Search Results

Magnetic Resonance Imaging (MRI) is a noninvasive technique used for diagnostic imaging. MRI with its soft tissue contrast capability enables the healthcare professional to differentiate between various soft tissues, for example, fat, water, and muscle, but can also visualize bone structures.

Depending on the region of interest, contrast agents may be used.

The MR system may also be used for imaging during interventional procedures and radiation therapy planning.

The PET images and measures the distribution of PET radiopharmaceuticals in humans to aid the physician in determining various metabolic (molecular) and physiologic functions within the human body for evaluation of diseases and disorders such as, but not limited to, cardiovascular disease, neurological disorders, and cancer.

The integrated system utilizes the MRI for radiation-free attenuation correction maps for PET studies. The integrated system provides inherent anatomical reference for the fused MR and PET images due to precisely aligned MR and PET image coordinate systems.

BIOGRAPH One with software Syngo MR XB10 includes new and modified hardware and software compared to the predicate device, Biograph mMR with software syngo MR E11P-AP01. A high level summary of the new and modified hardware and software is provided below:

Hardware

New Hardware

• Gantry offset phantom
• SDB (Smart Distribution Box)

New Coils

• BM Contour XL Coil
• BM Head/Neck Pro PET-MR Coil
• BM Spine Pro PET-MR Coil
• Transfer of up-to-date RF coils from the reference device MAGNETOM Vida.

Modified Hardware

• Main components such as:
  • Detector cassettes / DEA
  • Phantom holder
  • Gantry tube
  • Backplane
  • Magnet and cabling
  • Gradient coil
  • MaRS (measurement and reconstruction system)
  • MI MARS
  • PET electronics
  • RF transmitter TBX3 3T (TX Box 3)
• Other components such as:
  • Cover
  • Filter plate
  • Patient table
  • RFCEL_TEMP

Modified Coils

• Body coil
• Transfer of up-to-date RF coils from the reference device MAGNETOM Vida with some improvements.

Software

New Features and Applications

• Fast Whole-Body workflows
• Fast Head workflow
• myExam PET-MR Assist
• CS-Vibe
• myExam Implant Suite
• DANTE blood suppression
• SMS Averaging for TSE
• SMS Averaging for TSE_DIXON
• SMS without diffusion function
• BioMatrix Motion Sensor
• RF pulse optimization with VERSE
• Deep Resolve Boost for FL3D_VIBE and SPACE
• Deep Resolve Sharp for FL3D_VIBE and SPACE
• Preview functionality for Deep Resolve Boost
• EP2D_FID_PHS
• EP_SEG_FID_PHS
• ASNR recommended protocols for imaging of ARIA
• Open Workflow
• Ultra HD-PET
• "MTC Mode"
• OpenRecon 2.0
• Deep Resolve Boost for TSE
• GRE_PC
• The following functions have been migrated for the subject device without modifications from MAGNETOM Skyra Fit and MAGNETOM Sola Fit:
  • 3D Whole Heart
• Ghost reduction (Dual polarity Grappa (DPG))
• Fleet Reference Scan
• AutoMate Cardiac (Cardiac AI Scan Companion)
• Complex Averaging
• SPACE Improvement: high bandwidth IR pulse
• SPACE Improvement: increase gradient spoiling
• The following function has been migrated for the subject device without modifications from MAGNETOM Free.Max:
  • myExam Autopilot Spine
• The following functions have been migrated for the subject device without modifications from MAGNETOM Sola:
  • myExam Autopilot Brain
  • myExam Autopilot Knee
• Transfer of further up-to-date SW functions from the reference devices.

New Software / Platform

• PET-Compatible Coil Setup
• Select&GO
• PET-MR components communication

Modified Features and Applications

• HASTE_CT
• FL3D_VIBE_AC
• PET Reconstruction
• Transfer of further up-to-date SW functions from the reference devices with some improvements.

Modified Software / Platform

• Several software functions have been improved. Which are:
  • PET Group
  • PET Viewing
  • PET RetroRecon
  • PET Status and Tune-up/QA

Other Modifications and / or Minor Changes

• Indications for use
• Contraindications
• SAR parameter
• Off-Center Planning Support
• Flip Angle Optimization (Lock TR and FA)
• Inline Image Filter
• Marketing bundle "myExam Companion"
• ID Gain
• Automatic System Shutdown (ASS) sensor (Smoke Detector)
• Patient data display (PDD)

The FDA 510(k) Clearance Letter for BIOGRAPH One refers to several AI/Deep Learning features. However, the provided document does not contain explicit acceptance criteria for these AI features in a table format, nor does it detail a comparative effectiveness study (MRMC study) for human readers. It primarily focuses on demonstrating non-inferiority to the predicate device through various non-clinical tests.

