Eficiencia de la detección automática de potenciales evocados auditivos de estado estable a múltiples frecuencias evaluada mediante la metodología ROC

Here we evaluate comparatively using receiver operating characteristics (ROC) curves, the efficiency of the automatic detection of multiple auditory steady state responses (MSSR) with an established methodology such as click auditory brainstem responses (cABR). For the automatic detection of cABR two statistics were calculated: the standard deviation ratio (SDR) between the signal and the background noise and the correlation coefficient ratio (CCR) between two replicates, whereas for the MSSR the statistics in the frequency domain Hotelling T2 (HT2) and circular T2 (CT2) were used. The ROC curves were computed over 222 recordings (34 without response y 188 with response) obtained at a fixed intensity of 40 dB nHL. Although these evoked responses (cABR and MSSR) could be identified efficiently by the different statistics, the MSSR showed better efficiency with higher areas under the ROC curve (HT2: 0,91-0,95 and CT2: 0,90-0,91 for 0,5 and 2 KHz) compared to the cABR (0,60 for CCR and 0,87 for the SDR). However, the MSSR had lower time-efficiency than the cABR (MSSR: 5,4 +/-2,5 minutes, cABR: 2,5 +/-1,6 minutes). We conclude that the MSSR technique show significant advantages for the automated detection of near threshold responses over cABR and further improved could be valuable within a screening context.