(354 days)
PROView is an aiding tool for the clinicians to review multi-parametric resonance (MR) inages following PI-RADS guidelines. It displays acquired and reformatted data for visualization and provides tools for assessment of the prostate gland volume and findings analysis in patients with known or suspected prostate lesions. Measurements and associated scoring are included in a report for communication to referring physicians. It is intended for use by professionals, such as clinicians, radiologists, or physicians. The clinician remains ultinately responsible for the final assessment and diagnosis based on state-of-the-art practices, clinical judgment and interpretation of prostate images or quantitative data.
PROView offers a guided workflow for the review, assessment and reporting of multi-parametric MR prostate exams. From inputting clinical information, measuring prostate and lesion volume to scoring lesions to form a comprehensive MR report, PROView offers a simple workflow per PI-RADS™ v2.1 guideline. PROView Processes data from a single date. The PROView workflow includes: Prostate volume extracted from automatic organ segmentation, PSA Density, Lesion(s) mapping to sectors and measurement, Scoring of T2-weighted, diffusion weighted imaging (DWI) and, when applicable, dynamic contrast enhanced (DCE) acquisitions, Automatically generated report with all measurements and images. Prostate volume can be automatically calculated by defining the contours of the prostate gland with the use of a deep learning algorithm, or through a manual method. Users can cancel or switch to manual prostate gland volume definition if the automatic prostate gland segmentation fails or provides unsatisfactory results.
The provided text describes the PROView device, an aiding tool for clinicians to review multi-parametric MR images following PI-RADS guidelines for prostate imaging. The primary novel feature highlighted for evaluation is the fully automatic segmentation of the prostate based on a deep learning model.
Here's an analysis of the acceptance criteria and the study proving the device meets them:
1. Table of Acceptance Criteria and Reported Device Performance
The document explicitly states that the "algorithm meets the acceptance criteria and improves performance on volume accuracy." However, the specific quantitative acceptance criteria for volume accuracy (e.g., a percentage accuracy or a maximum deviation from ground truth) are not defined in the provided text. Similarly, the reported numerical performance that met these undefined criteria is also not provided.
Therefore, a table cannot be constructed with the required specificity from the given information. The document only offers a qualitative statement of success.
2. Sample Size Used for the Test Set and Data Provenance
- Test Set Sample Size: The document mentions "a database of MRI prostate exams" used for algorithm qualification (validation) but does not specify the sample size of this test set.
- Data Provenance: The text states, "This database of exams is considered as representative of the clinical scenarios where PROView is intended to be used, with consideration of the different protocols, practices and ethnics factors." This implies a diverse dataset, but does not explicitly state the country of origin or if it's retrospective or prospective data. Given the reference to "different protocols, practices and ethnics factors," it suggests a multi-site or internationally sourced retrospective database is likely, but this is an inference.
3. Number of Experts Used to Establish Ground Truth for the Test Set and Qualifications of Experts
The document does not provide any information regarding the number of experts used to establish ground truth or their qualifications.
4. Adjudication Method for the Test Set
The document does not specify any adjudication method used for establishing ground truth for the test set (e.g., 2+1, 3+1, none).
5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study
The provided text does not mention a multi-reader multi-case (MRMC) comparative effectiveness study, nor does it discuss the effect size of human readers improving with AI vs. without AI assistance. The study described focuses on the algorithm's standalone performance in prostate segmentation.
6. Standalone (Algorithm Only Without Human-in-the-Loop Performance) Study
Yes, a standalone study was done. The document explicitly states: "Engineering has validated PROView algorithm's capability of automatic segmentation based on deep learning technique by using a database of MRI prostate exams." This refers to evaluating the algorithm's performance independent of real-time human interaction during the segmentation process. The output of the algorithm (segmentation contour) is then presented to the user, who can adjust it.
7. Type of Ground Truth Used
The document states, "The results and feedback concluded that the algorithm meets the acceptance criteria and improves performance on volume accuracy." This strongly implies that the ground truth for prostate segmentation was established by expert consensus contouring or highly accurate manual segmentation that allows for the calculation of a reference prostate volume. While not explicitly stated, clinical practice for validating segmentation algorithms typically involves expert-drawn contours. It is not pathology or outcomes data in this context.
8. Sample Size for the Training Set
The document does not provide any information about the sample size used for the training set for the deep learning model.
9. How Ground Truth for the Training Set Was Established
The document states, "The algorithm provides a fully automatic segmentation of the prostate, based on a deep learning model." This implies that a training set with corresponding ground truth segmentations was used. However, the document does not provide any information on how this ground truth was established for the training data (e.g., by single expert, multiple experts, or a specific annotation protocol).
§ 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).