | 0.93 | 0.188 | 0.57 | Yes |
| Pancreas | 0.68 | 0.136 | 0.98 | 0.059 | 0.93 | Yes |
| Parotid Gland Left | 0.85 | 0.079 | 0.95 | 0.121 | 0.83 | Yes |
| Parotid Gland Right | 0.85 | 0.070 | 0.94 | 0.158 | 0.82 | Yes |
| Penile Bulb | 0.74 | 0.091 | 0.94 | 0.115 | 0.74 | Yes |
| Pharyngeal Constrictor Muscle Inferior | 0.78 | 0.048 | 0.96 | 0.022 | 0.91 | Yes |
| Pharyngeal Constrictor Muscle Middle | 0.67 | 0.075 | 0.93 | 0.086 | 0.86 | Yes |
| Pharyngeal Constrictor Muscle Superior | 0.66 | 0.055 | 0.88 | 0.157 | 0.53 | Yes |
| Pituitary Gland | 0.76 | 0.074 | 0.95 | 0.056 | 0.86 | Yes |
| Prostate | 0.87 | 0.055 | 0.97 | 0.106 | 0.86 | Yes |
| Proximal Bronchial Tree | 0.84 | 0.050 | 0.99 | 0.006 | 0.98 | Yes |
| Proximal Femur Left | 0.93 | 0.027 | 0.99 | 0.015 | 0.98 | Yes |
| Proximal Femur Right | 0.92 | 0.027 | 0.99 | 0.016 | 0.96 | Yes |
| Pulmonary Artery | 0.80 | 0.057 | 0.98 | 0.051 | 0.97 | Yes |
| Rectum | 0.82 | 0.090 | 0.97 | 0.102 | 0.88 | Yes |
| Rib Left 1 | 0.84 | 0.042 | 0.99 | 0.009 | 0.97 | Yes |
| Rib Left 10 | 0.77 | 0.184 | 0.98 | 0.028 | 0.96 | Yes |
| Rib Left 11 | 0.77 | 0.185 | 0.98 | 0.013 | 0.96 | Yes |
| Rib Left 12 | 0.80 | 0.071 | 0.97 | 0.031 | 0.94 | Yes |
| Rib Left 2 | 0.83 | 0.073 | 0.98 | 0.007 | 0.97 | Yes |
| Rib Left 3 | 0.82 | 0.037 | 0.98 | 0.007 | 0.97 | Yes |
| Rib Left 4 | 0.83 | 0.032 | 0.98 | 0.050 | 0.97 | Yes |
| Rib Left 5 | 0.83 | 0.031 | 0.98 | 0.009 | 0.97 | Yes |
| Rib Left 6 | 0.84 | 0.024 | 0.98 | 0.009 | 0.97 | Yes |
| Rib Left 7 | 0.83 | 0.030 | 0.97 | 0.081 | 0.97 | Yes |
| Rib Left 8 | 0.82 | 0.035 | 0.98 | 0.023 | 0.97 | Yes |
| Rib Left 9 | 0.82 | 0.023 | 0.98 | 0.066 | 0.97 | Yes |
| Rib Right 1 | 0.84 | 0.050 | 0.99 | 0.012 | 0.97 | Yes |
| Rib Right 10 | 0.79 | 0.189 | 0.98 | 0.064 | 0.96 | Yes |
| Rib Right 11 | 0.78 | 0.188 | 0.98 | 0.011 | 0.96 | Yes |
| Rib Right 12 | 0.82 | 0.043 | 0.97 | 0.044 | 0.92 | Yes |
| Rib Right 2 | 0.87 | 0.026 | 0.98 | 0.018 | 0.97 | Yes |
| Rib Right 3 | 0.85 | 0.038 | 0.98 | 0.045 | 0.97 | Yes |
| Rib Right 4 | 0.86 | 0.026 | 0.98 | 0.080 | 0.97 | Yes |
| Rib Right 5 | 0.86 | 0.025 | 0.98 | 0.068 | 0.97 | Yes |

