(116 days)
ANDI is intended for display of medical images and other healthcare data. It includes functions for image review, basic measurements, planning, and visualization (MPR reconstructions), Modules are available for image processing and atlas-assisted visualization and segmentation, where an output can be generated for use by a system capable of reading DICOM image sets.
ANDI is indicated for use in the processing of diffusion-weighted MRI sequences into 3D maps that represent white matter tracts. The information presented by ANDI is for measurement of brain white matter tracts only without making a prediction, diagnosis, or interpretation of brain health, It is the responsibility of the physician to review all clinical information associated with a patient in order to make a diagnosis and to determine next steps in the clinical care of the patient.
Typical users of ANDI are medical professionals, including but not limited to neurologists and radiologists. ANDI should be used only as adjunctive information. The decision made by trained medical professionals will be considered final. It is not a diagnostic aid.
ANDI is quantitative imaging software that extracts features from medical images to provide adjunctive information for use with the complete standard of care evaluation of the patient. The device processes diffusion weighted images using local (on a per-voxel basis) and global (for the whole brain) reconstruction algorithms, respectively called modeling, tractography, and white matter bundling, to map microstructural properties of the white matter. ANDI achieves its intended use by extracting white matter bundles that connect specific regions of the brain and then performing a microstructure analysis along those bundles.
ANDI combines two families of features derived from diffusion Magnetic Resonance Imaging (dMRI) along with T1-weighted images as an auxiliary input to augment processing. T1-weighted imaging is a general imaging modality that highlights differences in tissue types (skull, cerebrospinal fluid, white / gray matter), whereas dMRI is sensitive to the direction of white matter structures, based on the movements of water protons. Combining information from these MRI types gives ANDI data needed to quantify and visualize local water diffusion properties and to map these microstructural properties of the white matter along specific white matter bundles.
The ANDI analysis techniques provide quantitative insight into white matter microstructure, corresponding to the local environment of each neurological fiber population. The resulting report provides trained medical professionals with reference information as an adjunct to care. The information included in the report is intended to be used by the trained medical professionals as a comparison between a patient and a control population as well as a comparison to the subject himself, with an optional longitudinal comparison.
Based on the provided text, here's a detailed breakdown of the acceptance criteria and related study information for the ANDI device:
Acceptance Criteria and Reported Device Performance
The provided text does not explicitly list specific numerical acceptance criteria for the ANDI device's performance (e.g., a specific accuracy, sensitivity, or specificity threshold). Instead, it states that "Performance tests were conducted to assess the measured end points, AI-based brain extraction, and robustness of the processing pipeline." and "Through the performance test, it was confirmed that ANDI meets all performance test criteria and that all functions work as intended."
Given the information, the reported device performance is qualitative:
| Acceptance Criterion (Implicit) | Reported Device Performance |
|---|---|
| Accurate assessment of measured endpoints | Meets all performance test criteria and functions work as intended. |
| Effective AI-based brain extraction | Meets all performance test criteria and functions work as intended. |
| Robustness of the processing pipeline | Meets all performance test criteria and functions work as intended. |
| Software functions as intended and satisfies all requirements | Confirmed that all functions work as intended and satisfies all expected and previously defined system requirements and features. |
| Compliance with design specifications and technological characteristics | Device meets applicable requirements and standards for safety and effectiveness. |
Study Details
The document explicitly states, "No clinical studies were considered necessary and performed." This indicates that the evaluation of ANDI's performance was based solely on non-clinical (i.e., in-silico or bench-top) testing. Therefore, many of the typical clinical study elements listed in your prompt are not applicable.
Here's what can be extracted and what remains unknown based on the provided text:
-
Sample size used for the test set and the data provenance:
- Test Set Sample Size: Not explicitly stated. The text mentions "Performance tests were conducted," but it doesn't quantify the number of cases or images used in these tests.
- Data Provenance: Not specified. As no clinical studies were performed, the origin of any data used for non-clinical performance testing (if patient data was used) is not disclosed. It's possible simulated or publicly available datasets were used, but this is not confirmed.
-
Number of experts used to establish the ground truth for the test set and the qualifications of those experts:
- Since no clinical studies were performed, and no explicit ground truth establishment process for a "test set" in a clinical context is described, this information is not available. The "performance tests" mentioned are likely technical evaluations against predefined software requirements rather than clinical ground truth validation.
- The text does state, "Test results were reviewed by designated technical professionals," but their number or specific qualifications for establishing ground truth are not detailed.
-
Adjudication method for the test set:
- Not applicable/Not described, as no clinical test set requiring expert adjudication for ground truth establishment is mentioned.
-
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 done. The document explicitly states, "No clinical studies were considered necessary and performed." The device is intended to provide "adjunctive information," and the "decision made by trained medical professionals will be considered final," indicating it's not a diagnostic aid or intended to replace human interpretation, but rather to provide quantitative insights.
-
If a standalone (i.e. algorithm only without human-in-the-loop performance) was done:
- Yes, implicitly. The "Performance tests" described in the "Non-Clinical Test Summary" section would represent standalone algorithm performance, as no human readers are involved in these tests to validate the algorithm outputs against a clinical ground truth. The text confirms the device "meets all performance test criteria and that all functions work as intended" as a standalone software.
-
The type of ground truth used (expert consensus, pathology, outcomes data, etc):
- For the non-clinical performance tests, the "ground truth" would likely be design specifications and predetermined functional requirements established for the software, rather than clinical ground truth like pathology or expert consensus. The software was validated against these requirements.
-
The sample size for the training set:
- Not provided. The document describes the device's function and non-clinical testing but does not offer details about the training data or its size, which is common for regulatory summaries focusing on validation rather than development.
-
How the ground truth for the training set was established:
- Not provided. Similar to the training set size, the method for establishing ground truth for any potential training data is not discussed in this regulatory summary.
§ 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).