Search Results

The Longitudinal Brain Imaging (LoBI) is a post-processing application to be used for viewing and evaluating neurological images provided by a magnetic resonance diagnostic device.

The LoBI application is intended for viewing, manipulation and comparison of medical imaging and/or multiple time-points. The LoBI application enables visualization of information that would otherwise have to be visually compared disjointedly. The LoBI application provides analysis tools to help the user assess, and document changes in diagnostic and follow-up examinations. The LoBI application is designed to support the workflow by helping the user to confirm the absence or presence of lesions, including evaluation, follow-up and documentation of any such lesions.

The physician retains the ultimate responsibility for making the final diagnosis and treatment decision.

Philips Medical Systems' Longitudinal Brain Imaging application (LoBI) is a post processing software application intended to assist in the evaluation of serial brain imaging based on MR data.

The LoBI application allows the user to view images, perform segmentation of lesions, along with segmentation editing tool and volumetric quantification of segmented volumes and quantitative comparison between time points. LoBI application provides automatic registration between studies from different time points. for longitudinal comparison.

The LoBI application provides a supportive tool for visualization of subtle differences in the brain of the same individual across time, which can be used by clinicians as the assessment of disease progression.

The physician retains the ultimate responsibility for making the final diagnosis based on image visualization as well as any segmentation and measurement results obtained from the application.

The LoBI application is intended to be used for adult population only

Key Features
LoBI application has the following key features:

• 1. Longitudinal comparison between brain images in multiple studies
• 2. Support for multi-slice MR sequences (2D and 3D) and allow user to use basic viewing operations such as: Scroll, pan, zoom, windowing and annotation
• 3. Identify pre-defined data types (pre-sets) and user created hanging layouts
• 4. Automatic registration between studies (same patient, different time-points)
• 5. Single mode: allows reviewing each of the launched studies, showing multiple sequences of the same study, using the whole reading space
• 6. Tissue segmentation and editing tools allowing volumetric measurement of different lesion types
• 7. Lesion management tool allowing matching between lesions in different studies to facilitate the assessment of differences over time
• 8. CoBI feature (Comparative Brain Imaging) a supportive tool for visualization of subtle differences in lesions of the same individual across time for similar sequences. The CoBI feature provides a mathematical subtraction of scans yielding, after bias-field correction and intensity scaling, a colorcoded image of the differences in intensity between two registered scans.
• 9. Results are displayed in tabular and graphical formats.

Here's a summary of the acceptance criteria and study information for the Philips Longitudinal Brain Imaging (LoBI) application, based on the provided 510(k) summary:

1. Table of Acceptance Criteria and Reported Device Performance:

The document focuses on demonstrating substantial equivalence to predicate devices and adherence to regulatory standards rather than explicit quantitative acceptance criteria or detailed device performance metrics in a table format. The primary "acceptance criteria" are implied by compliance with:

• International and FDA-recognized consensus standards: ISO 14971, IEC 62304, IEC 62366-1, DICOM PS 3.1-3.18.
• FDA guidance document: "Guidance for the Content of Premarket Submissions for Software Contained in Medical Devices."
• Internal Philips verification and validation processes: Ensuring the device "meets the acceptance criteria and is adequate for its intended use and specifications."

Since specific numerical performance criteria (e.g., accuracy, sensitivity, specificity for particular lesion types) and corresponding reported performance are not provided in this 510(k) summary, the table below reflects what is broadly stated.

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

• Test Set Sample Size: Not explicitly stated as a number of cases or images. The validation activities used "real recorded clinical data cases." The quantity of these cases is not specified.
• Data Provenance: The data used for validation consisted of "real recorded clinical data cases." No specific country of origin is mentioned. It is indicated as retrospective, as they are "recorded" data.

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

• This information is not provided in the document. The general statement is that "The physician retains the ultimate responsibility for making the final diagnosis," suggesting human expert involvement in clinical practice, but not explicitly defining how ground truth for the test set was established or by whom.

4. Adjudication Method for the Test Set:

• This information is not provided in the document.

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 MRMC comparative effectiveness study was done or reported. The document states explicitly: "The subject of this premarket submission. Longitudinal Brain Imaging (LoBI) application did not require clinical studies to support equivalence." The testing focused on verification and validation of the software's functionality and compliance with standards.

