Analysis of Brain Stress in Response to Temperature Changes under Agricultural Work Using Electroencephalogram Measurement

Abstract

Agricultural workers suffer from various physical problems and mental stress, including depression and insomnia. Various factors affect brain stress, including harsh agricultural working conditions. Further brain stress may also be due to changes in working conditions such as temperature, noise, and vibration. This study aimed to determine the brain stress in response to ambient temperature changes under agricultural work using agricultural machinery. The electroencephalograms (EEGs) of 17 agricultural workers who work using various agricultural machines was measured and analyzed. The EEG was measured for 3 min at the inactive and active state of agricultural work, respectively, at 24 °C, 28 °C, and 32 °C. The EEG was analyzed using the stress indices such as spectral edge frequency 95% (SEF95%), relative gamma power (RGP), and EEG-based working index (EWI). The EEG analysis indicated that brain stress was generated when the subjects performed the agricultural work after an inactive state. Additionally, as the temperature increased to ≥24 °C, the brain regions where SEF95%, RGP, and EWI exhibited an increase were identified. The findings of this study can be used as basic data in determining the working stress in agricultural workers during work as the ambient temperature changes from 24 °C to 32 °C.

1. Introduction

Agriculture is categorized as one of the three most dangerous global industries, causing the 12th highest level of stress among 130 jobs with a high risk of stress [1,2]. Agricultural workers suffer not only from physical problems, such as musculoskeletal disease and addiction to therapeutic drugs, but also mental stress, including depression and insomnia [3,4]. In countries such as the United States and South Korea and countries in Europe, the stress of agricultural workers reduces work efficiency and crop productivity; thus, support is provided for policies and related research to address these problems [5,6,7,8]. In order for these policies to be implemented effectively, a method is needed to objectively evaluate the stress felt by farmers and express it in a quantitative numerical value.

Various factors affect brain stress, from harsh working conditions to physical injuries and accidents related to agricultural machines [2,9]. The common methods to assess brain stress associated with these factors are surveys and monitoring [10,11]. As these methods are qualitatively influenced by the subjective opinions of the subjects or observers, there is a limitation in identifying accurate and objective levels of brain stress [12]. In fields other than agriculture, quantitative studies have been conducted to assess the stress via biodata measurements, such as the brainwaves of workers via electroencephalogram (EEG), heart rate via electrocardiogram (ECG), blood pressure, pupil size via pupillometry, and hormones [13].

In the field of agriculture, studies on musculoskeletal disease, as a type of physical injury, have quantitatively assessed the agricultural work stress according to changes in workload by using an angular velocity sensor or ECG device attached to the bodies of agricultural workers [12,14]. However, there is no study, to our knowledge, that quantitatively assesses brain stress according to changes in working conditions, such as temperature, noise, and vibration, by analyzing the biodata.

A characteristic of agricultural work is that workers are often in outdoor fields or greenhouses in summer, which leads to frequent heat exhaustion or heatstroke owing to the long hours of exposure to high temperatures [15]. Hence, there is a need for quantitative studies to assess brain stress with respect to environmental stress factors, such as temperature. The following studies in fields other than agriculture have analyzed the stress level according to temperature using brainwave measurements. Lee et al. (2012) performed an EEG analysis to identify a suitable indoor air-conditioning temperature in the field of architecture [16]. EEG measurements were taken at three regions of the frontal lobe, and the level of attention was analyzed by comparing the theta waves, sensorimotor rhythm (SMR), and mid-beta waves at 20.5 °C and 24 °C. Kim et al. (2014) assessed the level of attention by extending the temperature range to 18 °C–31 °C [17]. Based on the level at 25 °C, a decrease in attention was observed with a rise or fall in temperature. Han and Chun (2019) analyzed the EEG data regarding the theta, beta, and gamma waves in the frontal and temporal lobes by extending the scope of brain regions to take EEG measurements [18]. This allowed the detection of increased thermal displeasure in the occupants as the temperature increased from 25 °C to 32 °C. Kim et al. (2014) analyzed the variation of SMR according to changes in light intensity in addition to changes in temperature, based on the prefrontal lobe [19]. Choi et al. (2014) conducted a study to analyze the stress in subjects by inducing complex changes in factors such as indoor temperature, noise, and malodor, and suggested that the stress caused by environmental changes could be assessed based on the measurements of high-beta waves [20].

