AiMIFY is an image processing software that can be used for image enhancement in MRI images. It can be used to increase contrast-to-noise ratio (CNR), contrast enhancement (CEP), and lesion-to-brain ratio (LBR) of enhancing tissue in brain MRI images acquired with a gadolinium-based contrast agent. It is intended to enhance MRI images acquired using standard approved dosage per the contrast agent's instructions for use.

Device Description

The AiMIFY device is a software as a medical device consisting of a machine learning software algorithm that enhances images taken by MRI scanners. AiMIFY consists of a software algorithm that improves contrast-to-noise ratio (CNR), contrast enhancement (CEP), and lesion-to-brain ratio (LBR) of Gadolinium-Based Contrast Agent (GBCA) enhanced T1-weighted images while maintaining diagnostic performance, using deep learning technology. It is a post-processing software that does not directly interact with the MR scanner and does not have a graphical user interface. It is intended to be used by radiologists in an imaging center, clinic, or hospital. The AiMIFY software uses T1 pre and post-contrast MR images acquired as part of standard of care contrast-enhanced MRI exams as the software input. The outputs are the corresponding images with enhanced contrast presence. AiMIFY enhances DICOM images.

AiMIFY image processing software uses a convolutional network based algorithm to enhance the AiMIFY-contrast images from pre-contrast and standard-dose post-contrast images. The image processing can be performed on MRI images with predefined or specific acquisition protocol settings as follows: gradient echo (pre- and post-contrast), 3D BRAVO (pre- and post-contrast), 3D MPRAGE (preand post-contrast), 2D T1 spin echo (pre- and post-contrast), T1 FLAIR/ inversion recovery spin echo (pre- and post-contrast).

The AiMIFY image is created by AiMIFY and sent back to the picture archiving and communication system (PACS) or other DICOM node by the compatible MDDS for clinical review.

Because the software runs in the background, it has no user interface. It is intended to be used by radiologists in an imaging center, clinic, or hospital.

Note, depending on the functionality of the compatible MDDS, AiMIFY can be used within the facility's network or remotely. The AiMFY device itself is not networked and therefore does not increase the cybersecurity risk of its users. Users are provided cybersecurity recommendations in labeling.

AI/ML Overview

Here's an analysis of the acceptance criteria and the study proving the device meets those criteria, based on the provided text.

Device: AiMIFY (1.x)
Indications for Use: Image processing software for enhancement of MRI images (increase CNR, CEP, LBR of enhancing tissue in brain MRI images acquired with gadolinium-based contrast agent).

1. Acceptance Criteria and Reported Device Performance

Table of Acceptance Criteria and Reported Device Performance:

Metric	Acceptance Criteria	Reported Device Performance
Quantitative Assessment
CNR (Contrast-to-Noise Ratio) Improvement	On average, improved by >= 50% after AiMIFY enhancement compared to traditionally acquired contrast images.	Achieved: 559.94% across all 95 cases; 831.70% for 57 lesion-only cases. Significantly higher than standard post-contrast images (Wilcoxon signed-rank test, p < 0.0001).
LBR (Lesion-to-Brain Ratio) Improvement	On average, improved by >= 50% after AiMIFY enhancement compared to traditionally acquired contrast images. (Inferred from primary endpoint definition encompassing CNR, LBR, CEP)	Achieved: 62.07% across all 95 cases; 58.80% for 57 lesion-only cases. Significantly better than standard post-contrast images (Wilcoxon signed-rank test, p-value < 0.0001).
CEP (Contrast Enhancement Percentage) Improvement	On average, improved by >= 50% after AiMIFY enhancement compared to traditionally acquired contrast images. (Inferred from primary endpoint definition encompassing CNR, LBR, CEP)	Achieved: 133.29% across all 95 cases; 101.80% for 57 lesion-only cases. Significantly better than standard post-contrast images (Wilcoxon signed-rank test, p-value < 0.0001).
Qualitative Assessment (Reader Study)
Perceived Visibility of Lesion Features (Lesion Contrast Enhancement, Border Delineation, Internal Morphology)	Statistically significantly better for AiMIFY processed images per the Wilcoxon signed-rank test by p < 0.05.	Achieved: Significantly better than standard post-contrast by p < 0.0001 for all three features.
Perceived Image Quality and Artifact Presence And Impact On Clinical Diagnosis	NOT statistically significantly worse than standard post-contrast images per the Wilcoxon signed-rank test by p < 0.05.	Achieved: Significantly not worse than standard post-contrast by p < 0.0001. Two of three readers demonstrated Perceived Image Quality is better than standard post-contrast by p < 0.0001.
Radiomics Analysis
CCC for Lesion Tissue (7 feature classes)	>= 0.65	Achieved: Ranged from 0.68 to 0.89 for lesion tissue.
CCC for Parenchyma Tissue (7 feature classes)	>= 0.8	Achieved: Ranged from 0.82 to 0.92 for parenchyma tissue.
SubtleMR Denoising Module Performance
Visibility of Small Structures	Average scores between original and SubtleMR enhanced images <= 0.5 Likert scale points.	Achieved: Average score difference was 0.05 points.
Perceived SNR, Image Quality, Artifacts	Average scores difference between original and SubtleMR enhanced images <= 0.5 Likert scale points. (Measured for Septum Pellucidum, Cranial Nerves, Cerebellar Folia)	Achieved: SNR differences: 0.05 (Septum Pellucidum), 0.08 (Cranial Nerves), 0.07 (Cerebellar Folia). Image quality/diagnostic confidence differences: 0.11 (Septum Pellucidum), 0.04 (Cranial Nerves), -0.05 (Cerebellar Folia). Imaging artifacts differences: 0.11 (Septum Pellucidum), 0.14 (Cranial Nerves), 0.05 (Cerebellar Folia).
SNR Improvement from SubtleMR	>= 5% (Acceptance criteria established in SubtleMR validation K223623)	Achieved: Average SNR improvement was 14%.

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

Test Set Sample Size: 95 T1 brain cases.
- Of these, 57 cases had identified lesions and were used for lesion-specific analyses (e.g., LBR, lesion-specific CNR).
Data Provenance: Retrospective, acquired from clinical sites or hospitals.
- Country of Origin: USA (California, New York, Nationwide), Beijing, China.
- Acquisition details: Variety of T1 input protocols (BRAVO, MPRAGE+, FLAIR, FSE), orientations (axial, sagittal, coronal), acquisition types (2D, 3D), field strengths (0.3T, 1.5T, 3.0T), and MR scanner vendors (GE, Philips, Siemens, Hitachi).
- Patient Demographics: Age (7 to 86, relatively even distribution), Sex (relatively even distribution of females and males), Pathologies (Cerebritis, Glioma, Meningioma, Metastases, Multiple Sclerosis, Neuritis, Inflammation, Other tumor related, other abnormalities).

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

Quantitative Assessment (ROI drawing): One board-certified radiologist.
Qualitative Assessment (Reader Study): Three board-certified neuro-radiologists.
- Specific years of experience are not mentioned, but "board-certified" implies a certain level of qualification and experience within their specialty.

4. Adjudication Method for the Test Set

Quantitative Assessment: ROIs were drawn by a single board-certified radiologist. No explicit mention of adjudication or multiple expert consensus for the initial ROI placement. The statistical analysis (Wilcoxon signed-rank test) focuses on the comparison of metrics derived from these ROIs.
Qualitative Assessment (Reader Study): The readers individually rated images on Likert scales. The results are presented as aggregated statistics (e.g., "significantly better/not worse by p<0.0001"). There is no mention of an adjudication process (e.g., 2+1, 3+1) to arrive at a single consensus ground truth or final rating for each case from the multiple radiologists.
- For exploratory endpoints, such as false lesion analysis, it's mentioned that "100% of cases received scores from all readers that the Standard-of-Care image was sufficient to identify the false lesion(s)," indicating agreement, but this is not a formal adjudication process for establishing ground truth from disagreements.