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 14 of 24

{18}

varian

A Siemens Healthineers Company

| Rib Right 6 | 0.86 | 0.025 | 0.98 | 0.015 | 0.97 | Yes |
| --- | --- | --- | --- | --- | --- | --- |
| Rib Right 7 | 0.85 | 0.030 | 0.98 | 0.015 | 0.97 | Yes |
| Rib Right 8 | 0.81 | 0.106 | 0.98 | 0.009 | 0.97 | Yes |
| Rib Right 9 | 0.79 | 0.195 | 0.98 | 0.046 | 0.97 | Yes |
| Sacrum | 0.91 | 0.040 | 0.98 | 0.066 | 0.96 | Yes |
| Seminal Vesicles | 0.77 | 0.097 | 0.93 | 0.112 | 0.76 | Yes |
| Sigmoid | 0.80 | 0.107 | 0.96 | 0.099 | 0.78 | Yes |
| Spinal Canal | 0.85 | 0.071 | 0.98 | 0.010 | 0.97 | Yes |
| Spinal Cord | 0.67 | 0.101 | 0.98 | 0.013 | 0.96 | Yes |
| Spleen | 0.93 | 0.021 | 1.00 | 0.008 | 0.99 | Yes |
| Sternum | 0.88 | 0.031 | 0.99 | 0.007 | 0.98 | Yes |
| Stomach | 0.91 | 0.042 | 0.99 | 0.057 | 0.97 | Yes |
| Submandibular Gland Left | 0.87 | 0.044 | 0.94 | 0.104 | 0.86 | Yes |
| Submandibular Gland Right | 0.85 | 0.064 | 0.93 | 0.111 | 0.79 | Yes |
| Supraglottic Larynx Brouwer et al. | 0.77 | 0.089 | 0.95 | 0.036 | 0.88 | Yes |
| Supraglottic Larynx DAHANCA | 0.77 | 0.089 | 0.96 | 0.038 | 0.91 | Yes |
| Thyroid | 0.84 | 0.034 | 0.96 | 0.045 | 0.88 | Yes |
| Trachea | 0.89 | 0.045 | 1.00 | 0.005 | 0.98 | Yes |
| Uterus | 0.88 | 0.044 | 0.95 | 0.113 | 0.84 | Yes |
| Vena Cava Inferior | 0.76 | 0.126 | 0.99 | 0.026 | 0.97 | Yes |
| Vena Cava Superior | 0.79 | 0.068 | 0.98 | 0.028 | 0.95 | Yes |
| Ventricle Left | 0.86 | 0.041 | 0.99 | 0.009 | 0.98 | Yes |
| Ventricle Left Endocardium | 0.84 | 0.053 | 0.99 | 0.006 | 0.99 | Yes |
| Ventricle Right | 0.85 | 0.017 | 0.99 | 0.005 | 0.99 | Yes |

Table 10: Robustness assessment of structures for different image contrasts.

| Structure | Reference Standard: syngo.via RTiS VC10 | | Subject Device | | | Robustness shown |
| --- | --- | --- | --- | --- | --- | --- |
| | DICE | | DICE | | | |
| | Mean | Std. Dev | Mean | Std. Dev | Lower 95th % Confidence Interval | |
| Abdominopelvic Cavity | 0.93 | 0.032 | 0.99 | 0.007 | 0.98 | Yes |
| Aorta | 0.87 | 0.032 | 0.97 | 0.065 | 0.94 | Yes |
| Atrium Left | 0.86 | 0.048 | 0.98 | 0.010 | 0.97 | Yes |
| Atrium Right | 0.79 | 0.088 | 0.98 | 0.010 | 0.96 | Yes |
| Bladder | 0.95 | 0.050 | 0.98 | 0.031 | 0.95 | Yes |
| Body | 0.99 | 0.003 | 1.00 | 0.000 | 1.00 | Yes |
| Bowel Bag | 0.95 | 0.036 | 0.99 | 0.017 | 0.97 | Yes |
| Bowel Large | 0.90 | 0.046 | 0.98 | 0.025 | 0.95 | Yes |
| Bowel Small | 0.89 | 0.044 | 0.98 | 0.031 | 0.94 | Yes |
| Brachial Plexus Left | 0.63 | 0.086 | 0.89 | 0.069 | 0.77 | Yes |
| Brachial Plexus Right | 0.66 | 0.050 | 0.89 | 0.071 | 0.77 | Yes |
| Brain | 0.98 | 0.005 | 0.99 | 0.012 | 0.97 | Yes |
| Brainstem Brouwer et al. | 0.89 | 0.023 | 0.94 | 0.111 | 0.88 | Yes |