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

• The document describes the LoBI application as a "post-processing software application intended to assist in the evaluation of serial brain imaging" and emphasizes that "The physician retains the ultimate responsibility for making the final diagnosis."
• While the software performs automated functions like registration, segmentation, and quantitative comparison, the validation process using "real recorded clinical data cases" seems to focus on the software's ability to provide accurate tools and information that a user would interpret.
• The description of "Full functionality test" and "RMF testing" could involve standalone algorithmic performance evaluation against predefined specifications. However, an explicit "standalone" performance study as a separate regulatory study with defined metrics (e.g., algorithm-only sensitivity/specificity against ground truth) is not detailed in this summary. The focus is on the tool's supportive role for the user.

7. The Type of Ground Truth Used (Expert Consensus, Pathology, Outcomes Data, etc.):

• The type of ground truth used for the validation data is not explicitly specified. It refers to "real recorded clinical data cases," implying that the medical imaging data came with existing clinical interpretations or diagnoses, which would have implicitly served as a form of reference or "ground truth" for evaluating the software's utility in "confirming the absence or presence of lesions, including evaluation, quantification, follow-up and documentation." However, the method of establishing this ground truth (e.g., expert consensus, pathology) is not detailed.

8. The Sample Size for the Training Set:

• The document does not provide information regarding a distinct training set sample size or how the LoBI application was developed using machine learning or AI. The product description focuses on its functionality as a post-processing application with features like automatic registration and tissue segmentation, which could be rule-based or machine learning-driven, but this is not specified, nor is training data mentioned.

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

• Since a training set is not mentioned, the method for establishing its ground truth is also not provided.

Ask a specific question about this device

The Lung Nodule Assessment and Comparison Option is intended for use as a diagnostic patient-imaging tool. It is intended for the review and analysis of thoracic CT images, providing quantitative and characterizing information about nodules in the lung in a single study, or over the time course of several thoracic studies. Characterizations include diameter, volume and volume over time. The system automatically performs the measurements, allowing lung nodules and measurements to be displayed.

The Lung Nodule Assessment and Comparison Option application is intended for use as a diagnostic patient-imaging tool. It is intended for the review and analysis of thoracic CT images, providing quantitative and characterizing information about nodules in the lung in a single study, or over the time course of several thoracic studies. The system automatically performs the measurements, allowing lung nodules and measurements to be displayed. The user interface and automated tools help to determine growth patterns and compose comparative reviews. The Lung Nodule Assessment and Comparison Option application requires the user to identify a nodule and to determine the type of nodule in order to use the appropriate characterization tool. Lung Nodule Assessment and Comparison Option may be utilized in both diagnostic and screening evaluations supporting Low Dose CT Lung Cancer Screening*.

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

Device Name: Lung Nodule Assessment and Comparison Option (LNA)

1. Table of Acceptance Criteria and Reported Device Performance

The provided 510(k) summary does not explicitly list quantified acceptance criteria with numerical targets. Instead, it indicates that the device was tested against its defined functional requirements and performance claims, and that it "meets the acceptance criteria and is adequate for its intended use and specifications." The "acceptance criteria" are implied by the verification and validation tests performed to ensure the device's design meets user needs and intended use, and that its technological characteristics claims are met.

However, based on the description of the device's capabilities, we can infer some key performance areas that would have been subject to acceptance criteria:

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

• Test Set Sample Size: The document does not specify a numerical sample size for the internal validation studies conducted by Philips for the LNA application. It states that the LNA application was validated "using real recorded clinical data cases in order to simulate the actual use of the software."
• Data Provenance for Philips' Internal Tests: The text implicitly suggests the data was retrospective, as it refers to "real recorded clinical data cases." The country of origin for these internal test cases is not specified.
• Data Provenance for the Risk Calculator (McWilliams et al. study):
  • Development Data Set: Participants from the Pan-Canadian Early Detection of Lung Cancer Study (PanCan).
  • Validation Data Set: Participants from chemoprevention trials at the British Columbia Cancer Agency (BCCA), sponsored by the U.S. National Cancer Institute.
  • This indicates the data was from Canada (PanCan, BCCA in British Columbia) and supported by the U.S. National Cancer Institute. Both were prospective population-based studies.