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

Yes, a MRMC study was performed. The "Qualitative Assessment (Reader Study)" involved three board-certified neuro-radiologists evaluating cases.
Effect Size of Human Reader Improvement with AI vs. Without AI Assistance:
- The study design presented is a comparison of standard post-contrast images vs. AiMIFY-enhanced images, evaluated by human readers. It assesses if AiMIFY improves perceived image quality and lesion features.
- The results show improvement in features like "Lesion Contrast Enhancement, Border Delineation, and Internal Morphology" (p < 0.0001 compared to standard post-contrast). Perceived Image Quality was "not worse" and even "better" for two of three readers (p < 0.0001).
- This study directly demonstrates the improvement in image characteristics for human readers when viewing AiMIFY-enhanced images. It does not, however, describe a comparative effectiveness study showing how much human readers' diagnostic accuracy or confidence improves when assisted by AI vs. not assisted. The study focuses on the image enhancement characteristics as perceived by readers rather than a change in diagnostic outcome or reader performance statistics.

6. If a Standalone (Algorithm Only Without Human-in-the-Loop Performance) Was Done

Yes, a standalone assessment was performed. The "Quantitative Assessment (Bench Test)" evaluated the algorithm's performance directly by comparing calculated metrics (CNR, LBR, CEP) from AiMIFY-processed images against standard post-contrast images. This assessment did not involve human readers' diagnostic interpretation of the images but rather quantifiable improvements generated by the algorithm itself.

7. The Type of Ground Truth Used

Quantitative Assessment: The ground truth for calculating CNR, LBR, and CEP was based on ROIs drawn by a single board-certified radiologist, identifying enhancing lesions and brain parenchyma. This can be considered a form of expert-defined ground truth based on anatomical and radiological characteristics. The lesions themselves were "identified" in the test datasets, suggesting a pre-existing clinical determination of their presence.
Qualitative Assessment: The ground truth for "lesion presence" in the Qualitative Assessment was presumably based on cases identified to "have lesions" in the initial test dataset (57 out of 95 cases). The evaluation itself was subjective (Likert scale ratings of perceived visibility, quality, etc.), with readers comparing the standard and AiMIFY images. This relies on the subjective judgment of multiple experts rather than an independent "true" ground truth like pathology.

8. The Sample Size for the Training Set

The document does not explicitly state the sample size of the training set.
It mentions that the training and validation datasets were compared for CNR increase, and that the training data compared low-dose to regular-dose post-contrast images, but provides no numerical size for the training set itself.

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

The document does not explicitly describe how the ground truth for the training set was established.
It implies that the training data involved "low-dose to regular-dose post-contrast images," suggesting that perhaps the ground truth for training the enhancement model was the "regular-dose" image, or that the model was trained to transform low-signal images into higher-signal enhanced images. However, specifics on how the "true" enhanced state or lesion characteristics within the training data were determined are not provided.

Summary

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

August 21, 2024

Image /page/0/Picture/1 description: The image shows the logo for 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.

Subtle Medical, Inc. % Jared Seehafer Regulatory Consultant Enzyme Corporation 611 Gateway Blvd Ste 120 South San Francisco, California 94080

Re: K240290

Trade/Device Name: AiMIFY (1.x) Regulation Number: 21 CFR 892.2050 Regulation Name: Medical Image Management And Processing System Regulatory Class: Class II Product Code: LLZ Dated: July 18, 2024 Received: July 18, 2024

Dear Jared Seehafer:

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.

FDA's substantial equivalence determination also included the review and clearance of your Predetermined Change Control Plan (PCCP). Under section 515C(b)(1) of the Act, a new premarket notification is not required for a change to a device cleared under section 510(k) of the Act, if such change is consistent with an

{1}------------------------------------------------

established PCCP granted pursuant to section 515C(b)(2) of the Act. Under 21 CFR 807.81(a)(3), a new premarket notification is required if there is a major change or modification in the intended use of a device, or if there is a change or modification in a device that could significantly affect the safety or effectiveness of the device, e.g., a significant change or modification in design, material, chemical composition, energy source, or manufacturing process. Accordingly, if deviations from the established PCCP result in a major change or modification in the intended use of the device, or result in a change or modification in the device that could significantly affect the safety or effectiveness of the a new premarket notification would be required consistent with section 515C(b)(1) of the Act and 21 CFR 807.81(a)(3). Failure to submit such a premarket submission would constitute adulteration and misbranding under sections 501(f)(1)(B) and 502(o) of the Act, respectively.