In terms of brain stress suffered by agricultural workers, differences have been found in comparison to the occupational stress related to other jobs [21]. This is because of the differences in the main cause of each type of stress. Furthermore, in the field of agriculture, most of the factors that influence brain stress are complex and associated with outdoor conditions. Thus, it is necessary for a study on brain stress to utilize different methods from those used in previous indoor studies.

If a method is established to analyze and quantitatively evaluate the brain stress felt by farmers, not only can it be used in policies to improve the welfare of farmers such as suggestion of work environment guidelines in response to the temperature, but it can also be used as an objective evaluation index in the ergonomic design of agricultural machinery. As a preliminary study to investigate the effects of various environmental factors on brain stress in the field of agriculture, EEG was measured and analyzed for brain stress in response to temperature changes, targeting farmers who perform agricultural work using agricultural machinery.

2. Materials and Methods

2.1. Experimental Design

2.1.1. Structure and Function of the Brain

The brain is critical with respect to human thoughts and behaviors. Neurons are the basic nerves that constitute the brain, and the recorded waveforms of potential differences that arise as the neurotransmitters are transported across neurons in the cerebral cortex are defined as brainwaves [22].

The brain can be divided into the prefrontal, frontal, temporal, parietal, and occipital lobe. The prefrontal lobe is responsible for mental activities such as reality perception, executive control, problem-solving, and impulse control [23]. Studies have also shown that stress, including depression, anger, and anxiety, activate the right prefrontal lobe [24].

The frontal lobe analyzes the received information to make judgments, playing a role in attention and behavioral control [25]. Compared to the temporal, occipital, and parietal lobes, the activity of the frontal lobe according to emotional changes are high; therefore, this region is most commonly examined when determining stress levels [26].

The parietal lobe integrates the acquired data from the tactile, auditory, and visual senses since it is composed of somatosensory and motor regions. In this way, the parietal lobe perceives the position of each body part and orchestrates the movements [27]. The temporal lobe receives the auditory information, and the occipital lobe receives and interprets the visual information [28]. The left hemisphere of the brain is activated in the processing of positive emotions, whereas the right hemisphere tends to be activated in the processing of negative emotions [29].

Therefore, an EEG analysis with the electrodes attached to the prefrontal, frontal, parietal, temporal, and occipital lobes constituting the cerebral cortex is presumed to allow the identification of the regions of the cerebral cortex that are activated in response to the ambient temperature changes, as well as the brain stress experienced by subjects.

2.1.2. EEG Measurement Test Method

To determine the brain stress in response to the ambient temperature changes, an EEG analysis was performed. The study included 17 adult agricultural workers (males, 20–30 ages: 2 people; 30–40 ages: 5 people; 40–50 ages: 2 people; 50–60 ages: 4 people; 60–70 ages 3 people; 70–80 ages: 1 person) whose work involves various agricultural machines. All subjects refrained from alcohol consumption, smoking, and drugs the day before the analysis. Agricultural workers with a history of neuropsychological disease were excluded. Before the EEG measurement experiment, the subjects were informed about the experiment method and possible risks. This study was approved by the Ethics Committee of Kangwon National University Institutional Review Board, and all subjects provided informed consent.

