(102 days)
Methinks NCCT Stroke is a radiological computer aided triage and notification software indicated for use in the analysis of (1) non-contrast head CT (NCCT) images. The device is intended to assist hospital networks and trained physicians in workflow triage by flagging and communicating suspected positive findings of (1) Intracranial Hemorrhage (ICH) and (2) Large Vessel Occlusion (LVO) of the ICA, MCA-M1 and MCA-M2.
Methinks NCCT Stroke uses an artificial intelligence algorithm to analyze images and highlight cases with suspected (1) ICH and (2) LVO in the cloud in parallel to the ongoing standard of care image interpretation. The user is presented with notifications for cases with suspected ICH or LVO findings via PACS and/or notifications. Notifications include preview images that are meant for informational purposes only, and are not intended for diagnostic use beyond notification.
The device does not alter the original medical image, and it is not intended to be used as a primary diagnostic device. The results of Methinks NCCT Stroke are intended to be used in conjunction with other patient information and based on professional judgment to assist with triage/prioritization of medical images. Notified clinicians are ultimately responsible for reviewing full images per the standard of care. Methinks NCCT Stroke is for adults only.
Methinks NCCT Stroke is a radiological computer-assisted triage and notification software device. The device receives Non-Contrast Computed Tomography (NCCT) images and processes them to provide triage and notification prioritization of suspected Intracranial Hemorrhage (ICH) and Large Vessel Occlusion (LVO) of the ICA, MCA-M1 and MCA-M2. The Methinks NCCT Stroke device is an AI/ML Software as a Medical Device. The outputs of the device are intended to be used by trained clinicians in the prioritization of patients with suspected ICH and/or LVO.
The provided FDA 510(k) clearance letter for the Methinks NCCT Stroke device details the acceptance criteria and the study that proves the device meets these criteria. Here's a breakdown of the requested information:
Acceptance Criteria and Reported Device Performance
Table of Acceptance Criteria and Reported Device Performance
The acceptance criteria are implied by the reported performance metrics, primarily sensitivity (Se) and specificity (Sp), for both Intracranial Hemorrhage (ICH) and Large Vessel Occlusion (LVO) detection. The document states that "Sensitivity and specificity exceed the pre-specified performance goals for ICH and LVO," although the exact numerical "goals" are not explicitly stated. The performance of the device against human readers is also an implicit acceptance criterion.
| Metric | Condition | Pre-specified Performance Goal (Implied Minimum) | Reported Device Performance | 95% Confidence Interval |
|---|---|---|---|---|
| ICH Detection | Sensitivity (Se) | > 89.3% | 94.7% | 89.3% - 97.8% |
| Specificity (Sp) | > 97.5% | 99.5% | 97.5% - 99.9% | |
| LVO Detection | Sensitivity (Se) | > 67.3% | 76.4% | 67.3% - 83.9% |
| Specificity (Sp) | > 86.6% | 91.1% | 86.6% - 94.5% | |
| LVO Reader Study (Versus Experts) | Sensitivity (Se) - Superiority | N/A (Device Se > Expert Se) | Device: 73.6% | 59.7% - 84.7% |
| Experts: 50.0% | 40.1% - 59.9% | |||
| LVO Reader Study (Versus Non-Experts) | Sensitivity (Se) - Superiority | N/A (Device Se > Non-Expert Se) | Device: 73.6% | 59.7% - 84.7% |
| Non-Experts: 37.7% | 28.5% - 47.7% | |||
| Time to Notification | NCCT-ICH | N/A | 1.43 minutes | 1.36 - 1.50 minutes |
| NCCT-LVO | N/A | 1.42 minutes | 1.36 - 1.48 minutes |
Study Information
-
Sample sizes used for the test set and the data provenance:
- ICH Test Set: 358 cases (132 ICH Positive, 226 ICH Negative)
- LVO Test Set: 335 cases (110 LVO Positive, 225 LVO Negative)
- Data Provenance: Retrospective, blinded, multicenter, multinational study.
-
Number of experts used to establish the ground truth for the test set and the qualifications of those experts:
- The document implies that ground truth for the initial performance evaluation (Se and Sp for ICH and LVO) was established through "expert reader truthing of the data." The number and qualifications of these specific experts for ground truth establishment are not explicitly stated beyond "expert reader."
- For the reader study, there were 4 readers involved: 2 "expert neuroradiologists" and 2 "general radiologists (non-experts)." Their specific years of experience or other detailed qualifications are not provided beyond these labels.