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 System (OS) regulation (21 CFR Part 820), which includes, but is not limited to, 21 CFR 820.30, Design controls; 21 CFR 820.90, Nonconforming product; and 21 CFR 820.100, Corrective and preventive action. Please note that regardless of whether a change requires premarket review, the QS regulation requires device manufacturers to review and approve changes to device design and production (21 CFR 820.30 and 21 CFR 820.70) and document changes and approvals in the device master record (21 CFR 820.181).

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 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-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 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.

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-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-regulatory

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

assistance/contact-us-division-industry-and-consumer-education-dice) for more information or contact DICE by email (DICE@fda.hhs.gov) or phone (1-800-638-2041 or 301-796-7100).

Sincerely,

for

Douglas W. Fletcher
Digitally signed by
Douglas W. Fletcher -S

Daniel M. Krainak, Ph.D.

Assistant Director Magnetic Resonance and Nuclear Medicine 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

510(k) Number (if known) K240290

Device Name AiMIFY (1.x)

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 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."

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

AiMIFY 510(k) Summary

Date Summary Prepared:	2024-08-20
Contact Details
Applicant Name:	Subtle Medical, Inc.
Applicant Address:	883 Santa Cruz Ave, Suite 205Menlo Park, CA 94025 United States
Applicant Contact:	Ms. Ronny Elor
Applicant Contact Telephone:	(925) 324-8467
Applicant Contact Email:	ronny@subtlemedical.com
Correspondent Name:	Enzyme Corporation
Correspondent Address:	611 Gateway Blvd, Ste 120South San Francisco, CA 94080 United States
Correspondent Contact:	Mr. Jared Seehafer
Correspondent Contact Telephone:	(415) 638-9554
Correspondent Contact Email:	jared@enzyme.com
Device Name
Device Trade Name:	AiMIFY (1.x)
Common Name:	Medical image management and processing system
Classification Name:	System, Image Processing, Radiological
Regulation Number:	892.2050
Product Code:	LLZ
Device Class:	Class II
Legally Marketed PredicateDevice:	Predicate #: K223623Predicate Trade Name: SubtleMRPredicate Manufacturer: Subtle Medical

Table 1. Contact Details & Device Name

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

Device Description Summary

The AiMIFY image is created by AiMIFY and sent back to the picture archiving and communication system (PACS) or other DICOM node by the compatible MDDS for clinical review.

Because the software runs in the background, it has no user interface. It is intended to be used by radiologists in an imaging center, clinic, or hospital.

Intended Use / Indications for Use

Intended Use / Indications for Use Comparison

AiMIFY and SubtleMR are both intended to enhance 2D/3D T1 imaging data from MRI imaging systems.

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

Technological Comparison

The following table shows the similarities and differences between the technological characteristics of the predicate and subject devices.

Topic	Predicate Device	Subject Device
PhysicalCharacteristics	Software package that operates on off-the-shelfhardware	Same
DICOMStandardCompliance	The software processes DICOM compliantimage data	Same
OperatingSystem	Linux	Same
Modalities	MRI	Same
User Interface	None - enhanced images are viewed onexisting PACS workstations	Same
Workflow	The software operates on DICOM files on thefile system, enhances the images, and storesthe enhanced images on the file system. Thereceipt of original DICOM image files anddelivery of enhanced images as DICOM filesdepends on other software systems. Enhancedimages co-exist with the original images.	Same
Algorithm	SubtleMR implements an image enhancementalgorithm using convolutional neural networkbased filtering. Original images are enhancedby running through a cascade of filter banks,where thresholding and scaling operations areapplied. Separate neural network based filtersare obtained for noise reduction and sharpnessincrease. The parameters of the filters wereobtained through an image-guidedoptimization process.	AiMIFY implements an imageenhancement algorithm usingconvolutional neural network basedfiltering. The model performs non-linearscaling of the contrast uptake between thepre- and post-contrast images to improvethe visualization of the areas in the brainthat have absorbed the contrast material.
Inputs	SubtleMR processes a single input image (withlow SNR or low resolution) and outputs asingle image with higher SNR or resolution.	AiMIFY processes two input images (T1pre-contrast and T1 post-contrast) andoutputs a single image with a higherrelative CNR.
Processing	SubtleMR processes the input image slice byslice in a 2D fashion using a ResNET styleneural network.	AiMIFY processes the input image in a2.5D fashion using a UNet style neuralnetwork.