Below is an attempt to extract and synthesize the information based on the provided text, while acknowledging gaps in the information regarding specific acceptance criteria metrics and clinical studies.

Acceptance Criteria and Study Details for BIOGRAPH One AI Features

1. Table of Acceptance Criteria and Reported Device Performance

The document does not explicitly state numerical acceptance criteria in a dedicated table format. Instead, it describes performance in terms of achieving "convergence of the training" and "improvements compared to conventional parallel imaging," or confirming "very similar metrics" to the predicate. The "acceptance criteria" are implied by these statements and the successful completion of the described tests.

2. Sample Sizes Used for Test Set and Data Provenance

The document primarily describes a validation dataset which serves as the "test set" for the AI models during development, and an additional "test dataset" for specific evaluations.

• Deep Resolve Boost for FL3D_VIBE and SPACE:

  • Test Set Description: The "collaboration partners (testing)" data is mentioned as the source for testing, implying an external, independent test set. No specific number for this test set is provided beyond the 1265 measurements for training/validation.
  • Sample Size (Validation/Training): 27,679 3D patches from 1265 measurements.
  • Data Provenance: "in-house measurements (training and validation) and collaboration partners (testing)." The country of origin is not specified but is likely Germany (Siemens Healthineers AG) and/or China (Siemens Shenzhen Magnetic Resonance LTD.) where the manufacturing is listed.
  • Retrospective/Prospective: "Input data was retrospectively created from the ground truth by data manipulation and augmentation." This indicates retrospective data use.

• Deep Resolve Sharp for FL3D_VIBE and SPACE:

  • Test Set Description: The document states, "The high-resolution datasets were split to 70% training and 30% validation datasets before training to ensure independence of them." This implies the 30% validation dataset is used as the test set.
  • Sample Size (Validation/Training): 27,679 3D patches from 1265 measurements (split into 70% training and 30% validation).
  • Data Provenance: "in-house measurements (training and validation) and collaboration partners (testing)."
  • Retrospective/Prospective: "Input data was retrospectively created from the ground truth by data manipulation." This indicates retrospective data use.

• Deep Resolve Boost for TSE (First Mention - General Performance):

  • Test Set Description: The "evaluation on the test dataset" is mentioned. The validation set is 30% of the 500 measurements.
  • Sample Size (Validation/Training): Approximately 13,000 high resolution 3D patches from 500 measurements (split into 70% training and 30% validation).
  • Data Provenance: "in-house measurements."
  • Retrospective/Prospective: "Input data was retrospectively created from the ground truth by data manipulation." This indicates retrospective data use.

• Deep Resolve Boost for TSE (Second Mention - Banding Artifacts):

  • Test Set Description: "Additional test dataset for banding artifact reduction: more than 2000 slices." This dataset was acquired after the release of the predicate network.
  • Sample Size: More than 2000 slices.
  • Data Provenance: "in-house measurements and collaboration partners."
  • Retrospective/Prospective: Not explicitly stated for this specific additional dataset, but the training/validation data for the predicate was retrospective.

3. Number of Experts and Qualifications for Ground Truth

• Radiologist Evaluation for Deep Resolve Boost for TSE (Second Mention): The document mentions "the radiologist evaluation revealed no difference in suitability for clinical diagnostics."

  • Number of Experts: Not specified (singular "radiologist" used, but typically multiple are implied for such evaluations).
  • Qualifications: "Radiologist." No specific years of experience or subspecialty are mentioned.

• Other features: For Deep Resolve Boost/Sharp for FL3D_VIBE and SPACE, and Deep Resolve Boost for TSE (first mention), the ground truth is derived directly from acquired image data (see section 7). No independent human expert ground truth establishment for these.

4. Adjudication Method (for Test Set)

• Radiologist Evaluation for Deep Resolve Boost for TSE (Second Mention): The adjudication method is not specified in the document (e.g., 2+1, 3+1). It only states "the radiologist evaluation."

• Other features: Adjudication methods are not applicable as human experts were not establishing ground truth for objective metrics.