The EEG measurements were taken three times for 3 min each in an inactive and active state of agricultural work. The measurement time was set considering the recommendations of the EEG measurement device manufacturer. The active state was defined as when the subject was performing actual agricultural work by manipulating various agricultural machines, such as a tractor, combine harvester, fertilizer distributor, or rice and vegetable transplanter. The subject in the inactive state was guided to sit on a chair, not on the agricultural machine, to rest. The ambient temperature conditions were set based on the average temperature range during agricultural work in rural areas and recommended temperature range for agricultural work, as reported in previous studies [30,31,32]. As a result, three temperature conditions were set as follows: 24 °C, 28 °C, and 32 °C. Through EEG measurement test at the active state, it was confirmed whether the stress felt by subjects increased when the temperature rose based on the 24 °C, which is the highest level of thermal comfort [32]. All subjects wore identical clothes. A demonstration of the EEG measurement test is shown in Figure 1.

2.1.3. EEG Measurement Device

The image and data of the device used in EEG measurements (Quick-20r/CGX/USA) are presented in Figure 2 and

Table 1. Specifications of the EEG measurement device.

Items	Specification
Model/Company/Nation	Quick-20r/CGX/USA
Length × Width × Height (mm)	200 × 180 × 190
Weight (g)	596
The number of electrodes	20
Sampling rate (samples/s)	500
A/D Resolution (bit)	24
Bandwidth (Hz)	0 to 131
Wireless range (mm)	10,000

[33,34]. As an EEG device allows for a non-invasive method of EEG measurements in subjects, there is no need to insert an apparatus into the body or apply substances such as a gel to the skin; therefore, it does not cause unnecessary discomfort to the subjects. The main components of the EEG device are electrodes, an analog-to-digital converter (ADC), and an amplifier [34]. The electrodes are attached to the scalp to measure the potential differences arising from brain activities. The measured values of the EEG are converted to electrical signals (μV) and amplified via the ADC and the amplifier, then wirelessly transmitted to software for brainwave analysis. Twenty electrodes were used in this study for EEG measurements (Figure 3). Among them, 18 were attached to the prefrontal lobe (Fp1 and Fp2), frontal lobe (F3, F4, F7, F8, and Fz), temporal lobe (T3, T4, T5, and T6), parietal lobe (P3, P4, Pz, C3, and C4), and occipital lobe (O1 and O2), following the international 10–20 system [35]; the reference electrode was attached to the right earlobe (A2) and the ground electrode was attached to the left earlobe (A1) in consideration of the ipsilateral ear reference [22,36]. The EEG signal was measured by setting the sampling rate to 500 Hz. The image and data of the thermometer (Testo-625/TESTO/Germany) used in this study are presented in Figure 4 and

Table 2. Specifications of the thermometer.

Items		Specification
Model/Company/Nation		Testo-625/TESTO/Germany
Length × Width × Height (mm)		182 × 64 × 40
Measurement range	Min.	−10
	Max.	60
Resolution (°C)		0.1
Accuracy (°C)		±0.5

.

2.2. EEG Data Analysis

In the EEG analysis, the EEG measurements were analyzed using Bioteck Analysis (Version 1.0, CGX, San Diego, CA, USA). To minimize the influence of electro-oculogram and electromyogram, subjects were asked to close their eyes (inactive state) and mouth (inactive/active state). The EEG is measured in the form of complex waves; hence, a process to classify the signals by the bandwidth through the Fast Fourier Transform (FFT) is necessary prior to the EEG analysis [37]. The classified bandwidths are 0.5–4 Hz (delta wave), 4–8 Hz (theta wave), 8–13 Hz (alpha wave), 13–30 Hz (beta wave), and 30–50 Hz (gamma wave). Each class of bandwidth is characterized as follows: delta waves are activated in a deep sleep or subconscious state [38]; theta waves grow stronger in a state of shallow sleep or deep rest and tend to increase during creative thinking and problem-solving [39,40]; alpha waves increase in a state of rest with a characteristic decline upon a specific stimulus [41]; beta waves increase in a state of tension or concentration on a perceived task, in response to an external stimulus, or in a decision-making process [42,43]; and gamma waves are mainly generated during the performance of higher cognitive functions, such as the detection of information, perception, and integration, as well as upon mental stress caused by excitement, intense anxiety, and high cognitive load [44].

