(248 days)
CuraRad-ICH is a software workflow tool designed to aid in prioritizing the clinical assessment of adult non-contrast head CT cases with features suggestive of acute intracranial hemorrhage. CuraRad-ICH analyzes cases using deep learning algorithms to identify suspected ICH findings. It makes case-level output available to a PACS/workstation for worklist prioritization or triage.
CuraRad-ICH is not intended to direct attention to specific portions of an image or to anomalies other than acute ICH. Its results are not intended to be used on a stand-alone basis for clinical decision-making nor is it intended to rule out hemorrhage or otherwise preclude clinical assessment of CT studies.
CuraRad-ICH is software as a medical device (SaMD) that detects intracranial hemorrhage (ICH) condition by analyzing non-contrast CT images. The software needs to be integrated with a third-party worklist application to receive analysis requests and the corresponding DICOM images, and return the ICH findings (whether suspected ICH is found) to the worklist to alert the radiologists.
To help radiologists triage and prioritize reading of images for patients with ICH, CuraRad-ICH uses deep learning methods to automatically detect acute ICH in non-contrast head CT scans. The software analyzes the input image and returns a binary prediction as to whether the exam suggests the presence of acute ICH.
Here's a breakdown of the acceptance criteria and the study proving the device meets them, based on the provided text:
1. Table of Acceptance Criteria & Reported Device Performance
| Metric | Acceptance Criteria (Implied) | Reported Device Performance (95% CI) |
|---|---|---|
| Sensitivity | Clinically Meaningful | 90.6% (85.9%-94.2%) |
| Specificity | Clinically Meaningful | 93.1% (88.3%-96.4%) |
| Positive Predictive Value (PPV) | Reasonable given prevalence | Varies by prevalence: |
| 1% prevalence: 11.8% (7.2%-18.8%) | ||
| 15% prevalence: 70% (57.4%-80.1%) | ||
| 54.9% prevalence (actual): 89.1% (84.3%-92.5%) | ||
| Negative Predictive Value (NPV) | Very High | Varies by prevalence: |
| 1% prevalence: 99.9% (99.8%-100%) | ||
| 15% prevalence: 98.3% (97.4%-98.8%) | ||
| 54.9% prevalence (actual): 91.8% (88.6%-94.3%) | ||
| System Processing Time | Not explicitly stated but implies quick processing | Average: 43 seconds (39-46) |
| Median: 33 seconds | ||
| Min: 16 seconds | ||
| Max: 301 seconds |
Note: The document states that the performance "met the pre-specified criteria for study success," implying that the reported values for sensitivity and specificity were the acceptance criteria.
2. Sample Size Used for the Test Set and Data Provenance
- Sample Size: 388 CT studies (213 positives and 175 negatives)
- Data Provenance: Retrospective, collected from 296 imaging facilities across 48 states in the US.
3. Number of Experts Used to Establish Ground Truth for the Test Set and Qualifications
The document states ground truth was "provided by experienced radiologists" but does not specify the number of experts or their specific qualifications (e.g., years of experience).
4. Adjudication Method for the Test Set
The document does not explicitly state the adjudication method used for establishing ground truth for the test set. It only mentions "ground truth provided by experienced radiologists."
5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study Was Done
No, an MRMC comparative effectiveness study was not reported. The study focused on the standalone performance of the AI algorithm.
6. If a Standalone (i.e., Algorithm Only Without Human-in-the-Loop Performance) Was Done
Yes, a standalone performance study was done. The reported sensitivity, specificity, PPV, and NPV are for the CuraRad-ICH algorithm's performance in identifying ICH.
7. The Type of Ground Truth Used
The ground truth was established by "experienced radiologists," implying an expert consensus or expert-determined ground truth.
8. The Sample Size for the Training Set
The document does not specify the sample size used for the training set. It only states that the deep learning algorithm was "trained on non-contrast head CT scans with ICH ground truth."
9. How the Ground Truth for the Training Set Was Established
The ground truth for the training set was "provided by experienced radiologists." This implies expert labeling or consensus, similar to the test set, though no further details are given.
§ 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.