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

• Was an MRMC study done? No, the document does not describe an MRMC comparative effectiveness study where human readers' performance with and without AI assistance is compared. The evaluation for Deep Resolve Boost for TSE mentions "radiologist evaluation" but not in a comparative MRMC study context.
• Effect Size: Not applicable, as no MRMC study was performed.

6. Standalone (Algorithm Only) Performance

• Was standalone performance done? Yes, the performance testing for all Deep Resolve features (Boost and Sharp for FL3D_VIBE, SPACE, and TSE) was conducted "algorithm only" by evaluating metrics like PSNR, SSIM, MSE, and LPIPS, and then visual inspection/radiologist evaluation. These refer to the algorithm's direct output performance.

7. Type of Ground Truth Used

• Deep Resolve Boost for FL3D_VIBE and SPACE: "The acquired datasets (as described above) represent the ground truth for the training and validation."
• Deep Resolve Sharp for FL3D_VIBE and SPACE: "The acquired datasets represent the ground truth for the training and validation." Input data was manipulated (cropped k-space) to create low-resolution input and high-resolution output/ground truth from the same dataset.
• Deep Resolve Boost for TSE (First Mention): "The acquired datasets represent the ground truth for the training and validation." Input data was manipulated (cropped k-space) to create low-resolution input and high-resolution output/ground truth from the same dataset.
• Deep Resolve Boost for TSE (Second Mention): "The acquired training/validation datasets... represent the ground truth for the training and validation." Input data was manipulated by undersampling k-space, adding noise, and mirroring k-space.
• Summary: The ground truth for all AI features was derived from acquired, high-resolution original image data (retrospectively manipulated to simulate inputs). For Deep Resolve Boost for TSE (second mention), there was also an implicit "expert consensus" or "expert reading" component for the "radiologist evaluation" regarding clinical suitability.

8. Sample Size for the Training Set

• Deep Resolve Boost for FL3D_VIBE and SPACE: 81% of 1265 measurements (for 27,679 3D patches).
• Deep Resolve Sharp for FL3D_VIBE and SPACE: 70% of 1265 measurements (for 27,679 3D patches).
• Deep Resolve Boost for TSE (First Mention): 70% of 500 measurements (for approx. 13,000 high resolution 3D patches).
• Deep Resolve Boost for TSE (Second Mention): More than 23,250 slices (93% of the combined training/validation dataset from K213693).

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

• Deep Resolve Boost for FL3D_VIBE and SPACE: The "acquired datasets" represent the ground truth. "Input data was retrospectively created from the ground truth by data manipulation and augmentation. This process includes further undersampling of the data by discarding k-space lines as well as creating sub-volumes of the acquired data."
• Deep Resolve Sharp for FL3D_VIBE and SPACE: The "acquired datasets represent the ground truth." "Input data was retrospectively created from the ground truth by data manipulation. k-space data has been cropped such that only the center part of the data was used as input. With this method corresponding low-resolution data as input and high-resolution data as output / ground truth were created for training and validation."
• Deep Resolve Boost for TSE (First Mention): Similar to Deep Resolve Sharp for FL3D_VIBE and SPACE: "The acquired datasets represent the ground truth for the training and validation. Input data was retrospectively created from the ground truth by data manipulation. k-space data has been cropped such that only the center part of the data was used as input. With this method corresponding low-resolution data as input and high-resolution data as output / ground truth were created for training and validation."
• Deep Resolve Boost for TSE (Second Mention): "The acquired training/validation datasets... represent the ground truth for the training and validation. Input data was retrospectively created from the ground truth by data manipulation and augmentation. This process includes further undersampling of the data by discarding k-space lines, lowering of the SNR level by addition of noise and mirroring of k-space data."

In summary, for all AI features, the ground truth for training was established by using high-quality, originally acquired MRI data that was then retrospectively manipulated (e.g., undersampled, cropped, noise added) to create synthetic "lower quality" input data for the AI model to learn from, with the original high-quality data serving as the target output or ground truth.

Ask a specific question about this device

The MAGNETOM system is indicated for use as a magnetic resonance diagnostic device (MRDD) that produces transverse, sagittal, coronal and oblique cross-sectional images, spectroscopic images and/or spectra, and that displays, depending on optional local coils that have been configured with the system, the internal structure and/or function of the head, body, or extremities. Other physical parameters derived from the images and/or spectra may also be produced. Depending on the region of interest, contrast agents may be used. These images and/or spectra and the physical parameters derived from the images and/or spectra when interpreted by a trained physician yield information that may assist in diagnosis.

