Pitch Processing Can Indicate Cognitive Alterations in Chronic Liver Disease: An fNIRS Study

Abstract

Early detection and evaluation of cognitive alteration in chronic liver disease is important for predicting the subsequent development of hepatic encephalopathy. While visuomotor tasks have been rigorously employed for cognitive evaluation in chronic liver disease, there is a paucity of auditory processing task. Here we focused on auditory perception and examined behavioral and haemodynamic responses to a melodic contour identification task (CIT) to compare cognitive abilities in patients with chronic liver disease (CLD, N = 30) and healthy controls (N = 25). Further, we used support vector machines to examine the optimal combination of channels of functional near-infrared spectroscopy that can classify cognitive alterations in CLD. Behavioral findings showed that CIT performance was significantly worse in the patient group and CIT significantly correlated with neurocognitive evaluation (i.e., number connection test, digit span test). The findings indicated that CIT can measure auditory cognitive capacity and its difference existing between patient group and healthy controls. Additionally, optimal subsets classified the 16-dimensional haemodynamic data with 78.35% classification accuracy, yielding markers of cognitive alterations in the prefrontal regions (CH6, CH7, CH10, CH13, CH14, and CH16). The results confirmed the potential use of behavioral as well as haemodynamic responses to music perception as an alternative or supplementary method for evaluating cognitive alterations in chronic liver disease.

Keywords: chronic liver disease (CLD); cognitive alteration; functional near-infrared spectroscopy (fNIRS); haemodynamic response; melodic contour identification; nonverbal auditory perception/music perception.

Figure 1

Sample melodic contours. (A) Ascending and stationary contours, (B) ascending and descending contours.

Figure 2

An example of an answer page given in a monitor with the musical stimuli (CIT3). The boxes were presented prior to presentation of each contour.

Figure 3

Spectratech OEG-16. The locations of the 16 fNIRS channels along the frontal cortex. The center of the measurement unit was placed on the frontopolar (Fp) region according to the international 10–20 system.

Figure 4

The flow of the experiment.

Figure 5

SVM classification.

Figure 6

Classification performance for the bilateral hemisphere between CL and HC. The upper panel (A) represents classification performance for the best 50 out of 65,565 subsets. The lower panel (B) shows the weight vector of each channel matched in the subsets. MCC, Matthew Correlation Coefficient. Red indicates higher HbO₂ in the CL, while blue indicates higher HbO₂ in the HC.

Figure 7

Differences of haemodynamic responses between the HC and CL groups.