-
Adjudication method for the test set:
- The document mentions "expert reader truthing of the data" for establishing ground truth but does not specify a detailed adjudication method (e.g., 2+1, 3+1). For the reader study, the individual performance of the readers is provided, implying that their interpretations were compared against the established ground truth, but not that they formally adjudicated for the ground truth itself within the study.
-
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 MRMC comparative study was done comparing the device's performance to human readers (radiologists) without AI assistance.
- Effect Size of AI vs. Human Readers (Standalone AI vs. Human Alone):
- LVO Sensitivity:
- Methinks NCCT-LVO: 73.6%
- Expert Neuroradiologists (R1 + R2): 50.0%
- General Radiologists (R3 + R4): 37.7%
- Difference in Sensitivity (Effect Size):
- Methinks NCCT-LVO vs. Experts: 23.6% (95%CI: 8.5% - 38.7%), showing superiority of the device.
- Methinks NCCT-LVO vs. Non-experts: 35.9% (95%CI: 16.0% - 42.9%), also showing superiority of the device.
- LVO Sensitivity:
- The study does not report how much human readers improve with AI assistance (i.e., human-in-the-loop performance). It focuses on the standalone performance of the AI compared to human readers working without AI.
-
If a standalone (i.e. algorithm only without human-in-the-loop performance) was done:
- Yes, a standalone performance evaluation of the Methinks NCCT Stroke algorithm was done for both ICH and LVO detection. The reported sensitivity and specificity metrics (e.g., ICH Se: 94.7%, Sp: 99.5%; LVO Se: 76.4%, Sp: 91.1%) are for the algorithm only.
-
The type of ground truth used:
- The ground truth for the test set was established by "expert reader truthing of the data." This implies a consensus of medical experts, likely radiologists or neuroradiologists, reviewing the images. It is not explicitly stated if pathology, surgical findings, or long-term clinical outcomes were used to confirm the ground truth.
-
The sample size for the training set:
- The document does not specify the sample size for the training set. It only mentions the test set sizes.
-
How the ground truth for the training set was established:
- The document does not specify how the ground truth for the training set was established. It only mentions the "expert reader truthing of the data" in the context of the performance validation (test set).
§ 892.2080 Radiological computer aided triage and notification software.
(a)
Identification. Radiological computer aided triage and notification software is an image processing prescription device intended to aid in prioritization and triage of radiological medical images. The device notifies a designated list of clinicians of the availability of time sensitive radiological medical images for review based on computer aided image analysis of those images performed by the device. The device does not mark, highlight, or direct users' attention to a specific location in the original image. The device does not remove cases from a reading queue. The device operates in parallel with the standard of care, which remains the default option for all cases.(b)
Classification. Class II (special controls). The special controls for this device are:(1) Design verification and validation must include:
(i) A detailed description of the notification and triage algorithms and all underlying image analysis algorithms including, but not limited to, a detailed description of the algorithm inputs and outputs, each major component or block, how the algorithm affects or relates to clinical practice or patient care, and any algorithm limitations.
(ii) A detailed description of pre-specified performance testing protocols and dataset(s) used to assess whether the device will provide effective triage (
e.g., improved time to review of prioritized images for pre-specified clinicians).(iii) Results from performance testing that demonstrate that the device will provide effective triage. The performance assessment must be based on an appropriate measure to estimate the clinical effectiveness. The test dataset must contain sufficient numbers of cases from important cohorts (
e.g., subsets defined by clinically relevant confounders, effect modifiers, associated diseases, and subsets defined by image acquisition characteristics) such that the performance estimates and confidence intervals for these individual subsets can be characterized with the device for the intended use population and imaging equipment.(iv) Stand-alone performance testing protocols and results of the device.
(v) Appropriate software documentation (
e.g., device hazard analysis; software requirements specification document; software design specification document; traceability analysis; description of verification and validation activities including system level test protocol, pass/fail criteria, and results).(2) Labeling must include the following:
(i) A detailed description of the patient population for which the device is indicated for use;
(ii) A detailed description of the intended user and user training that addresses appropriate use protocols for the device;
(iii) Discussion of warnings, precautions, and limitations must include situations in which the device may fail or may not operate at its expected performance level (
e.g., poor image quality for certain subpopulations), as applicable;(iv) A detailed description of compatible imaging hardware, imaging protocols, and requirements for input images;
(v) Device operating instructions; and
(vi) A detailed summary of the performance testing, including: test methods, dataset characteristics, triage effectiveness (
e.g., improved time to review of prioritized images for pre-specified clinicians), diagnostic accuracy of algorithms informing triage decision, and results with associated statistical uncertainty (e.g., confidence intervals), including a summary of subanalyses on case distributions stratified by relevant confounders, such as lesion and organ characteristics, disease stages, and imaging equipment.