The MAGNETOM system may also be used for imaging during interventional procedures when performed with MR compatible devices such as in-room displays and MR Safe biopsy needles.

MAGNETOM Flow.Ace and MAGNETOM Flow.Plus are 60cm-bore MRI systems with quench pipe-free, sealed magnets utilizing DryCool technology. They are equipped with BioMatrix technology and run on Siemens' syngo MR XA70A software platform. The systems include Eco Power Mode for reduced energy and helium consumption. They have different gradient configurations suitable for all body regions, with stronger configurations supporting advanced cardiac imaging. Compared to the predicate device, new hardware includes a new magnet, gradient coil, RF system, local coils, patient tables, and computer systems. New software features include AutoMate Cardiac, Quick Protocols, BLADE with SMS acceleration for non-diffusion imaging, Deep Resolve Swift Brain, Fast GRE Reference Scan, Ghost reduction, Fleet Reference Scan, SMS Averaging, Select&GO extension, myExam Spine Autopilot, and New Startup-Timer. Modified features include improvements for Pulse Sequence Type SPACE, improved Gradient ECO Mode Settings, and Inline Image Filter switchable for users.

The provided 510(k) clearance letter and summary describe the acceptance criteria and supporting studies for the MAGNETOM Flow.Ace and MAGNETOM Flow.Plus devices, particularly focusing on their AI features: Deep Resolve Boost, Deep Resolve Sharp, and Deep Resolve Swift Brain.

Here's a breakdown of the requested information:

1. Table of Acceptance Criteria and Reported Device Performance

The document uses quality metrics like PSNR, SSIM, and NMSE as indicators of performance and implicitly as acceptance criteria. Visual inspection and clinical evaluations are also mentioned.

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

The document uses "Training and Validation data" and often refers to the datasets used for both. It is not explicitly stated what percentage or how many cases from these datasets were strictly reserved for a separate "test set" and what came from the "validation sets." However, given the separation in slice count, the "Validation" slices for Deep Resolve Swift Brain might be considered the test set.

• Deep Resolve Boost:
  • TSE: >25,000 slices
  • HASTE: >10,000 HASTE slices (refined)
  • EPI Diffusion: >1,000,000 slices
  • Data Provenance: Retrospectively created from acquired datasets. Data covered a broad range of body parts, contrasts, fat suppression techniques, orientations, and field strength.
• Deep Resolve Sharp:
  • Data: >10,000 high resolution 2D images
  • Data Provenance: Retrospectively created from acquired datasets. Data covered a broad range of body parts, contrasts, fat suppression techniques, orientations, and field strength.
• Deep Resolve Swift Brain:
  • 1.5T Validation: 3,616 slices (This functions as a test set for 1.5T)
  • 3T Validation: 6,048 slices (This functions as a test set for 3T)
  • Data Provenance: Retrospectively created from acquired datasets.

The document does not explicitly state the country of origin for the data.

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

The document does not explicitly state the number or qualifications of experts used to establish the ground truth for the test sets. For Deep Resolve Swift Brain, it mentions "evaluated in clinical settings with collaboration partners," implying clinical experts were involved in the evaluation, but details are not provided. For Boost and Sharp, the "acquired datasets...represent the ground truth," suggesting the raw imaging data itself, rather than expert annotations on that data, served as ground truth.

4. Adjudication Method for the Test Set

The document does not describe a formal adjudication method (e.g., 2+1, 3+1). For Deep Resolve Swift Brain, it mentions "visually inspected" and "evaluated in clinical settings with collaboration partners," suggesting a qualitative assessment, but details on consensus or adjudication are missing.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study was Done, If so, what was the effect size of how much human readers improve with AI vs without AI assistance

A formal MRMC comparative effectiveness study demonstrating human reader improvement with AI vs. without AI assistance is not described in the provided text. The studies focus on the AI's standalone performance in terms of image quality metrics and internal validation.

6. If a Standalone (i.e. algorithm only without human-in-the-loop performance) was Done

Yes, standalone performance was done for the AI features. The "Test Statistics and Test Results Summary" for Deep Resolve Boost, Deep Resolve Sharp, and Deep Resolve Swift Brain describe the evaluation of the algorithm's output using quantitative metrics (PSNR, SSIM, NMSE) and visual inspection against reference standards, which is characteristic of standalone performance evaluation.