Table 2. Summary of Technological Characteristics Comparison

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

Fundamentally, both devices enhance images from input data provided by MR imaging systems. Though there are technological differences, they do not present different questions of safety and effectiveness, and the methods for assessing the safety and effectiveness of the subject device are substantively similar to those of the predicate device.

Predetermined Change Control Plan

AiMIFY has a Predetermined Change Control Plan for planned changes to improve the generalizability of the AiMIFY Artificial Intelligence/Machine Learning (AI/ML) model, to reduce processing time, and improve perceived image quality.

Improved generalizability of the AiMIFY AI/ML model is expected to affect the labeled limitations of AiMIFY, as the model will be more generalizable than the cleared version of the device. Improvements planned include additional training and/or performance data to add more gadolinium-based contrast agents, add additional patient ages (infants and children), add additional clinical conditions, and add additional scanner models. The AiMFY AI/ML model is planned to be retrained with additional output channels to reduce processing time, and shall be subject to the same performance testing endpoints and acceptance criteria as the cleared version of the device.

Additionally, an optional pre-processing feature is planned that consists of an algorithm which will suppress vessel enhancement of the AiMIFY AI/ML model to reduce vessel conspicuity in the AiMIFY-enhanced images and resulting in improved perceived image quality. The vessel suppression algorithm shall be subject to the same performance testing performance testing endpoints and acceptance criteria as the cleared version of the device, along with the additional endpoints that lesions in AiMIFY-enhance images with vessel suppression should have equivalent CNR, LBR, CEP improvements than images without suppression and that the CNR, LBR, CEP around small and large blood vessels are lower in AiMIFY-enhance images with vessel suppression than without. The vessel suppression algorithm shall be subject to the same performance testing dataset used in the cleared version of the device, along with additional data containing lesions that appear like blood vessels to demonstrate that the vessel suppression algorithm doesn't incorrectly suppress enhancement of the lesion in this edge case situation.

All changes shall be performed as a part of Subtle Medical's Design Change Control process in a Quality Management System that is ISO 13485:2016 and MDSAP certified. All impacted design documentation, testing documentation, and labeling shall be updated appropriately for each change. None of the changes shall result in a change to the Indications for Use, Cautions, performance endpoints, or acceptance criteria of AiMIFY. Each change to AiMIFY shall be accompanied by a software version-specific Customer Release Notes describing the change along with software version-specific User Manual (with updated Limitations description if applicable) and other updated labeling.

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

Non-Clinical and/or Clinical Tests Summary & Conclusions Dataset Characteristics:

The dataset used for the quantitative and qualitative tests consists of 95 T1 brain cases with a variety of protocols, contrast agent, MRI scanners, and patient demographics.

The images varied in acquisition including:

T1 input protocols: BRAVO, MPRAGE+, FLAIR, FSE. ●
· Orientation: axial, sagittal, coronal
. Acquisition type: 2D, 3D

The images were acquired from various scanners:

Field strength: 0.3T, 1.5T, 3.0T ●
Vendors: GE, Philips, Siemens, Hitachi

The images were sourced from a variety of patients:

. Age: Relatively even distribution from age 7 to 86.
. Sex: Relative event distribution of females and males.
· Pathologies: Cerebritis, Glioma, Meningioma, Metastases, Multiple Sclerosis, Neuritis, Inflammation, Other tumor related (resection / radiation necrosis / suspected cyst / etc. ), other abnormalities
. Population Demographics*: California, USA; New York, USA; Nationwide, USA; Beijing, China