The power spectrum analysis is a representative method of EEG analysis [45]. The level of electrical power can be calculated for each bandwidth classified through the FFT. The proportion of the power spectrum of a given frequency against the overall power spectrum can be estimated subsequently. In this way, the power spectrums of delta, theta, alpha, beta, and gamma waves against the overall power spectrum, which fall in the range of 0.5–50 Hz, can be compared to estimate and evaluate the levels of brain stress across the subjects [28]. A higher intensity of the power spectrums of beta and gamma waves indicates higher brain stress.

The indices used to express the stress felt by the subject in the active state were the spectral edge frequency (SEF)95%, relative gamma power (RGP), and EEG-based working index (EWI). The levels of these indices are proportional to the intensity of brain stress felt by the subject.

The SEF95% is a stress index based on the boundary frequency at 95% power spectral density within the range of 0.5–50 Hz; i.e., a higher level of SEF95% indicates a higher level of brain stress in the subject [46].

The RGP is an index representing the proportion of the power spectrum of gamma waves in the overall power spectrum; i.e., a higher level of RGP indicates a higher level of brain stress in the subject. The RGP is determined by Equation (1) [44].

$$RGP=\frac{{PS}_{\gamma }}{{PS}_{\delta }+{PS}_{\theta }+{PS}_{\alpha }+{PS}_{\beta }+{PS}_{\gamma }}$$

(1)

where,

RGP = relative gamma power

PS_δ = power spectrum of delta waves, μV²

PS_θ = power spectrum of theta waves, μV²

PS_α = power spectrum of alpha waves, μV²

PS_β = power spectrum of beta waves, μV²

PS_γ = power spectrum of gamma waves, μV²

The EWI is an index representing the sum of the power spectrums of beta and gamma waves against the sum of the power spectrums of theta and alpha waves. Through the EWI, the theta, alpha, beta, and gamma activities can be deduced. In sum, a higher level of EWI indicates a higher level of brain stress in the subject. The EWI is determined by Equation (2) [47].

$$EWI=\frac{{PS}_{\beta }+{PS}_{\gamma }}{{PS}_{\theta }+{PS}_{\alpha }}$$

(2)

A t-test was performed to evaluate whether there were significant differences in SEF95%, RGP, and EWI values when the subjects performed agricultural work after an inactive state. The subordination variables were brain stress indices (SEF95%, RGP, and EWI values) derived through EEG measured from 18 electrodes, excluding the reference electrode and ground electrode, and independent variables were inactive and active state. Through this statistical analysis, it was determined whether the subjects felt brain stress when performing the farm work.

In addition, one-way analysis of variance (ANOVA) was performed to evaluate whether the SEF95%, RGP, and EWI values showed significant differences in response to the ambient temperature changes in the active state.

Subordination variables were set to SEF95%, RGP, and EWI values, and independent variables were set to three levels of temperature (24 °C, 28 °C, and 32 °C). For the statistical analysis, the Statistical Package for the Social Sciences software (SPSS, Version 26, IBM Corp., Armonk, NY, USA) was used.

In order to visualize the change in brain stress according to the presence or absence of work and temperature rises, the SEF95%, RGP, and EWI values of each subject were expressed as time series data and averaged. Then, the calculated levels of stress were expressed as Z-scores ranging from 0–1 to increase the visual effects of the graphs [22,48]. Additionally, the moving average filter on the graphs was estimated for 20 samples. The process of EEG analysis is shown in Figure 5.

3. Results

3.1. Results of EEG Analysis According to the Inactive and Active States

3.1.1. SEF95% Values According to the Inactive and Active States

A t-test was performed to confirm whether there were significant differences in the levels of SEF95% when the subjects performed agricultural work after inactive state. The results showed that the SEF95% values in the Fp2 location of the prefrontal lobe, the F4 and F8 locations of the frontal lobe, and the C4 location of the parietal lobe of the 18 electrode locations were significantly different (p < 0.05) according to the presence or absence of work (

Table 3. T-test results based on SEF95% at each electrode in the inactive and active states (p < 0.05).