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 15 of 24

{19}

Varian
A Siemens Healthineers Company

| Brainstem DAHANCA | 0.89 | 0.023 | 0.94 | 0.097 | 0.87 | Yes |
| --- | --- | --- | --- | --- | --- | --- |
| CA Left Anterior Descending Artery (LAD) | 0.48 | 0.097 | 0.81 | 0.083 | 0.69 | Yes |
| Chest Wall Left | 0.92 | 0.015 | 0.98 | 0.014 | 0.97 | Yes |
| Chest Wall Right | 0.93 | 0.012 | 0.98 | 0.012 | 0.97 | Yes |
| Cochlea Left | 0.75 | 0.161 | 0.82 | 0.215 | 0.32 | Yes |
| Cochlea Right | 0.80 | 0.048 | 0.83 | 0.203 | 0.32 | Yes |
| Duodenum | 0.77 | 0.101 | 0.95 | 0.069 | 0.84 | Yes |
| Esophagus DAHANCA | 0.81 | 0.059 | 0.94 | 0.036 | 0.88 | Yes |
| Esophagus RTOG | 0.81 | 0.059 | 0.96 | 0.029 | 0.91 | Yes |
| Eye Globe Left | 0.89 | 0.037 | 0.98 | 0.012 | 0.96 | Yes |
| Eye Globe Right | 0.89 | 0.026 | 0.98 | 0.033 | 0.92 | Yes |
| Female Breast Left ESTRO | 0.89 | 0.046 | 0.98 | 0.063 | 0.95 | Yes |
| Female Breast Left RTOG | 0.89 | 0.046 | 0.96 | 0.040 | 0.91 | Yes |
| Female Breast Right ESTRO | 0.85 | 0.061 | 0.99 | 0.014 | 0.96 | Yes |
| Female Breast Right RTOG | 0.85 | 0.061 | 0.96 | 0.048 | 0.89 | Yes |
| Femoral Head Left | 0.95 | 0.011 | 0.98 | 0.043 | 0.97 | Yes |
| Femoral Head Right | 0.95 | 0.013 | 0.99 | 0.015 | 0.96 | Yes |
| Glottis Brouwer et al. | 0.70 | 0.100 | 0.91 | 0.089 | 0.77 | Yes |
| Glottis DAHANCA | 0.70 | 0.100 | 0.93 | 0.097 | 0.80 | Yes |
| Heart | 0.92 | 0.027 | 0.99 | 0.015 | 0.97 | Yes |
| Hip Bone Left | 0.93 | 0.016 | 0.99 | 0.010 | 0.97 | Yes |
| Hip Bone Right | 0.94 | 0.013 | 0.99 | 0.009 | 0.97 | Yes |
| Humeral Head Left | 0.94 | 0.021 | 0.98 | 0.056 | 0.97 | Yes |
| Humeral Head Right | 0.95 | 0.032 | 0.98 | 0.063 | 0.96 | Yes |
| Kidney Left | 0.93 | 0.018 | 0.99 | 0.033 | 0.97 | Yes |
| Kidney Right | 0.91 | 0.067 | 0.99 | 0.023 | 0.98 | Yes |
| Lacrimal Gland Left | 0.72 | 0.073 | 0.96 | 0.033 | 0.90 | Yes |
| Lacrimal Gland Right | 0.70 | 0.123 | 0.94 | 0.070 | 0.83 | Yes |
| Lens Left | 0.68 | 0.188 | 0.93 | 0.130 | 0.74 | Yes |
| Lens Right | 0.67 | 0.118 | 0.95 | 0.098 | 0.87 | Yes |
| Lips | 0.79 | 0.070 | 0.92 | 0.135 | 0.86 | Yes |
| Liver | 0.96 | 0.013 | 0.99 | 0.008 | 0.98 | Yes |
| LN Axilla Level I Left | 0.81 | 0.054 | 0.97 | 0.017 | 0.94 | Yes |
| LN Axilla Level I Right | 0.80 | 0.071 | 0.97 | 0.018 | 0.94 | Yes |
| LN Axilla Level II Left | 0.79 | 0.067 | 0.97 | 0.025 | 0.93 | Yes |
| LN Axilla Level II Right | 0.78 | 0.069 | 0.97 | 0.019 | 0.94 | Yes |
| LN Axilla Level III Left | 0.75 | 0.041 | 0.96 | 0.028 | 0.91 | Yes |
| LN Axilla Level III Right | 0.76 | 0.076 | 0.96 | 0.025 | 0.91 | Yes |
| LN Common Iliac Left | 0.85 | 0.052 | 0.88 | 0.162 | 0.53 | Yes |
| LN Common Iliac Right | 0.82 | 0.047 | 0.88 | 0.161 | 0.54 | Yes |
| LN External Iliac Left | 0.89 | 0.035 | 0.93 | 0.106 | 0.83 | Yes |
| LN External Iliac Right | 0.88 | 0.030 | 0.94 | 0.060 | 0.85 | Yes |
| LN Internal Iliac Left | 0.83 | 0.070 | 0.93 | 0.092 | 0.83 | Yes |
| LN Internal Iliac Right | 0.84 | 0.061 | 0.92 | 0.083 | 0.81 | Yes |
| LN Internal Mammary Left | 0.59 | 0.080 | 0.93 | 0.061 | 0.81 | Yes |
| LN Internal Mammary Right | 0.63 | 0.094 | 0.93 | 0.078 | 0.85 | Yes |