7. The Type of Ground Truth Used

For Deep Resolve Boost, Deep Resolve Sharp, and Deep Resolve Swift Brain, the ground truth used was the acquired high-quality datasets themselves. The input data for training and validation was then retrospectively created from this ground truth by manipulating or augmenting it (e.g., undersampling k-space, adding noise, cropping, using only the center part of k-space). This means the original, higher-quality MR images or k-space data served as the reference for what the AI models should reconstruct or improve upon.

8. The Sample Size for the Training Set

• Deep Resolve Boost:
  • TSE: >25,000 slices
  • HASTE: pre-trained on the TSE dataset and refined with >10,000 HASTE slices
  • EPI Diffusion: >1,000,000 slices
• Deep Resolve Sharp: >10,000 high resolution 2D images
• Deep Resolve Swift Brain: 20,076 slices

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

For Deep Resolve Boost, Deep Resolve Sharp, and Deep Resolve Swift Brain, the "acquired datasets (as described above) represent the ground truth for the training and validation." This implies that high-quality, fully acquired MRI data was considered the ground truth. The input data used during training (e.g., undersampled, noisy, or lower-resolution versions) was then derived or manipulated from this original ground truth. Essentially, the "ground truth" was the optimal, full-data acquisition before any degradation was simulated for the AI's input.

Ask a specific question about this device

The MAGNETOM system is indicated for use as a magnetic resonance diagnostic device (MRDD) that produces transverse, sagittal, coronal and oblique cross sectional images, spectroscopic images and/or spectra, and that displays the internal structure and/or function of the head, body, or extremities. Other physical parameters derived from the images and/or spectra may also be produced. Depending on the region of interest, contrast agents may be used. These images and/or spectra and the physical parameters derived from the images and/or spectra when interpreted by a trained physician yield information that may assist in diagnosis.

The MAGNETOM system may also be used for imaging during interventional procedures when performed with MR compatible devices such as in-room displays and MR Safe biopsy needles.

The subject device, MAGNETOM Avanto Fit with software syngo MR XA70A, consists of new and modified software and hardware that is similar to what is currently offered on the predicate device, MAGNETOM Avanto Fit with syngo MR XA50A (K220151).

A high-level summary of the new and modified hardware and software is provided below:

For MAGNETOM Avanto Fit with syngo MR XA70:

Hardware

New Hardware:
myExam 3D Camera
BM Head/Neck 20

Modified Hardware:
Sanaflex (cushions for patient positioning)

Software

New Features and Applications:
myExam Autopilot Brain
myExam Autopilot Knee
3D Whole Heart
HASTE_interactive
GRE_PC
Open Recon
Deep Resolve Gain
Fleet Reference Scan
Physio logging
complex averaging
AutoMate Cardiac
Ghost Reduction
BLADE diffusion
Beat Sensor
Deep Resolve Sharp
Deep Resolve Boost and Deep Resolve Boost (TSE)
Deep Resolve Boost HASTE
Deep Resolve Boost EPI Diffusion

Modified Features and Applications:
SPACE improvement (high band)
SPACE improvement (incr grad)
Brain Assist
Eco power mode
myExam Angio Advanced Assist (Test Bolus)

The subject device, MAGNETOM Skyra Fit with software syngo MR XA70A, consists of new and modified software and hardware that is similar to what is currently offered on the predicate device, MAGNETOM Skyra Fit with syngo MR XA50A (K220589).

A high-level summary of the new and modified hardware and software is provided below:

For MAGNETOM Skyra Fit with syngo MR XA70:

Hardware

New Hardware:
myExam 3D Camera

Modified Hardware:
Sanaflex (cushions for patient positioning)

Software

New Features and Applications:
Beat Sensor
HASTE_interactive
GRE_PC
3D Whole Heart
Deep Resolve Gain
Open Recon
Ghost Reduction
Fleet Reference Scan
BLADE diffusion
HASTE diffusion
Physio logging
complex averaging
Deep Resolve Swift Brain
Deep Resolve Sharp
Deep Resolve Boost and Deep Resolve Boost (TSE)
Deep Resolve Boost HASTE
Deep Resolve Boost EPI Diffusion
AutoMate Cardiac
SVS_EDIT

Modified Features and Applications:
SPACE improvement (high band)
SPACE improvement (incr grad)
Brain Assist
Eco power mode
myExam Angio Advanced Assist (Test Bolus)

The subject device, MAGNETOM Sola Fit with software syngo MR XA70A, consists of new and modified software and hardware that is similar to what is currently offered on the predicate device, MAGNETOM Sola Fit with syngo MR XA51A (K221733).