EEG Electrode Position		State	N	Mean	Standard Deviation	t (p)
Prefrontal lobe	Fp2	Inactive	51	37.014	3.742	−2.386 (0.019)
		Active	51	39.155	5.203
Frontal lobe	F4	Inactive	51	38.619	4.705	2.726 (0.008)
		Active	51	35.605	6.343
	F8	Inactive	51	35.842	4.635	−5.066 (0.000)
		Active	51	39.915	3.388
Parietal lobe	C4	Inactive	51	39.510	5.156	2.467 (0.015)
		Active	51	36.826	5.815

).

The SEF95% value measured at the location of the Fp2 increased by 5.78% in the active state compared to the inactive state. Furthermore, the SEF95% value at the location of the F8 increased by 11.36% as the activity changed from the inactive state to the active state. This confirmed that brain stress in the subjects was generated in the active state. Therefore, the prefrontal lobe (Fp2) and frontal lobe (F8) were presumably stimulated due to the activation of the EEG in the high bandwidth range at 0.5–50 Hz in the subjects performing agricultural work after an inactive state.

On the other hand, the SEF95% values at the F4 location of the frontal lobe and C4 location of the parietal lobe in the active state tended to decrease by 7.8% and 6.79%, respectively, compared to the inactive state. The SEF95% values derived from other electrodes did not show a significant difference according to the presence or absence of work at the 5% significance level. Taken together, these results confirmed the occurrence of brain stress through the analysis of the EEG measured in the right prefrontal lobe (Fp2) and right frontal lobe (F8). Figure 6 shows the change in SEF95% values as time series data in the right prefrontal lobe (Fp2) and right frontal lobe (F8) when subjects engaged in agricultural work after inactive state. Three-dimensional (3D) mapping for SEF95% according to the inactive and active states are shown in Figure 7.

3.1.2. RGP Values According to the Inactive and Active States

A t-test was performed to confirm whether there were significant differences in the levels of RGP when the subjects performed agricultural work after inactive state. The results showed that the RGP values in the Fp2 location of the prefrontal lobe, the F4 and F8 locations of the frontal lobe, and the C4 location of the parietal lobe of the 18 electrode locations were significantly different (p < 0.05 for all) according to the presence or absence of work (

Table 4. T-test results based on RGP at each electrode in the inactive and active states (p < 0.05).

EEG Electrode Position		State	N	Mean	Standard Deviation	t (p)
Prefrontal lobe	Fp2	Inactive	51	0.100	0.037	−3.186 (0.002)
		Active	51	0.127	0.048
Frontal lobe	F4	Inactive	51	0.112	0.038	2.536 (0.013)
		Active	51	0.090	0.048
	F8	Inactive	51	0.093	0.047	−2.209 (0.029)
		Active	51	0.113	0.048
Parietal lobe	C4	Inactive	51	0.127	0.052	2.067 (0.041)
		Active	51	0.104	0.061

).

The RGP value measured at the location of the Fp2 increased by 27% in the active state compared to the inactive state. In addition, the RGP value at the location of the F8 increased by 21.51% as it changed from the inactive state to the active state. This confirmed that brain stress in the subjects was generated in the active state. Therefore, the prefrontal lobe (Fp2) and frontal lobe (F8) were presumably stimulated due to the activation of gamma waves compared to delta, theta, alpha, and beta waves in the subjects performing agricultural work after an inactive state.

On the other hand, the RGP values at the F4 location of the frontal lobe and C4 location of the parietal lobe in the active state tended to decrease by 19.64% and 18.11%, respectively, compared to the inactive state. The RGP values derived from other electrodes did not show a significant difference according to the presence or absence of work at the 5% significance level. These results confirmed the occurrence of brain stress through the analysis of the EEG measured in the right prefrontal lobe (Fp2) and the right frontal lobe (F8). Figure 8 shows the change of RGP values as time series data in the right prefrontal lobe (Fp2) and right frontal lobe (F8) when subjects engaged in agricultural work after an inactive state. Three-dimensional mapping for RGP according to the inactive and active states is shown in Figure 9.