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

The document does not specify the number of experts or their qualifications for establishing ground truth specifically for Philips' internal V&V test set. It mentions the LNA application was validated to address "user needs" and simulate "actual use of the software," which implies expert input, but no details are provided.

For the Risk Calculator, the ground truth for malignancy in the McWilliams et al. study was established through tracking the final outcomes of all detected nodules. This likely involved pathology reports and clinical follow-up, adjudicated by clinical experts, but the exact number and qualifications of these experts are not detailed in this summary.

4. Adjudication Method for the Test Set

The document does not describe a specific adjudication method (e.g., 2+1, 3+1) for Philips' internal V&V test set. The validation process involved evaluating each test case for the complete clinical workflow and ensuring the design meets user needs, which might involve expert review, but the formal adjudication protocol is not elaborated upon in this summary.

For the Risk Calculator's underlying study (McWilliams et al.), the "final outcomes of all nodules" suggests a definitive ground truth based on pathology or long-term clinical stability/progression, but the adjudication method for these biological outcomes is not specified within this document.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study Was Done, and the Effect Size of How Much Human Readers Improve with AI vs. Without AI Assistance

The document does not report an MRMC comparative effectiveness study where human readers' performance with and without AI assistance was evaluated. The studies described focus on the standalone performance and validation of the LNA application's features and the underlying model for the risk calculator.

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

Yes, standalone performance was evaluated for various features of the LNA application:

• The automatic segmentation capabilities (lungs, lobes, nodules) were validated to be "adequate."
• The automatic measurement calculations (diameters, volume, mean HU) were tested to comply with "defined functionality requirements and performance claims."
• The comparison and matching functionality and "Prefill functionality for the Lung RADS score and the risk prediction" were assured to be "adequate."
• The Risk Calculator tool itself (based on McWilliams et al.) demonstrated standalone predictive performance with "excellent discrimination and calibration, with areas under the receiver-operating-characteristic curve of more than 0.90." This indicates strong standalone performance of the algorithm in predicting malignancy.

7. The Type of Ground Truth Used

• For Philips' Internal V&V: The ground truth appears to be based on "real recorded clinical data cases," implying clinical diagnoses, measurements, and potentially pathology results where applicable, as evaluated against the software's specified functionality and user needs. The specific hierarchy or gold standard used for each feature's ground truth (e.g., expert consensus for segmentation, pathology for nodule type) is not explicitly detailed.
• For the Risk Calculator (McWilliams et al. study): The ground truth for malignancy was established by tracking "the final outcomes of all nodules," which would primarily be pathology results for cancerous nodules and long-term clinical outcome data (stability or benign diagnosis) for non-cancerous ones.

8. The Sample Size for the Training Set

The document does not specify the sample size for the training set used for the LNA application's algorithms, including the segmentation, measurement, and comparison features.

For the Risk Calculator's underlying model (McWilliams et al.):

• The "development data set" (training set) included participants from the Pan-Canadian Early Detection of Lung Cancer Study (PanCan). The exact number of participants or nodules is not provided in this summary but the PanCan study is a large, population-based study.

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

For the Risk Calculator's underlying model (McWilliams et al.):

• The ground truth for the development data set (PanCan study) was established by tracking "the final outcomes of all nodules of any size that were detected on baseline low-dose CT scans." This indicates that the ground truth for malignancy was based on definitive pathological diagnosis or long-term clinical follow-up confirming benignity or stability.

Ask a specific question about this device

Multi-Modality Tumor Tracking (MMTT) application is a post processing software application used to display, process, analyze , quantify and manipulate anatomical and functional images, for CT, MR PET/CT and SPECT/CT images and/or multiple time-points. The MMTT application is intended for use on tumors which are known/confirmed to be pathologically diagnosed cancer. The results obtained may be used as a tool by clinicians in determining the diagnosis of patient disease conditions in various organs, tissues, and other anatomical structure.

Philips Medical Systems' Multi-Modality Tumor Tracking (MMTT) application is a post - processing software. It is a non-organ specific, multi-modality application which is intended to function as an advanced visualization application. The MMTT application is intended for displaying, processing, analyzing, quantifying and manipulating anatomical and functional images, from multi-modality of CT ,MR PET/CT and SPECT/CT scans.

