Domain adaptation for robust workload level alignment between sessions and subjects using fNIRS

Abstract

We demonstrated the potential of using domain adaptation on functional Near-Infrared Spectroscopy (fNIRS) data to detect and discriminate different levels of n-back tasks that involve working memory across different experiment sessions and subjects. Aim: To address the domain shift in fNIRS data across sessions and subjects for task label alignment, we exploited two domain adaptation approaches -Gromov-Wasserstein (G-W) and Fused Gromov-Wasserstein (FG-W). Approach: We applied G-W for session-by-session alignment and FG-W for subject-by-subject alignment with Hellinger distance as underlying metric to fNIRS data acquired during different n-back task levels. We also compared with a supervised method -Convolutional Neural Network (CNN). Results: For session-by-session alignment, using G-W resulted in alignment accuracy of 70 ± 4 % (weighted mean ± standard error), whereas using CNN resulted in classification accuracy of 58 ± 5 % across five subjects. For subject-by-subject alignment, using FG-W resulted in alignment accuracy of 55 ± 3 %, whereas using CNN resulted in classification accuracy of 45 ± 1 %. Where in both cases 25 % represents chance. We also showed that removal of motion artifacts from the fNIRS data plays an important role in improving alignment performance. Conclusions: Domain adaptation is potential for session-by-session and subject-by-subject alignment using fNIRS data.

Four-Class Classification of Neuropsychiatric Disorders by Use of Functional Near-Infrared Spectroscopy Derived Biomarkers

“…The major advantages of these techniques rely on their ability to capture the complexity of neural information embedded in the HBO signal patterns through optimization of the network structures [ 81 ]. Indeed, there exists some successfully implemented DL classifiers with fNIRS and EEG signals [ 82 , 83 , 84 , 85 , 86 ]. However, we avoided testing the utility of DL algorithms in the presented work because of the limited cohort size of each group.…”

A General and Scalable Vision Framework for Functional Near-Infrared Spectroscopy Classification

“…The CNN [39], CNN-3b [40], RNN [39], fNIRS-T [9], and fNIRS-PreT [9] are used as baseline networks. A threebranch CNN network decodes consumers' preference levels from viewing commercial advertisement videos of different durations (15, 30, and 60 s) [40].…”

Brain–Computer Interfacing Using Functional Near-Infrared Spectroscopy (fNIRS)

“…In a typical fNIRS experimental setup, a predetermined number of optodes, i.e., light sources/emitters and detectors, are placed on the subject's head, covering a specific region or the whole brain, called montage. The choice of the montage is determined by the nature of the experiment and based on the prior knowledge of the activation region in the brain for a given stimulation [12]. For, e.g., experiments involving motor tasks, the pairs are placed and the hemodynamic response is acquired from the motor cortex (Figure 8), whereas for higher cognitive functions or mental imagery, the emitter/detector pairs are placed around the prefrontal cortex (Figure 9).…”

“…This leads to the insertion of implicit assumptions in the characteristics of motion artifacts and subsequently filtered fNIRS signals. Multiple approaches guided by statistical signal processing methods including adaptive filtering, independent component analysis (ICA), and time-frequency analysis have been employed to formulate algorithms that can help fix for motion artifacts in fNIRS signals [12,41,[46][47][48][49][50][51][52][53][54].…”