Search Results

The ICTA software is intended as a review tool to mark previously acquired sections of the adult (greater than or equal to 18 years) EEG recordings (surface or intracranial) that may correspond to electrographic seizures, in order to assist qualified clinical practitioners, who will exercise professional judgment in using the information, in the assessment of EEG traces.

• Surface recordings must be obtained with full montage according to the standard 10/20 . system.
• Intracranial recordings must be obtained with depth electrodes (strips and/or grids). .
  This device does not provide any diagnostic conclusion about the patient's condition to the user.

ICTA is a software only product. It runs on a personal computer and requires no specialized hardware. It identifies electroencephalographic activity that might correspond to seizures (referred as "events"). These events are then reviewed, accepted, modified and/or deleted by the qualified medical practitioner. The software does not make any final decisions that result in any automatic diagnosis or treatment. The EEG input is read from a file on the personal computer (or available across the network).
ICTA employs Bayesian formulation to provide a detection variable based on the probabilities that a given section of EEG contains a seizure-like activity. The a priori probabilities that a certain set of features represent seizure or non-seizure data were computed from the training data set. These probabilities are used by the detection method for all seizure detections.
The software has two components: ICTA-S for analysis of surface EEG recordings and ICTA-D for analysis of intracranial recordings. Whether a particular module is active is determined by the user. The user also determines parameters that are needed for the algorithm to perform its intended task. None of the components is responsible for data acquisition, review or any other function different from analysis.

Here's a breakdown of the acceptance criteria and study details for the ICTA device, based on the provided 510(k) summary:

1. Table of Acceptance Criteria and Reported Device Performance

The acceptance criteria for ICTA were established through a comparison with a predicate device (NeuroWorks Seizure Detector, K090019) and the "gold standard" of expert neurophysiologists. The key performance metrics are Positive Percent Agreement (PPA) and False Detection Rate (FDR).

Note: The document states "Equivalent" for both metrics when comparing to the predicate, even though the FDRs are numerically different. This suggests the FDA considers these values acceptable within the context of seizure detection assistance tools.

2. Sample Sizes Used for the Test Set and Data Provenance

• ICTA-S (Surface EEG):
  • Number of Seizures: 615
  • Total Number of Patients: 102
  • Total Number of Hours: 395
  • Data Provenance: Retrospective, patients with medically refractory seizures admitted to an Epilepsy Monitoring Unit. The specific country of origin is not explicitly stated, but Natus Medical Incorporated DBA Excel-Tech Ltd. is based in Oakville, Ontario, Canada.
• ICTA-D (Intracranial EEG):
  • Number of Seizures: 429
  • Total Number of Patients: 93 (57 Male, 36 Female)
  • Total Number of Hours: 619 hours
  • Data Provenance: Retrospective, adult patients seen for routine clinical evaluation at Epilepsy Monitoring Units of Toronto Western General Hospital (Canada) and NewYork-Presbyterian Hospital (USA).

3. Number of Experts Used to Establish Ground Truth for the Test Set and Their Qualifications

• Number of Experts: Three independent, blinded EEG experts were used for both ICTA-S and ICTA-D studies.
• Qualifications: All experts were board-certified Neurophysiologists (or neurologists/epileptologists). The document does not specify their years of experience.

4. Adjudication Method for the Test Set

• Adjudication Method: A "majority rule (at least 2 out of 3)" was applied. This means that for a seizure to be considered a "true" electrographic seizure (ground truth), at least two of the three independent experts had to agree on its presence.

5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study

• The document does not describe a multi-reader multi-case (MRMC) comparative effectiveness study where human readers' performance with and without AI assistance was evaluated. The study focuses on evaluating the standalone performance of the ICTA algorithm against a human-established ground truth and comparing it to a predicate device's reported performance.

6. Standalone Performance Study

• Yes, a standalone (algorithm only without human-in-the-loop performance) study was done. The entire clinical testing section describes the evaluation of the ICTA-S and ICTA-D algorithms' performance (PPA and FDR) independently against the ground truth established by the expert panel. The results presented in the tables (e.g., PPA 75% / FDR 2.0 FP/h for ICTA-S) are for the algorithm in standalone mode.

7. Type of Ground Truth Used

• Type of Ground Truth: Expert consensus. Specifically, electrographic seizures identified by a panel of three board-certified Neurophysiologists, with a majority rule for final determination.

8. Sample Size for the Training Set

• The document states that Bayesian formulation was used, and "The a priori probabilities that a certain set of features represent seizure or non-seizure data were computed from the training data set."
• However, the specific sample size for the training set is not provided in the summary.

9. How the Ground Truth for the Training Set Was Established

• The document states that probabilities were computed from the "training data set." It does not explicitly detail the method for establishing ground truth for this training set. However, given the nature of the device and the methods described for the test set, it is highly probable that the ground truth for the training set was also established through expert review and annotation of EEG recordings, likely by qualified medical practitioners. The summary implies that this training data was used to establish the "a priori probabilities" for the Bayesian model.

Ask a specific question about this device