A Development of Clinical Decision Support System for Video Head Impulse Test Based on Fuzzy Inference System

Figure 1

System block diagram. After loading the data file (.csv), the head impulse and eye response are first smoothed, then used for detecting parameters. Those include onset, minimum and maximum values, significant peaks, and catch-up saccades (CUS). If the CUS appears before the head impulse end, it is labeled as covert CUS, and the gain is calculated via desaccaded position gain algorithm. Otherwise, it can be a normal response, or contains overt CUS, which can be separated due to the number and time of significant peaks. After all, the gain and CUS proportion is calculated and inputted to fuzzy inference system to obtain normality and artifact index. Artifact handling included smoothing filter, peaks (#, amplitude, and times), and gain. Artifact index is theoretically contradicting such as low gain but no CUS, or gain within normal range but still have CUS.