*Due to privacy practices, ethnicities of the training and testing data is unknown but can be inferred based on the population demographics from where they were sourced. The population demographics of data sources used for training and testing are as follows per the Census.gov July 2022 Estimates. In California, USA, the population of 39m people are distributed by race and hispanic origin as: White alone 70.7%, Black or African American alone 6.5%, American Indian and Alaskan Native alone 1.7%, Asian alone 16.3%, Native Hawaiian and Other Pacific Islander alone, percent 0.5%, Two or More Race 4.3%, Hispanic or Latino 40.3%, and White alone (not Hispanic or Latino) 34.7%. In New York, USA, the population of 20m people are distributed by race and hispanic origin as: White alone 68.6%, Black or African American alone 17.7%. American Indian and Alaskan Native alone 1%. Asian alone 9.6%. Native Hawaiian and Other Pacific Islander alone, percent 0.1%. Two or More Race 2.8%. Hispanic or Latino 19.7%, and White alone (not Hispanic or Latino) 54.2%. In the overall USA, the population of 333m people are distributed by race and hispanic origin as: White alone 75.5%, Black or African American alone 13.6%, American Indian and Alaskan Native alone 1.3%, Asian alone 6.3%, Native Hawaiian and Other Pacific Islander alone, percent 0.3%, Two or More Race 3%, Hispanic or Latino 19.1%, White alone (not Hispanic or Latino) 58.9%.

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

Quantitative Assessment (Bench Test):

In addition to software verification & validation, performance bench testing was conducted to demonstrate the safety and efficacy of applying AiMIFY to enhance contrasted images using a convolutional neural network on clinically acquired pre- and post-contrast brain images. To evaluate the performance of the software, Subtle quantitatively evaluated the 95 acquired studies recruited from clinical sites or hospitals. Of these cases, it was found during the Reader Study that 57 of them had lesions. The acquired studies include pre-contrast, and AiMIFY images from various sites, age ranges, manufacturers, scanner models, orientations, and magnitudes of field strengths. They were used to assess the superiority of the AiMIFY against the acquired post-contrast image using Contrast Enhancement Percentage (CEP), Contrast-to-Noise Ratio (CNR), and Lesion-to-Brain Ratio (LBR) to perform the analysis.

The primary endpoint of the quantitative test was that the CNR, LBR, and CEP of a selected region of interest (ROI) in each test dataset is on average improved by greater than or equal to 50% after AiMIFY enhancement compared to the traditionally acquired contrast images.

Two ROIs were drawn by a board-certified radiologist for each case to capture the enhancing lesion and brain parenchyma. An ROI was drawn at the superior sagittal sinus region without an enhancing lesion. The post-contrast image was labeled where the ROI was copied to the AiMIFY image with relevant metrics on the case. The ROI is designed to estimate the signal intensity value.

Contrast-to-Noise Ratio (CNR) was determined to be 559.94% across all 95 cases, and 831.70% for the 57 lesion-only cases. The CNR of the AiMIFY images is significantly higher (Wilcoxon signed-rank test, p < 0.0001) than that of the standard post-contrast images. The high increase of CNR of AiMIFY images compared to standard post-contrast images is due to the fact that both the contrast between a lesion and the normal parenchyma is increased by AiMIFY and also the images is reduced by the denoising module of AiMIFY.

Lesion-to-Brain Ratio (LBR) was determined to be 62.07% across all 95 cases, and 58.80% for the 57 lesion-only cases. The Contrast Enhancement Percentage (CEP) between the acquired post-contrast image and AiMIFY was determined to be 133.29% across all 95 cases, and 101.80% for the 57 lesion only cases. The LBR and the CEP metric of the AiMIFY images are both significantly better (Wilcoxon signed-rank test, p-value < 0.0001) than that of the standard post-contrast images. Both metrics are measuring the main effects of AiMIFY, which is the increase of signal intensity of enhancing tissue, measured as CEP and leading to higher LBR.

In summary, these results prove that the AiMIFY images are enhanced with the acquired post-contrast in terms of the signal intensity value from the brain parenchyma to the enhanced lesion.