3.1.3. EWI Values According to the Inactive and Active States

A t-test was performed to confirm whether there were significant differences in the levels of EWI when the subjects performed agricultural work after inactive state. The results showed that the EWI values in the P3, P4, and Pz locations of the parietal lobe, T5 and T6 locations of the temporal lobe, and O1 and O2 locations of the occipital lobe of the 18 electrode locations were significantly different (p < 0.05 for all) according to the presence or absence of work (

Table 5. T-test results based on EWI at each electrode in the inactive and active states (p < 0.05).

EEG Electrode Position		State	N	Mean	Standard Deviation	t (p)
Parietal lobe	P3	Inactive	51	1.273	0.429	−3.459 (0.001)
		Active	51	1.637	0.617
	P4	Inactive	51	1.212	0.447	−3.556 (0.001)
		Active	51	1.585	0.602
	Pz	Inactive	51	1.132	0.341	−2.993 (0.003)
		Active	51	1.446	0.667
Temporal lobe	T5	Inactive	51	1.272	0.495	−3.054 (0.003)
		Active	51	1.740	0.975
	T6	Inactive	51	1.187	0.424	−2.086 (0.040)
		Active	51	1.444	0.769
Occipital lobe	O1	Inactive	51	1.141	0.430	−2.919 (0.004)
		Active	51	1.464	0.662
	O2	Inactive	51	1.138	0.425	−2.712 (0.008)
		Active	51	1.445	0.687

). This is presumably due to the higher theta and alpha activities than the beta and gamma activities in the subjects in the inactive state, whereas the beta and gamma activities increased and the theta and alpha activities decreased as agricultural work was performed.

The EWI values of the subjects measured at the T5 location of the temporal lobe increased by 36.79% in the active state compared to the inactive state, showing the highest increase rate among the electrodes that showed a significant difference. Furthermore, the increase rates of EWI values from largest to smallest were found to be P4, P3, O1, Pz, O2, and T6, and the increase rates were 30.78%, 28.59%, 28.3%, 27.74%, 26.97%, and 21.65%, respectively. The EWI values derived from other electrodes did not show a significant difference according to the presence or absence of work at the 5% significance level.

These results confirmed the occurrence of brain stress through the analysis of the EEG measured in the parietal lobe (P3, P4, and Pz), temporal lobe (T5 and T6), and occipital lobe (O1 and O2). Figure 10 shows the change of EWI values as time series data in the parietal lobe (P3, P4, and Pz), temporal lobe (T5 and T6), and occipital lobe (O1 and O2) during the active periods. Three-dimensional mapping for the relative theta power (RTP), relative alpha power (RAP), and relative beta power (RBP) according to the inactive and active states are shown in Figure 11.

3.2. Results of EEG Analysis in Response to the Temperature Changes

3.2.1. SEF95% Values in Response to the Temperature Changes

One-way ANOVA was performed to confirm whether there were significant differences in the levels of SEF95% in response to the ambient temperature changes in the active state (

Table 6. ANOVA results based on SEF95% at each electrode in the active state in response to the ambient temperature changes (p < 0.05).

EEG Electrode Position		Sum of Squares	Degree of Freedom	Mean Square	F-Value	p-Value
Prefrontal lobe	Fp2	172.660	2	86.330	3.508	0.038
Frontal lobe	F8	218.352	2	109.176	14.735	0.000
Parietal lobe	P4	125.980	2	62.990	3.417	0.041
Temporal lobe	T4	153.963	2	76.982	4.378	0.018

).

