(78 days)
It is used by radiation oncology department to segment (non-contrast) CT images, to generate needed information for treatment planning, treatment evaluation and treatment adaptation
The proposed device, RT-Mind-AI, is a standalone software which used by radiation oncology department to segment (non-contrast) CT images, to generate needed information for treatment planning. treatment evaluation and treatment adaptation.
The proposed device has four main function:
- Deep learning contouring:
Automatic segment on desktop: it can automatically contour the organ-at-risk (ORA), including Head and Neck, thorax and abdominal and pelvic.
Automatic segment on the Web: it can realize automatic contouring the OAR on the web and A sending to the specified network node in a local area network (LAN). Note: only the administrator account and operator accounts can carry out the automatic segment on the Web.
Manual segment: Adjust the segment result after automatic segment. 2)
It also has the following general functions:
A Preset ROIs
Preset templates
Transmit DICOM data:
Desktop patient management
Review images;
A ROI management;
Web-based patient management
A Open and save of files.
The device is RT-Mind-AI, a software used by radiation oncology departments to segment non-contrast CT images for treatment planning, evaluation, and adaptation. The study aims to demonstrate that RT-Mind-AI meets the acceptance criteria for segmentation performance, primarily through Dice Similarity Coefficient (DSC) comparisons.
Here's an analysis of the acceptance criteria and the study that proves the device meets them:
1. Table of Acceptance Criteria and Reported Device Performance
The acceptance criteria for the RT-Mind-AI device's segmentation performance are implicitly based on demonstrating non-inferiority to its predicate device (K191928, AccuContour™) for common segmentable organs, and demonstrating acceptable performance for additional organs segmented by RT-Mind-AI. While specific numerical thresholds for DSC are not explicitly stated as "acceptance criteria" in a target value format within the provided text, the methodology for demonstrating non-inferiority and acceptable performance serves as the de-facto acceptance criteria.
| Acceptance Criterion | Reported Device Performance (RT-Mind-AI) |
|---|---|
| Non-inferiority for organs common to predicate device | DSC of proposed device (RT-Mind-AI) was non-inferiority compared with that of the predicate device (AccuContour™) for the same segment organs. |
| Acceptable performance for additional organs | The average DSC of additional segment organs was concluded to be non-inferiority compared to the average DSC of other (common) segment organs of the proposed device. |
2. Sample Sizes and Data Provenance
- Test Set Sample Size: Not explicitly stated as a number of images or patients. The text mentions "two sets of images for test group and control group, respectively" for the non-inferiority comparison with the predicate, and images for the additional organs. It states, "For each body parts, all intended organs were included in images."
- Data Provenance:
- Country of Origin: Images were generated in healthcare institutions in US.
- Retrospective/Prospective: Not explicitly stated, but typically, data collected for these types of validation studies are retrospective.
3. Number of Experts and Qualifications for Ground Truth
- Number of Experts: At least three licensed physicians.
- Qualifications of Experts: They were "licensed physicians." Specific specialties (e.g., radiation oncologist, radiologist) or years of experience are not mentioned, but the context of "radiation oncology department" suggests expertise in this area.
4. Adjudication Method for the Test Set
- Method: Consensus of at least three licensed physicians was used to establish the ground truth. This implies that discrepancies among the initial segmentations by the experts were resolved to reach a single agreed-upon segmentation for each image. This is a form of 3-reader consensus.
5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study
- Was it done? No. The provided text describes a technical validation of the segmentation software itself, comparing its output to expert consensus ground truth and to a predicate device's output. It does not describe a study involving human readers using the AI vs. not using the AI to assess improvements in human performance.
- Effect Size of Human Reader Improvement: Not applicable, as an MRMC study was not performed.
6. Standalone (Algorithm only without Human-in-the-loop) Performance
- Was it done? Yes. The study directly evaluates the automated segmentation accuracy of the RT-Mind-AI. The Dice Similarity Coefficient (DSC) was calculated between the device's automatic segmentation results and the expert-generated ground truth. This is a standalone performance evaluation.
7. Type of Ground Truth Used
- Type: Expert Consensus. Specifically, "manual segmentation was generated from the consensus of at least three licensed physicians."
8. Sample Size for the Training Set
- The sample size for the training set is not mentioned in the provided text.
9. How Ground Truth for the Training Set Was Established
- The method for establishing ground truth for the training set is not mentioned in the provided text. The document focuses exclusively on the test set validation.
§ 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).