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

Qualitative Assessment (Reader Study):

The AiMIFY Reader Study aimed to demonstrate image enhancement of contrasted MRI images by applying AiMIFY post-processing enhancement software using convolutional neural networks. Specifically, 3 board-certified neuro-radiologists evaluated the Perceived Visibility of anatomic features and lesions. The primary diagnostic categories were (1) Meningioma, (2) Glioma, (4) Metastases, (5) Infection, (6) Cerebritis, (7) Multiple Sclerosis, (8) Neuritis, (9) Other tumor-related (resection / radiation necrosis / suspected cyst / etc.), and (10) Other abnormality. The readers were presented with both the pre- and post-contrast input images and the AiMIFY enhanced image. Readers were asked to evaluate the Perceived Visibility of Lesion Contrast Enhancement, Lesion Border Delineation, and Lesion Internal Morphology on a 4-point Likert scale. Additionally, readers were asked to evaluate Perceived Image Quality, Artifact Presence, and Vessel Conspicuity on a 4-point Likert scale.

The primary endpoint of the qualitative test was the visibility of lesion features (lesion contrast enhancement, border delineation, and internal morphology) for cases with identified lesions in the test datasets should be statistically significantly better for AiMIFY processed images per the Wilcoxon signed rank test by p<0.05.

Lesion Contrast Enhancement, Border Delineation, and Internal Morphology were significantly better than standard post-contrast by p<0.0001. Therefore, results provide evidence that AiMIFY provides a significant improvement of lesion contrast enhancement, border delineation, and internal morphology visualization.

The secondary endpoint of the qualitative test was the visibility of perceived image quality and image artifacts for all cases in the test datasets for AiMIFY processed images should NOT be statistically significantly worse than standard post-contrast images per the Wilcoxon signed rank test by p<0.05.

Perceived Image Quality and Artifact Presence And Impact On Clinical Diagnosis were significantly not worse than standard post-contrast by p<0.0001. Additionally, two of three readers demonstrated that Perceived Image Quality is better than standard post-contrast by p<0.0001. Therefore, these results provide evidence that AiMIFY does not have worse perceived image quality or artifacts that impact diagnosis, and may have higher perceived image quality.

In addition to the primary and secondary endpoints, the study contained several exploratory endpoints to characterize AiMIFY's performance.

Exploratory Endpoints:

Vessel Conspicuity

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

The first exploratory endpoint was the evaluation of user preference of vessel conspicuity for all cases in the test datasets for AiMIFY processed images to establish the limitations of AiMIFY.

Compared to 3.51% of standard post-contrast cases, 11.58% of AiMIFY cases were rated as likely to impact diagnosis due to vessel conspicuity. However, in 90% of cases vessel conspicuity was assessed as unlikely to impact or having no impact on diagnosis. This shows that the potential negative impact of the higher vessel conspicuity with AiMIFY is limited. Additionally, a further breakdown of results by reader revealed that there was significant variability in reader preference regarding vessel conspicuity.

False Lesion Analysis

The second exploratory endpoint was the evaluation of false lesion frequency, cause, and severity for all cases in the test datasets for AiMIFY processed images to establish a caution statement for AiMIFY.

73% of cases did not have false lesions, 4% of cases with false lesions did not have impact on diagnosis, 9% of cases with false lesions had minor impact on diagnosis, 14% of cases with false lesions had major impact on diagnosis (4% being due to false lesions that are introduced by AiMIFY, 2% due to quality of input images, and 8% being due to a combination of both causes). 100% of cases received scores from all readers that the Standard-of-Care image was sufficient to identify the false lesion(s).

Lesion Size

To establish the range of lesions that AiMIFY is validated to be used on the size of lesions in the 95 test subject validation datasets used for AiMIFY quantitative and qualitative assessment was performed.

AiMIFY has been validated with lesion sizes ranging from 1.23 cc to 7.08 cc with 95% CI, and has been tested on lesions as small as 0.02 cc and as large as 59.83 cc. AiMIFY processed images had greater lesion contrast enhancement than standard-of-care post-contrast images for both small lesions (<1cc AiMIFY = 3.57 versus Original = 2.71) and large lesions (>1cc: AiMIFY = 3.77 versus Original = 3.23). This establishes the range of lesion sizes over which AiMIFY performance has been validated, and that enhancement is seen in both small (<1cc) and large (>1cc) lesions.