# **510(k) Summary**

Traditional 510(k) Application

AI Contouring

Page 16 of 24

{20}

varian

A Siemens Healthineers Company

| LN Level Ia Submental Triangle | 0.64 | 0.156 | 0.88 | 0.140 | 0.60 | Yes |
| --- | --- | --- | --- | --- | --- | --- |
| LN Level Ib Submandibular Triangle Left | 0.80 | 0.062 | 0.92 | 0.136 | 0.76 | Yes |
| LN Level Ib Submandibular Triangle Right | 0.77 | 0.085 | 0.91 | 0.131 | 0.62 | Yes |
| LN Level II Upper Jugular Nodes Left | 0.82 | 0.051 | 0.93 | 0.099 | 0.83 | Yes |
| LN Level II Upper Jugular Nodes Right | 0.80 | 0.062 | 0.92 | 0.089 | 0.71 | Yes |
| LN Level III Middle Jugular Nodes Left | 0.79 | 0.067 | 0.92 | 0.070 | 0.79 | Yes |
| LN Level III Middle Jugular Nodes Right | 0.77 | 0.085 | 0.85 | 0.214 | 0.28 | Yes |
| LN Level IVa Lower Jugular Group Left | 0.71 | 0.108 | 0.87 | 0.167 | 0.55 | Yes |
| LN Level IVa Lower Jugular Group Right | 0.73 | 0.084 | 0.82 | 0.226 | 0.30 | Yes |
| LN Level IVb Medial Supraclavicular Group Left | 0.66 | 0.157 | 0.89 | 0.125 | 0.62 | Yes |
| LN Level IVb Medial Supraclavicular Group Right | 0.70 | 0.109 | 0.90 | 0.093 | 0.72 | Yes |
| LN Level IX Bucco-facial Group Left | 0.63 | 0.115 | 0.92 | 0.091 | 0.83 | Yes |
| LN Level IX Bucco-facial Group Right | 0.60 | 0.123 | 0.92 | 0.068 | 0.84 | Yes |
| LN Level V Posterior Triangle Group Left | 0.71 | 0.116 | 0.91 | 0.117 | 0.71 | Yes |
| LN Level V Posterior Triangle Group Right | 0.69 | 0.121 | 0.90 | 0.124 | 0.71 | Yes |
| LN Level Vc Lateral Supraclavicular Group Left | 0.56 | 0.154 | 0.90 | 0.108 | 0.77 | Yes |
| LN Level Vc Lateral Supraclavicular Group Right | 0.61 | 0.122 | 0.88 | 0.112 | 0.69 | Yes |
| LN Level VIa Anterior Jugular Nodes | 0.70 | 0.066 | 0.86 | 0.175 | 0.58 | Yes |
| LN Level VIb Prelaryngeal, Pretracheal, & Paratracheal Nodes | 0.66 | 0.108 | 0.86 | 0.192 | 0.45 | Yes |
| LN Level VIIa Retropharyngeal Nodes Left | 0.53 | 0.100 | 0.86 | 0.148 | 0.61 | Yes |
| LN Level VIIa Retropharyngeal Nodes Right | 0.49 | 0.148 | 0.84 | 0.162 | 0.60 | Yes |
| LN Level VIIb Retro-styloid Nodes Left | 0.73 | 0.079 | 0.90 | 0.141 | 0.67 | Yes |
| LN Level VIIb Retro-styloid Nodes Right | 0.74 | 0.094 | 0.89 | 0.146 | 0.55 | Yes |

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 17 of 24

{21}

varian

A Siemens Healthineers Company

| LN Level VIII Parotid Group Left | 0.85 | 0.037 | 0.93 | 0.133 | 0.80 | Yes |
| --- | --- | --- | --- | --- | --- | --- |
| LN Level VIII Parotid Group Right | 0.84 | 0.042 | 0.93 | 0.091 | 0.81 | Yes |
| LN Level Xa Retroauricular & Subauricular Nodes Left | 0.69 | 0.099 | 0.93 | 0.095 | 0.86 | Yes |
| LN Level Xa Retroauricular & Subauricular Nodes Right | 0.73 | 0.074 | 0.94 | 0.048 | 0.86 | Yes |
| LN Level Xb Occipital Nodes Left | 0.57 | 0.110 | 0.91 | 0.119 | 0.74 | Yes |
| LN Level Xb Occipital Nodes Right | 0.56 | 0.134 | 0.89 | 0.140 | 0.70 | Yes |
| LN Obturator Left | 0.83 | 0.026 | 0.95 | 0.063 | 0.88 | Yes |
| LN Obturator Right | 0.83 | 0.040 | 0.94 | 0.069 | 0.86 | Yes |
| LN Presacral | 0.68 | 0.140 | 0.90 | 0.079 | 0.77 | Yes |
| LN Supraclavicular Left | 0.79 | 0.083 | 0.96 | 0.027 | 0.92 | Yes |
| LN Supraclavicular Right | 0.80 | 0.067 | 0.96 | 0.032 | 0.91 | Yes |
| Lung Lobe Left Lower | 0.90 | 0.121 | 0.99 | 0.029 | 0.98 | Yes |
| Lung Lobe Left Upper | 0.94 | 0.071 | 1.00 | 0.006 | 0.99 | Yes |
| Lung Lobe Right Lower | 0.93 | 0.044 | 0.99 | 0.065 | 0.99 | Yes |
| Lung Lobe Right Middle | 0.89 | 0.077 | 0.98 | 0.091 | 0.93 | Yes |
| Lung Lobe Right Upper | 0.93 | 0.052 | 0.99 | 0.015 | 0.98 | Yes |
| Mandible | 0.88 | 0.027 | 0.95 | 0.068 | 0.89 | Yes |
| N1 Station 10: Hilar Nodes Left | 0.53 | 0.097 | 0.88 | 0.163 | 0.57 | Yes |
| N1 Station 10: Hilar Nodes Right | 0.50 | 0.095 | 0.91 | 0.108 | 0.79 | Yes |
| N2 Station 1: Highest Mediastinal Nodes Left | 0.71 | 0.100 | 0.91 | 0.107 | 0.73 | Yes |
| N2 Station 1: Highest Mediastinal Nodes Right | 0.66 | 0.128 | 0.90 | 0.126 | 0.70 | Yes |
| N2 Station 2: Upper Paratracheal Nodes Left | 0.68 | 0.076 | 0.89 | 0.169 | 0.60 | Yes |
| N2 Station 2: Upper Paratracheal Nodes Right | 0.56 | 0.136 | 0.90 | 0.119 | 0.69 | Yes |
| N2 Station 3A: Prevascular Nodes | 0.76 | 0.063 | 0.88 | 0.223 | 0.23 | Yes |
| N2 Station 3P: Retrotracheal Nodes | 0.66 | 0.060 | 0.82 | 0.264 | 0.04 | Yes |
| N2 Station 4: Lower Paratracheal Nodes Left | 0.66 | 0.084 | 0.91 | 0.118 | 0.79 | Yes |
| N2 Station 4: Lower Paratracheal Nodes Right | 0.73 | 0.080 | 0.92 | 0.131 | 0.82 | Yes |
| N2 Station 5: Subaortic Nodes | 0.61 | 0.148 | 0.90 | 0.134 | 0.79 | Yes |
| N2 Station 6: Para-aortic Nodes | 0.59 | 0.148 | 0.90 | 0.102 | 0.78 | Yes |