A high-level summary of the new and modified hardware and software is provided below:

For MAGNETOM Sola Fit with syngo MR XA70:

Hardware

New Hardware:
myExam 3D Camera

Modified Hardware:
Sanaflex (cushions for patient positioning)

Software

New Features and Applications:
GRE_PC
3D Whole Heart
Ghost Reduction
Fleet Reference Scan
BLADE diffusion
Physio logging
Open Recon
Complex averaging
Deep Resolve Sharp
Deep Resolve Boost and Deep Resolve Boost (TSE)
Deep Resolve Boost HASTE
Deep Resolve Boost EPI Diffusion
AutoMate Cardiac
Implant suite

Modified Features and Applications:
SPACE improvement (high band)
SPACE improvement (incr grad)
Brain Assist
Eco power mode

The subject device, MAGNETOM Viato.Mobile with software syngo MR XA70A, consists of new and modified software and hardware that is similar to what is currently offered on the predicate device, MAGNETOM Viato.Mobile with syngo MR XA51A (K240608).

A high-level summary of the new and modified hardware and software is provided below:

For MAGNETOM Viato.Mobile with syngo MR XA70:

Hardware

New Hardware:
n.a.

Modified Hardware:
Sanaflex (cushions for patient positioning)

Software

New Features and Applications:
GRE_PC
3D Whole Heart
Ghost Reduction
Fleet Reference Scan
BLADE diffusion
Physio logging
Open Recon
Complex averaging
Deep Resolve Sharp
Deep Resolve Boost and Deep Resolve Boost (TSE)
Deep Resolve Boost HASTE
Deep Resolve Boost EPI Diffusion
AutoMate Cardiac
Implant suite

Modified Features and Applications:
SPACE improvement (high band)
SPACE improvement (incr grad)
Brain Assist
Eco power mode

Furthermore, the following minor updates and changes were conducted for the subject devices:

Low SAR Protocol minor update (for all subject devices but MAGNETOM Skyra Fit): the goal of the SAR adaptive protocols was to be able to perform knee, spine, heart and brain examinations with 50% of the max allowed SAR values in normal mode for head and whole-body SAR. The SAR reduction was achieved by parameter adaptations like Flip angle, TR, RF Pulse Type, Turbo Factor, concatenations. For cardiac clinically accepted alternative imaging contrasts are used (submitted with K232494).

Implementation of image sorting prepare for PACS (submitted with K231560).

Implementation of improved DICOM color support (submitted with K232494).

Needle intervention AddIn was added all subject device (submitted with K232494).

Inline Image Filter switchable for users: in the subject device, users have the ability to switch the "Inline image filter" (implicite Filter) on or off. This filter is an image-based filter that can be applied to specific pulse sequence types. The function of the filter remains unchanged from the previous device MAGNETOM Sola with syngo MR XA61A (K232535).

SVS_EDIT is newly added for MAGNETOM Skyra Fit, but without any changes (submitted with K203443)

Brain Assist received an improvement and is identical to that of snygo MR XA61A (K232535)

Open Recon is introduced for all systems. The function of Open Recon remains unchanged from the previous submissions (submitted with K221733).

Lock TR and FA in Bold received a minor UI update

Implant Suite is newly introduced for MAGNETOM Sola Fit and MAGNETOM Viato.Mobile, but without any changes (submitted with K232535)

myExam Autopilot Brain and myExam Autopilot Knee are newly introduced for the subject device MAGNETOM AVANTO Fit and are unchanged from previous submissions (submitted with K221733).

myExam Angio Advanced Assist (Test Bolus) received a bug fixing and minimal UI improvements.

The provided text is an FDA 510(k) clearance letter for various MAGNETOM MRI Systems. While it details new and modified software and hardware features, it does not include specific acceptance criteria or a study that "proves the device meets the acceptance criteria" in terms of performance metrics like sensitivity, specificity, or accuracy for a diagnostic task.

