Neural Network Based Response Prediction of rTMS in Major Depressive Disorder Using QEEG Cordance

Abstract

ObjectiveThe combination of repetitive transcranial magnetic stimulation (rTMS), a non-pharmacological form of therapy for treating major depressive disorder (MDD), and electroencephalogram (EEG) is a valuable tool for investigating the functional connectivity in the brain. This study aims to explore whether pre-treating frontal quantitative EEG (QEEG) cordance is associated with response to rTMS treatment among MDD patients by using an artificial intelligence approach, artificial neural network (ANN).MethodsThe artificial neural network using pre-treatment cordance of frontal QEEG classification was carried out to identify responder or non-responder to rTMS treatment among 55 MDD subjects. The classification performance was evaluated using k-fold cross-validation.ResultsThe ANN classification identified responders to rTMS treatment with a sensitivity of 93.33%, and its overall accuracy reached to 89.09%. Area under Receiver Operating Characteristic (ROC) curve (AUC) value for responder detection using 6, 8 and 10 fold cross validation were 0.917, 0.823 and 0.894 respectively.ConclusionPotential utility of ANN approach method can be used as a clinical tool in administering rTMS therapy to a targeted group of subjects suffering from MDD. This methodology is more potentially useful to the clinician as prediction is possible using EEG data collected before this treatment process is initiated. It is worth using feature selection algorithms to raise the sensitivity and accuracy values.

Neurobiological Predictors of Response to Dorsolateral Prefrontal Cortex Repetitive Transcranial Magnetic Stimulation in Depression: A Systematic Review

“…[22] Further, one study demonstrated that by applying the artificial neural network machine learning method, frontal slow bands (delta and theta) cordance at baseline could be a predictor of response to rTMS with approximately 90% overall accuracy. [59] However, one study concluded that these reported quantitative EEG (QEEG) variables of alpha, beta, and theta power were not significantly correlated with clinical improvement as measured by HAM-D. [20] Lastly, intact Baeken et al [18] No stats 2 HF-left DLPFC E2/P ratio Significant positive correlation found between E2/P ratio and percentage of HAM-D reduction in female patients…”

“…The artificial neural network (ANN) classification method for slow bands at baseline using frontal pretreatment cordance could detect responders to rTMS treatment with approximately 90% overall classification accuracy Erguzel et al [59] Area under ROC curve values for responder detection using 6-, 8-, and 10-fold cross-validation were 0.917, 0.823, and 0.894, respectively 2 H F -l e f t DLPFC…”

Early symptom improvement at 10 sessions as a predictor of rTMS treatment outcome in major depression

“…For example, group-level predictive correlations to outcome have been reported for serotonin transporter and brain-derived neurotrophic factor genetic polymorphisms [15,16], biochemical measures such as cortisol [17,18], baseline regional cerebral perfusion or metabolism on PET and SPECT [19e24], resting-state functional MRI [8,25,26], quantitative EEG metrics [27], and electrophysiological measures such as motor evoked potential amplitude, cortical silent period, or short-interval cortical inhibition [28]. A smaller number of studies have actually reported individual-level outcome prediction accuracies >80% for specific biomarkers, including quantitative EEG [29,30], and most recently, resting-state functional MRI [31]. Such developments are promising, but measures such as fMRI and TMS-electrophysiology are not readily available outside academic centers, limiting clinical utility.…”

Non-linear Entropy Analysis in EEG to Predict Treatment Response to Repetitive Transcranial Magnetic Stimulation in Depression

“…Our data consistently show that NR patients have significantly lower entropy values in the prefrontal areas, and particularly the DLPFC region, compared to RP and HC subjects (Figure 3 ). Entropy indicates the complexity in a system, ( Erguzel et al, 2015 ) and is also associated with the amount of “information” the signal carries. In the nervous system, higher levels of entropy have been consistently associated with healthy states where the nervous system is able to respond and adapt to dynamic changes.…”

“…Several frequency-based rsEEG measures have been proposed as predictors of response in TRD in the context of rTMS. Examples include theta (4–7 Hz) activity in the subgenual zone of the anterior cingulate cortex ( Narushima et al, 2010 ), anterior alpha (8–12 Hz) peak frequency ( Arns et al, 2012 ), prefrontal cordance (combination of absolute and relative EEG power at different bands), ( Bares et al, 2015 ; Erguzel et al, 2015 ) and Lempel-Ziv analysis on the alpha band ( Arns et al, 2014 ). However, these frequency-based methods are susceptible to artifacts and are more suitable for the analysis of stationary signals.…”