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 18 of 24

{22}

varian

A Siemens Healthineers Company

| N2 Station 7: SubCarinal Nodes | 0.59 | 0.076 | 0.81 | 0.251 | 0.06 | Yes |
| --- | --- | --- | --- | --- | --- | --- |
| N2 Station 8: Paraesophageal Nodes | 0.66 | 0.080 | 0.88 | 0.156 | 0.56 | Yes |
| N2 Station 9: Pulmonary Ligament Nodes Left | 0.38 | 0.211 | 0.81 | 0.147 | 0.52 | Yes |
| N2 Station 9: Pulmonary Ligament Nodes Right | 0.39 | 0.156 | 0.82 | 0.146 | 0.57 | Yes |
| Optic Chiasm | 0.37 | 0.178 | 0.68 | 0.251 | 0.09 | Yes |
| Optic Nerve Left | 0.63 | 0.118 | 0.89 | 0.132 | 0.67 | Yes |
| Optic Nerve Right | 0.63 | 0.100 | 0.86 | 0.181 | 0.45 | Yes |
| Oral Cavity | 0.89 | 0.048 | 0.93 | 0.189 | 0.81 | Yes |
| Pancreas | 0.68 | 0.136 | 0.92 | 0.109 | 0.78 | Yes |
| Parotid Gland Left | 0.85 | 0.079 | 0.93 | 0.128 | 0.83 | Yes |
| Parotid Gland Right | 0.85 | 0.070 | 0.93 | 0.142 | 0.76 | Yes |
| Penile Bulb | 0.74 | 0.091 | 0.90 | 0.159 | 0.60 | Yes |
| Pharyngeal Constrictor Muscle Inferior | 0.78 | 0.048 | 0.92 | 0.064 | 0.81 | Yes |
| Pharyngeal Constrictor Muscle Middle | 0.67 | 0.075 | 0.92 | 0.092 | 0.80 | Yes |
| Pharyngeal Constrictor Muscle Superior | 0.66 | 0.055 | 0.89 | 0.120 | 0.66 | Yes |
| Pituitary Gland | 0.76 | 0.074 | 0.91 | 0.148 | 0.75 | Yes |
| Prostate | 0.87 | 0.055 | 0.91 | 0.190 | 0.46 | Yes |
| Proximal Bronchial Tree | 0.84 | 0.050 | 0.99 | 0.009 | 0.97 | Yes |
| Proximal Femur Left | 0.93 | 0.027 | 0.98 | 0.057 | 0.94 | Yes |
| Proximal Femur Right | 0.92 | 0.027 | 0.98 | 0.038 | 0.94 | Yes |
| Pulmonary Artery | 0.80 | 0.057 | 0.96 | 0.060 | 0.93 | Yes |
| Rectum | 0.82 | 0.090 | 0.95 | 0.137 | 0.85 | Yes |
| Rib Left 1 | 0.84 | 0.042 | 0.96 | 0.020 | 0.92 | Yes |
| Rib Left 10 | 0.77 | 0.184 | 0.95 | 0.094 | 0.90 | Yes |
| Rib Left 11 | 0.77 | 0.185 | 0.95 | 0.115 | 0.89 | Yes |
| Rib Left 12 | 0.80 | 0.071 | 0.93 | 0.127 | 0.82 | Yes |
| Rib Left 2 | 0.83 | 0.073 | 0.97 | 0.037 | 0.94 | Yes |
| Rib Left 3 | 0.82 | 0.037 | 0.97 | 0.020 | 0.94 | Yes |
| Rib Left 4 | 0.83 | 0.032 | 0.96 | 0.091 | 0.94 | Yes |
| Rib Left 5 | 0.83 | 0.031 | 0.96 | 0.081 | 0.94 | Yes |
| Rib Left 6 | 0.84 | 0.024 | 0.97 | 0.080 | 0.95 | Yes |
| Rib Left 7 | 0.83 | 0.030 | 0.96 | 0.108 | 0.91 | Yes |
| Rib Left 8 | 0.82 | 0.035 | 0.95 | 0.143 | 0.91 | Yes |
| Rib Left 9 | 0.82 | 0.023 | 0.96 | 0.121 | 0.92 | Yes |
| Rib Right 1 | 0.84 | 0.050 | 0.96 | 0.022 | 0.92 | Yes |
| Rib Right 10 | 0.79 | 0.189 | 0.96 | 0.095 | 0.92 | Yes |
| Rib Right 11 | 0.78 | 0.188 | 0.96 | 0.094 | 0.92 | Yes |
| Rib Right 12 | 0.82 | 0.043 | 0.93 | 0.124 | 0.75 | Yes |
| Rib Right 2 | 0.87 | 0.026 | 0.96 | 0.075 | 0.93 | Yes |
| Rib Right 3 | 0.85 | 0.038 | 0.96 | 0.086 | 0.93 | Yes |