Instead, the document focuses on demonstrating substantial equivalence to predicate devices. This is achieved by:

• Stating that the indications for use are the same.
• Listing numerous predicate and reference devices.
• Detailing hardware and software changes.
• Mentioning non-clinical tests like software verification and validation, sample clinical images, and image quality assessment to show that the new features maintain an "equivalent safety and performance profile" to the predicate devices.
• Referencing scientific publications for certain features to support their underlying principles and utility.
• Briefly describing the training and validation data for two AI features: Deep Resolve Boost and Deep Resolve Sharp, but without performance acceptance criteria or detailed results.

Therefore, much of the requested information cannot be extracted from this document because it is not a study report detailing clinical performance against predefined acceptance criteria for a specific diagnostic outcome.

However, I can extract the information related to the AI features as best as possible from the "AI Features/Applications training and validation" section (Page 16).

Acceptance Criteria and Study Details (Limited to AI Features)

1. Table of Acceptance Criteria and Reported Device Performance

2. Sample size used for the test set and the data provenance

• Deep Resolve Boost:
  • Test Set Sample Size: Not explicitly stated as a separate "test set" size. The document mentions "training and validation data" for over 25,000 TSE slices, over 10,000 HASTE slices (for refinement), and over 1,000,000 EPI Diffusion slices. It's unclear what proportion of this was used specifically for final testing, or if the "validation" mentioned includes the final performance evaluation.
  • Data Provenance: Retrospective, described as "Input data was retrospectively created from the ground truth by data manipulation and augmentation." Country of origin is not specified.
• Deep Resolve Sharp:
  • Test Set Sample Size: Not explicitly stated as a separate "test set" size. The document mentions "training and validation" on more than 10,000 high resolution 2D images. Similar to Deep Resolve Boost, it's unclear what proportion was specifically for final testing.
  • Data Provenance: Retrospective, described as "Input data was retrospectively created from the ground truth by data manipulation." Country of origin is not specified.

3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts

This information is not provided in the document. The definition of "ground truth" for the AI features refers to the acquired datasets themselves rather than expert-labeled annotations. Visual comparisons are mentioned as part of the evaluation, but without details on expert involvement or qualifications.

4. Adjudication method for the test set

This information is not provided in the document. While "visual comparisons" and "visual rating" are mentioned, no specific adjudication method (e.g., 2+1, 3+1) is described.

5. If a multi reader multi case (MRMC) comparative effectiveness study was done, If so, what was the effect size of how much human readers improve with AI vs without AI assistance

No, a MRMC comparative effectiveness study demonstrating human reader improvement with AI assistance is not described in this document. The focus of the AI features (Deep Resolve Boost and Deep Resolve Sharp) is on image quality enhancement (denoising, sharpness) and reconstruction rather than assisting human readers in a diagnostic task that can be quantified by an effect size.

6. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done

Yes, the evaluation of Deep Resolve Boost and Deep Resolve Sharp, based on metrics like PSNR, SSIM, and perceptual loss, and "visual comparisons" or "visual rating" appears to be an assessment of the algorithm's performance in enhancing image quality in a standalone capacity, without direct human-in-the-loop interaction for diagnosis.

7. The type of ground truth used (expert consensus, pathology, outcomes data, etc.)

• Deep Resolve Boost: "The acquired datasets (as described above) represent the ground truth for the training and validation." This implies the original, full-quality, unaltered MRI scan data. Further, "Input data was retrospectively created from the ground truth by data manipulation and augmentation. This process includes further under-sampling of the data by discarding k-space lines, lowering of the SNR level by addition Restricted of noise and mirroring of k-space data."
• Deep Resolve Sharp: "The acquired datasets represent the ground truth for the training and validation." Similar to Boost, this refers to original, high-resolution MRI scan data. For training, "k-space data has been cropped such that only the center part of the data was used as input. With this method corresponding low-resolution data as input and high-resolution data as output / ground truth were created for training and validation."

8. The sample size for the training set

• Deep Resolve Boost:
  • TSE: more than 25,000 slices
  • HASTE (for refinement): more than 10,000 HASTE slices
  • EPI Diffusion: more than 1,000,000 slices
• Deep Resolve Sharp: more than 10,000 high resolution 2D images.

9. How the ground truth for the training set was established

• Deep Resolve Boost: The ground truth was established by the "acquired datasets" themselves (full-quality MRI scans). The training input data was then derived from this ground truth by simulating degraded images (e.g., under-sampling, adding noise).
• Deep Resolve Sharp: Similarly, the ground truth was the "acquired datasets" (high-resolution MRI scans). The training input data was derived by cropping k-space data to create corresponding low-resolution inputs.

Ask a specific question about this device