Normative Database

Firefly Neuroscience’s proprietary commercial database is based on the normative dataset that is the bedrock of our new 510(K) cleared BNA™ Platform V1.0 product. The FDA-reviewed, multi-paradigm, high-quality normative database covers the age range 12-85. Thanks to this unique database, the BNA™ reports present the BNA™ ERP/Resting EEG scores at the right context – in comparison to age-matched normative data as standardized scores (z-scores). This allows physicians to answer the patient’s primary fundamental question of ‘Am I Normal?’ and to understand where the differences may lie.  Further, after a patient undergoes their first BNA™ test, a baseline measurement is established.  In conjunction with the advanced analysis and reporting capabilities, the changes are easily seen and summarized graphically.

The following (Figure 1) is a diagram breaking down the dataset that underpins the commercial BNA™ V1.0 product:

Firefly Neuroscience’s Normative Database (used with BNA™)

The fact that the FDA has approved both the ERP (Auditory and Visual) and Resting-EEG normative databases puts Firefly Neuroscience in a unique competitive position; as illustrated in Figure 2, the BNA™ product empowers clinicians by providing them with a multi-perspective, objective evaluation of the patient’s brain in one product.  This is highly useful to clinicians as the information from the Resting-EEG and ERP tasks are complementary to each other in many ways.

The BNA™ report is accompanied by information about expected deviances in different brain disorders that we provide in form of Clinician EEG/ERP Atlas. Together, the objective BNA™ scores and the ERP/EEG atlas increase outcome interpretability and enable better decision-making in the patient flow.

The BNA™ Algorithm

The BNA^TM-ERP algorithm aims to detect and quantify the characteristics of ERP peaks of well-known ERP components (AOB task: P50, N100, P200, P3a, P3b; VGNG task: P200, N200, P3a, P3b). The BNA™ algorithm is described in more detail in the IFU document (available on request)  and in Stern et al., 2016 publication. However, we describe here shortly the main advantage of the BNA algorithm and how it leverages the normative database to extract clinically meaningful information.

EEG recordings are high-dimensional and complex. As figure 3 illustrates, this is the result of the dynamic spatiotemporal nature of the neurophysiological signals. For example, the event-related potential wave measured at first, at a relatively focal location in time and space on the scalp, and then it evolves and propagates in the cortex while its amplitude changes simultaneously.

Traditional methods for EEG and event-related potential (ERP) analysis follow waveform morphology over time at selected electrode locations, using either time-domain or frequency-domain tools, while neglecting the spatiotemporal dynamics associated with the electric field at the scalp. Moreover, focusing on a specific electrode or subset of electrodes, the traditional methods ignore the expected variability between patients in terms of where the ERP component appears on the scalp due to natural differences between cortex and scalp anatomy.

The aim of the BNA™-ERP algorithm is to improve the accuracy of the ERP component detection by considering the spatiotemporal nature of the signal. It achieves this improvement by parceling the EEG activity into major spatiotemporal events- ERP peaks and its surrounding – spatiotemporal parcels (or ‘STEPs’). Figure 4 describes the algorithm process; as a prerequisite step, a set of main spatiotemporal events representing the reference group was generated by clustering the spatiotemporal event of all individual subjects in each normative group age-bin. Figure 4 (panel to the right) illustrates the ERP peaks clusters in a 3D graph with two spatial axes – ‘left-right’ and ‘posterior-anterior’ and time axis. These reference group clusters are utilized to automatically match between the patient’s ERP peak and the subset of ERP components of interest.

The fact that the ERP peaks clusters were calculated per age-bin enables a more fine-tuned ERP peak matching process- each main ERP peak is identified by proximity in time and location relative to the age-matched, clusters. 

To summarize the key strengths of the BNA™ ERP algorithm are:

1. Adaptive peak detection – as described, the ERP peaks are detected in a more patient-specific way, by incorporating some degree of freedom in both time and space in a way that is proportional to the normal variability in time and space of the ERP peaks.
2. Displaying ERP-peak location – thanks to the adaptive peak detection by the BNA algorithm, the ERP peak is identified and displayed in the BNA report on a topographical map together with the group ERP-peak position. This adds clinically useful information to the physician.
3. Neural-Consistency score – as mentioned in answer 1, another unique feature of the BNA report is the presentation of the Neural-Consistency score. Since this is a differentiating score, we will describe here shortly what is this score and why we believe it is important.

Since the ERP signal is an average of multiple single-trial ERP signals which can differ largely from each other, assessing the consistency between them may provide complementary functional information to the average measure. As part of the new peak-analysis, the algorithm calculates a new score – ‘Neural Consistency’ based on the similarity of the amplitude activation between single-ERP trials (epochs, Figure 5). The score is calculated based on averaging the inter-single trials variability of the ERP-peak and its surrounding points.           

Trial-by-trial variability of specific ERP components has been studied within multiple domains: it has been suggested as an index of the cognitive and information processing capacity of the brain, which may not be reflected by standard behavioral measures (McIntosh, 2008); it has been shown to increase in patients with schizophrenia  (Anderson, 1991); (Roth, 2007)), ADHD  (Gonen-Yaacovi, 2016); (Lazzaro, 1997)), autism spectrum disorder  (Milne, 2011); (Weinger, 2014), and may prove useful for understanding key periods of age-related decline marking the transition from normal to pathological aging  (Hogan, 2006).