The Multi-Modality Tumor Tracking (MMTT) application allows the user to view imaging, perform segmentation and measurements and provides quantitative and characterizing information of oncology lesions, such as solid tumor and lymph node, for a single study or over the time course of several studies (multiple time-points). Based on the measurements, the MMTT application provides an automatic tool which may be used by clinicians in diagnosis, management and surveillance of solid tumors and lymph node, conditions in various organs, tissues, and other anatomical structures, based on different oncology response criteria.

The provided text does not contain detailed information about a study that proves the device meets specific acceptance criteria, nor does it include a table of acceptance criteria and reported device performance.

The submission is a 510(k) premarket notification for the "Multi-Modality Tumor Tracking (MMTT) application." For 510(k) submissions, the primary goal is to demonstrate substantial equivalence to a legally marketed predicate device, rather than proving a device meets specific, pre-defined performance acceptance criteria through a rigorous clinical or non-clinical study that would be typical for a PMA (Premarket Approval) application.

Here's what can be extracted and inferred from the document regarding the device's validation:

Key Information from the Document:

• Study Type: No clinical studies were required or performed to support equivalence. The validation was based on non-clinical performance testing, specifically "Verification and Validation (V&V) activities."
• Demonstration of Compliance: The V&V tests were intended to demonstrate compliance with international and FDA-recognized consensus standards and FDA guidance documents, and that the device "Meets the acceptance criteria and is adequate for its intended use and specifications."
• Acceptance Criteria (Implied): While no quantitative table is provided, the acceptance criteria are implicitly tied to:
  • Compliance with standards: ISO 14971, IEC 62304, IEC 62366-1, DICOM PS 3.1-3.18.
  • Compliance with FDA guidance documents for software in medical devices.
  • Addressing intended use, technological characteristics claims, requirement specifications, and risk management results.
  • Functionality requirements and performance claims as described in the device description (e.g., longitudinal follow-up, multi-modality support, automated/manual registration, segmentation, measurement calculations, support for oncology response criteria, SUV calculations).
• Performance (Implied): "Testing performed demonstrated the Multi-Modality Tumor Tracking (MMTT) meets all defined functionality requirements and performance claims." Specific quantitative performance metrics are not given.

Information NOT present in the document:

The following information, which would typically be found in a detailed study report proving acceptance criteria, is not available in this 510(k) summary:

1. A table of acceptance criteria and the reported device performance: This document states the device "Meets the acceptance criteria and is adequate for its intended use and specifications," but does not list these criteria or the test results.
2. Sample sizes used for the test set and the data provenance: No details on the number of images, patients, or data characteristics used for non-clinical testing.
3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts (e.g., radiologist with 10 years of experience): Since it was non-clinical testing, there's no mention of expert involvement in establishing ground truth for a test set.
4. Adjudication method (e.g., 2+1, 3+1, none) for the test set: Not applicable as no expert-adjudicated clinical test set 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 MRMC study was performed as no clinical studies were undertaken.
6. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done: The V&V activities would have included testing the software's functionality, which could be considered standalone performance testing, but specific metrics are not provided. The device is a "post processing software application" used "by clinicians," implying a human-in-the-loop tool rather than a fully autonomous AI diagnostic device.
7. The type of ground truth used (expert consensus, pathology, outcomes data, etc.): Not detailed for the non-clinical V&V testing. For the intended use, the device is for "tumors which are known/confirmed to be pathologically diagnosed cancer," suggesting that the "ground truth" for the intended use context is pathological diagnosis. However, this is not the ground truth for the V&V testing itself.
8. The sample size for the training set: Not applicable; this is a 510(k) for a software application, not specifically an AI/ML product where a training set size would be relevant for model development. The document does not describe any machine learning model training.
9. How the ground truth for the training set was established: Not applicable for the same reason as above.

In summary, this 510(k) submission relies on a demonstration of substantial equivalence to existing predicate devices and internal non-clinical verification and validation testing, rather than a clinical study with specific, quantifiable performance metrics against an established ground truth.

Ask a specific question about this device