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 19 of 24

{23}

varian

A Siemens Healthineers Company

| Rib Right 4 | 0.86 | 0.026 | 0.96 | 0.117 | 0.93 | Yes |
| --- | --- | --- | --- | --- | --- | --- |
| Rib Right 5 | 0.86 | 0.025 | 0.96 | 0.092 | 0.93 | Yes |
| Rib Right 6 | 0.86 | 0.025 | 0.96 | 0.080 | 0.94 | Yes |
| Rib Right 7 | 0.85 | 0.030 | 0.96 | 0.078 | 0.93 | Yes |
| Rib Right 8 | 0.81 | 0.106 | 0.96 | 0.075 | 0.93 | Yes |
| Rib Right 9 | 0.79 | 0.195 | 0.96 | 0.089 | 0.93 | Yes |
| Sacrum | 0.91 | 0.040 | 0.96 | 0.112 | 0.92 | Yes |
| Seminal Vesicles | 0.77 | 0.097 | 0.86 | 0.204 | 0.39 | Yes |
| Sigmoid | 0.80 | 0.107 | 0.92 | 0.156 | 0.59 | Yes |
| Spinal Canal | 0.85 | 0.071 | 0.97 | 0.016 | 0.95 | Yes |
| Spinal Cord | 0.67 | 0.101 | 0.95 | 0.026 | 0.90 | Yes |
| Spleen | 0.93 | 0.021 | 0.99 | 0.010 | 0.98 | Yes |
| Sternum | 0.88 | 0.031 | 0.96 | 0.042 | 0.90 | Yes |
| Stomach | 0.91 | 0.042 | 0.97 | 0.073 | 0.89 | Yes |
| Submandibular Gland Left | 0.87 | 0.044 | 0.92 | 0.142 | 0.80 | Yes |
| Submandibular Gland Right | 0.85 | 0.064 | 0.91 | 0.148 | 0.67 | Yes |
| Supraglottic Larynx Brouwer et al. | 0.77 | 0.089 | 0.95 | 0.044 | 0.86 | Yes |
| Supraglottic Larynx DAHANCA | 0.77 | 0.089 | 0.93 | 0.092 | 0.83 | Yes |
| Thyroid | 0.84 | 0.034 | 0.91 | 0.083 | 0.77 | Yes |
| Trachea | 0.89 | 0.045 | 0.99 | 0.007 | 0.98 | Yes |
| Uterus | 0.88 | 0.044 | 0.88 | 0.188 | 0.54 | Yes |
| Vena Cava Inferior | 0.76 | 0.126 | 0.96 | 0.059 | 0.93 | Yes |
| Vena Cava Superior | 0.79 | 0.068 | 0.96 | 0.052 | 0.90 | Yes |
| Ventricle Left | 0.86 | 0.041 | 0.95 | 0.041 | 0.92 | Yes |
| Ventricle Left Endocardium | 0.84 | 0.053 | 0.97 | 0.015 | 0.95 | Yes |
| Ventricle Right | 0.85 | 0.017 | 0.98 | 0.012 | 0.96 | Yes |

### MR Autocontouring

The algorithms were evaluated on 294 (252 unique) subjects. The test data collection is subdivided into three cohorts i.e., cohort A for Brain Metastases algorithm, cohort B for Brain OAR algorithm and cohort C for Pelvis OAR algorithms (T1 and T2 models) testing. The MR Brain Metastases algorithm was tested on 60 subjects The MR Brain OAR algorithm was tested on 81 subjects. The MR Pelvis OAR algorithms were tested on 153 subjects.

### Brain Metastases

Our acceptance criteria involve the following statistical tests - only structures that pass both the quantitative metrics are included in the final models:

1. Statistical non-inferiority of the Lesionwise DICE compared with the reference device.
2. Statistical non-inferiority of the Lesionwise Sensitivity compared with the reference device.