Radiomics

Radiomics analysis was performed to assess the similarity in the texture related features of the lesions as well as parenchyma tissue / healthy tissue, using the concordance correlation coefficient (CCC) as the metric, on 95 subject datasets (the same data from the AiMIFY Quantitative Assessment described above).

{12}------------------------------------------------

The similarity in the texture measured as Concordance Correlation Coefficient (CCC) between AiMIFY input and output images in seven radiomics feature classes were all >= 0.65 for lesion tissue. Specifically, the CCC in the seven radiomics feature classes ranged from 0.68 to 0.89, which are all >= 0.65, for lesion tissue. Seven radiomics feature classes of SOC and AiMIFY-enhanced have a good correlation for lesion tissue.

The similarity in the texture measured as CCC between AiMIFY input and output images in seven radiomics feature classes were all >= 0.8 for parenchyma tissue / healthy tissue. Specifically, the CCC in the seven radiomics feature classes ranged from 0.82 to 0.92, which are all >= 0.8, for parenchyma tissue / healthy tissue. Seven radiomics feature classes of SOC and AiMIFY-enhanced have a good correlation for parenchyma tissue / healthy tissue.

Dose Appearance

To assess whether AiMIFY-enhanced images have a high-dose-like appearance, the 'dose concentration' of AiMIFY-enhanced images to standard post-contrast images was compared using a physics-based formula. "BRAVO" sequence images were chosen for this evaluation due to the widely accepted signal evolution equation which is well established in the literature. On average, AiMIFY-enhanced images had 93% more dose concentration than standard post-contrast images, indicating they do have a high-dose-like appearance.

In addition, the training and validation datasets were compared in terms of CNR increase, where the training dataset compared CNR increase from low-dose to regular-dose post-contrast images while the testing dataset compared CNR increase from standard regular-dose post-contrast images and the AiMIFY-enhanced post-contrast images. Both datasets had significant increases in CNR increase.

SubtleMR

The AiMIFY device makes use of the SubtleMR device (K223623) as a denoising module in its post-processing engine. As the SubtleMR device was not previously tested or trained on AiMIFY-processed images, validation of SubtleMR using the validation tests described in K223623 was conducted to ensure performance of that device on this new type of input data.

When using SubtleMR in AiMIFY post-processing, average scores between the original and SubtleMR enhanced images are <= 0.5 Likert scale points for visibility of small structures. Specifically, the average score between original and SubtleMR enhanced images was 0.05 points. These results demonstrate that SubtleMR's denoising does not over-smooth images processed by the AiMIFY algorithm.

When using SubtleMR in AiMIFY post-processing, average scores difference between the original and SubtleMR enhanced images are <= 0.5 Likert scale points for perceived SNR, Image Quality, Image

{13}------------------------------------------------

Artifact Scores. Specifically, the average SNR score difference between original and SubtleMR enhanced images was 0.05 for Septum Pellucidum, 0.08 for Cranial Nerves, and 0.07 for Cerebellar Folia. The average image quality/diagnostic confidence score difference between original and SubtleMR enhanced images was 0.11 for Septum Pellucidum, 0.04 for Cranial Nerves, and -0.05 for Cerebellar Folia. The average imaging artifacts score difference between original and SubtleMR enhanced images was 0.11 for Septum Pellucidum, 0.14 for Cranial Nerves, and 0.05 for Cerebellar Folia.

These results demonstrate that SubtleMR's denoising does not negatively impact perceived SNR, perceived image quality/diagnostic confidence, or imaging artifacts in images processed by the AiMIFY algorithm.

When evaluating for SNR improvement, the average SNR improvement was 14%, greater than the 5% acceptance criteria established in the SubtleMR validation, indicating that SubtleMR can denoise images processed by the AiMIFY algorithm.

Conclusion:

These results demonstrate that the AiMIFY is substantially equivalent to the predicate device.

Regulation Number and Section

§ 892.2050 Medical image management and processing system.

(a)
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.(b)
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).