Augmented Intelligence-Based Interference Pattern Analysis (AI-IPA) in Concentric Needle Electromyography.

Abstract

Introduction/aims: To add objectivity to the routine needle electromyography examination, we describe an "Augmented Intelligence" based interference pattern (IP) analysis method that mimics the subjective assessment by quantifying IP fullness, discreteness, amplitude, pitch, and motor unit firing rate (FR).

Methods: IP recordings from 20 control subjects and other patients with neuropathy and myopathy were analyzed. The IP was divided into three groups: low, intermediate, and full to mimic visual appearance. Reference values (RVs) were defined for each group. "Fence" pattern was defined based on the discreteness and amplitude. Upper limit of FR was defined. Various technical artifacts were detected and excluded from analysis.

Results: In control subjects, a total of 2435 recordings from 119 commonly tested muscles were analyzed. The single set of RVs was satisfactory across the tested muscles. Amplitude increased when the pattern changed from low to full. Pitch did not correlate with fullness and its RVs were same for all groups. In patients with neuropathy, an intermediate or low pattern, high amplitude, fence pattern, low pitch, and high FR were demonstrated. In patients with myopathy, a full pattern with low amplitude and high pitch was demonstrated.

Discussion: The algorithm makes simple measurements that are readily interpreted by the electromyographer. In this respect, it augments analysis by providing quantitative data. If implemented in an "on-line" manner, it can provide guidance to the operator without adding to the procedure time or changing the recording technique. The measurements can also be included in the report to support the study's findings.