Table 11: Quantitative evaluation results of Lesionwise DICE for brain metastases in the subject device.

| Structure | Reference DICE Mean | Subject DICE Mean | Subject DICE Std | Lower 95thConfidence Interval | Equivalence Shown |
| --- | --- | --- | --- | --- | --- |
| Brain metastases | 0.79 | 0.74 | 0.17 | 0.72 | Yes |

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 20 of 24

{24}

varian

A Siemens Healthineers Company

Table 12: Quantitative evaluation results of Lesionwise Sensitivity for brain metastases in the subject device.

| Structure | Reference Sensitivity Mean (%) | Subject Sensitivity Mean (%) | Subject Sensitivity Std (%) | Lower 95^{th} Confidence Interval | Equivalence Shown |
| --- | --- | --- | --- | --- | --- |
| Brain metastases | 90.3 | 92.5 | 3.4 | 85.80 | Yes |

# Brain and Pelvis OAR

The acceptance criteria combine the statistical tests and the user evaluation - only structures that pass two or more tests could be included in the final models:

1. Statistical non-inferiority of the DICE score compared with the reference device.
2. Statistical non-inferiority of the ASSD score compared with the reference device. Average user evaluation of 3 or higher (3: usable with minor edits, 4: clinically usable).

Table 13: Quantitative evaluation results of DICE for Brain OAR structures in the subject device.

| Organ | Reference Standard: syngo.via RTiS VC10 | Subject Device | | | Equivalence shown |
| --- | --- | --- | --- | --- | --- |
| | Dice Mean | Dice Mean | Dice Std | Lower 95^{th} Confidence Interval | |
| **Brain OAR** | | | | | |
| Brainstem | 0.90 | 0.93 | 0.02 | 0.92 | Yes |
| Cochlea Left | 0.5 | 0.50 | 0.18 | 0.46 | Yes |
| Cochlea Right | 0.5 | 0.49 | 0.19 | 0.45 | Yes |
| Cornea Left | 0.5 | 0.52 | 0.16 | 0.48 | Yes |
| Cornea Right | 0.5 | 0.53 | 0.15 | 0.50 | Yes |
| Eye Left | 0.89±0.04 | 0.92 | 0.05 | 0.90 | Yes |
| Eye Right | 0.89±0.03 | 0.92 | 0.03 | 0.92 | Yes |
| Hippocampus Left | 0.66 | 0.77 | 0.07 | 0.75 | Yes |
| Hippocampus Right | 0.71 | 0.76 | 0.08 | 0.75 | Yes |
| Lacrimal Gland Left | 0.46 | 0.56 | 0.20 | 0.51 | Yes |
| Lacrimal Gland Right | 0.55 | 0.57 | 0.20 | 0.52 | Yes |
| Lens Left | 0.68±0.19 | 0.75 | 0.14 | 0.71 | Yes |
| Lens Right | 0.66±0.12 | 0.77 | 0.15 | 0.73 | Yes |
| Optic Chiasm | 0.55 | 0.66 | 0.11 | 0.63 | Yes |
| Optic Nerve Left | 0.49 | 0.54 | 0.15 | 0.51 | Yes |
| Optic Nerve Right | 0.56 | 0.57 | 0.14 | 0.54 | Yes |
| Pituitary Gland | 0.61 | 0.65 | 0.17 | 0.61 | Yes |
| Retina Left | 0.5 | 0.62 | 0.15 | 0.59 | Yes |
| Retina Right | 0.5 | 0.63 | 0.11 | 0.60 | Yes |

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 21 of 24

{25}

varian

A Siemens Healthineers Company

| Spinal cord | 0.68±0.10 | 0.86 | 0.09 | 0.84 | Yes |
| --- | --- | --- | --- | --- | --- |
| T1 Pelvis OAR | | | | | |
| Body | 0.98 | 0.98 | 0.01 | 0.98 | Yes |
| Femur head Left | 0.94 | 0.93 | 0.03 | 0.92 | Yes |
| Femur head Right | 0.94 | 0.94 | 0.02 | 0.93 | Yes |
| T2 Pelvis OAR | | | | | |
| Anus | 0.76 | 0.73 | 0.10 | 0.71 | Yes |
| Bladder | 0.91 | 0.90 | 0.08 | 0.88 | Yes |
| Penile Bulb | 0.82 | 0.81 | 0.09 | 0.79 | Yes |
| Prostate | 0.85 | 0.86 | 0.07 | 0.84 | Yes |
| Rectum | 0.85 | 0.86 | 0.06 | 0.85 | Yes |
| Seminal Vesicles | 0.66 | 0.72 | 0.14 | 0.69 | Yes |

Table 14: Quantitative evaluation results of ASSD for Brain OAR structures in the subject device.

| Structure | Reference Standard | Subject Device | | | Equivalence shown |
| --- | --- | --- | --- | --- | --- |
| | MD/ASSD Mean±Std | ASSD Mean (mm) | ASSD Std (mm) | Upper 95% Confidence Interval (mm) | |
| Brain OAR | | | | | |
| Brainstem | 0.9±0.16 | 0.54 | 0.15 | 0.57 | Yes |
| Cochlea Left | NA | 1.11 | 0.84 | 1.30 | - |
| Cochlea Right | NA | 1.07 | 0.76 | 1.23 | - |
| Cornea Left | 0.60±0.21 | 0.59 | 0.35 | 0.66 | Yes |
| Cornea Right | 0.60±0.21 | 0.66 | 0.80 | 0.83 | No |
| Eye Left | 0.60±0.24 | 0.40 | 0.24 | 0.45 | Yes |
| Eye Right | 0.62±0.17 | 0.36 | 0.15 | 0.40 | Yes |
| Hippocampus Left | 0.75±0.75 | 0.72 | 0.36 | 0.79 | Yes |
| Hippocampus Right | 0.67±0.47 | 0.71 | 0.31 | 0.77 | Yes |
| Lacrimal Gland Left | 1.39±0.97 | 1.60 | 1.58 | 1.95 | Yes |
| Lacrimal Gland Right | 1.31±0.95 | 1.48 | 1.30 | 1.77 | Yes |
| Lens Left | 0.61±0.56 | 0.43 | 0.30 | 0.50 | Yes |
| Lens Right | 0.60±0.25 | 0.42 | 0.36 | 0.50 | Yes |
| Optic Chiasm | 0.69±0.80 | 0.69 | 0.66 | 0.84 | Yes |
| Optic Nerve Left | 1.47±2.17 | 1.21 | 2.70 | 1.80 | Yes |
| Optic Nerve Right | 0.95±0.65 | 0.85 | 0.86 | 1.04 | Yes |

510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 22 of 24

{26}

A Siemens Healthineers Company

| Pituitary Gland | 0.94±0.26 | 0.76 | 0.47 | 0.86 | Yes |
| --- | --- | --- | --- | --- | --- |
| Retina Left | 0.61±0.21 | 0.43 | 0.50 | 0.54 | Yes |
| Retina Right | 0.61±0.21 | 0.38 | 0.14 | 0.40 | Yes |
| Spinal cord | 1.70±0.72 | 0.51 | 0.47 | 0.61 | Yes |
| **T1 Pelvis OAR** | | | | | |
| Body | 2.07±1.81 | 1.40 | 1.56 | 1.82 | Yes |
| Femur head Left | 0.85±0.59 | 1.10 | 0.59 | 1.26 | Yes |
| Femur head Right | 0.88±0.54 | 0.96 | 0.48 | 1.09 | Yes |
| **T2 Pelvis OAR** | | | | | |
| Anus | 2.09±0.9 | 2.24 | 1.16 | 2.47 | Yes |
| Bladder | 1.44±0.89 | 1.36 | 0.85 | 1.53 | Yes |
| Penile Bulb | 0.79±0.75 | 0.91 | 0.76 | 1.07 | Yes |
| Prostate | 1.56±0.54 | 1.66 | 2.01 | 2.08 | Yes |
| Rectum | 2.34±2.17 | 1.92 | 2.01 | 2.32 | Yes |
| Seminal Vesicles | 2.43±2.61 | 1.78 | 1.25 | 2.05 | Yes |

#### Use of Consensus Standards

The following list of FDA-recognized, voluntary consensus standards were utilized in the design and evaluation of the subject device's safety and efficacy.

| ISO 14971:2022 | Medical devices - Application of risk management to medical devices |
| --- | --- |
| ISO 15223-1:2022 | Medical devices - Symbols to be used with medical device labels, labelling and information to be supplied - Part 1: General requirements |
| ISO 20417:2022 | Information supplied by the manufacturer of medical devices |
| IEC 62304:2006 + A1:2016 | Medical Device Software - Software Lifecycle processes |
| IEC 62366-1:2021 | Application of Usability Engineering to Medical Devices |
| IEC 82304-1:2018 | Health software Part 1: General requirements for product safety |

#### Risk Management

Risk management is conducted in compliance with EN ISO 14971:2019 and Siemens Healthineers' internal process (QR7). Hazards and risks are systematically identified, evaluated, and mitigated through defined control measures. These are verified, validated, and continuously monitored through post-market surveillance (PMS) and post-market clinical follow-up (PMCF).

The latest risk documentation reflects adjustments to risk evaluations, removal of obsolete hazards, and validation of updated control measures. Usability-related risks and software warnings have been reviewed, with relevant IFU content and GUI messages updated accordingly.

#### IX. Determination of Substantial Equivalence to the Predicate

##### 510(k) Summary

Traditional 510(k) Application

AI Contouring

Page 23 of 24

{27}

**varian**
A Siemens Healthineers Company

AI Contouring contains a subset of software features and characteristics of the predicate and references devices. The principle of operation of the subject device is the same as that of the existing predicate device. Verification and validation demonstrate that the subject device is as safe and effective as the predicate. AI Contouring VA10A is comparable to AI-Rad Companion Organs RT and has similar performance metrics. Varian therefore believes that the subject device is substantially equivalent to the predicate device.

**510(k) Summary**
Traditional 510(k) Application
AI Contouring

Page 24 of 24