The results showed that SEF95% measured in the Fp2 location of the prefrontal lobe, the F8 location of the frontal lobe, the P4 location of the parietal lobe, and the T4 location of the temporal lobe were significantly different (p < 0.05 for all) in response to temperature changes. As the temperature increased based on 24 °C, the SEF95% value showed a tendency to increase.

As the temperature increased from 24 °C to 32 °C by 4 °C intervals, the SEF95% values measured in the prefrontal lobe (Fp2) increased by 6.83% and 5.04%, respectively. In addition, the frontal lobe (F8) increased by 7.63% and 5.51%, the parietal lobe (P4) by 4.3% and 5.4%, and the temporal lobe (T4) by 5.49% and 5.09%, respectively (

Table 7. SEF95% values in response to ambient temperature changes in the active state.

EEG Electrode Position			Ambient Temperature (°C)
			24	28	32
Prefrontal lobe	Fp2	Average	36.818	39.333	41.315
		S.D.	6.267	4.370	3.930
Frontal lobe	F8	Average	37.282	40.125	42.337
		S.D.	3.081	2.732	2.297
Parietal lobe	P4	Average	38.637	40.297	42.475
		S.D.	5.004	5.018	2.253
Temporal lobe	T4	Average	39.178	41.330	43.434
		S.D.	5.278	3.909	3.101

). It was determined that the heat discomfort felt by the subjects increased as the temperature increased from 24 °C to 32 °C by 4 °C intervals and the high-frequency bandwidths of EEG were relatively activated [32]. The prefrontal lobe (Fp2) is activated upon stress [27], and the frontal lobe (F8) is a significant role in the processing of emotion [25,29]. Considering that the SEF95% value of the right parietal lobe (P4) and temporal lobe (T4) did not show a significant difference (p < 0.05) in response to the presence or absence of work, as shown in Table 3, the brain stress did not occur in the right parietal lobe (P4) and temporal lobe (T4) based on SEF95% value. Therefore, it was determined that the right prefrontal lobe (Fp2) and right frontal lobe (F8) can be effective measurement locations to evaluate brain stress in response to the temperature change. The SEF95% values derived from other electrodes did not show a significant difference in response to the temperature changes at the 5% significance level. Figure 12 shows the change of SEF95% values as time series data in the prefrontal lobe (Fp2), frontal lobe (F8), parietal lobe (P4), and temporal lobe (T4) during temperature changes. Three-dimensional mapping for SEF95% in response to the temperature changes is shown in Figure 13.

3.2.2. RGP Values in Response to the Temperature Changes

One-way ANOVA was performed to confirm whether there were significant differences in the levels of RGP in response to the ambient temperature changes in the active state (

Table 8. ANOVA results based on RGP at each electrode in the active state in response to the ambient temperature changes (p < 0.05).

EEG Electrode Position		Sum of Squares	Degree of Freedom	Mean Square	F-Value	p-Value
Prefrontal lobe	Fp2	0.020	2	0.010	5.052	0.010
Frontal lobe	F8	0.039	2	0.020	12.503	0.000
Temporal lobe	T4	0.044	2	0.022	4.934	0.011

).

The results showed that the RGP measured in the Fp2 location of the prefrontal lobe, the F8 location of the frontal lobe, and the T4 location of the temporal lobe were significantly different (p < 0.05 for all) in response to temperature changes. As the temperature increased based on 24 °C, the RGP value showed a tendency to increase.

As the temperature increased from 24 °C to 32 °C by 4 °C intervals, the RGP values measured in the prefrontal lobe (Fp2) increased by 19.84% and 22.42%, respectively. In addition, the frontal lobe (F8) increased by 44.49% and 28.78% and the temporal lobe (T4) by 20.98% and 29.8%, respectively (

Table 9. RGP values in response to ambient temperature changes in the active state.

EEG Electrode Position			Ambient Temperature (°C)
			24	28	32
Prefrontal lobe	Fp2	Average	0.104	0.124	0.152
		S.D.	0.040	0.038	0.054
Frontal lobe	F8	Average	0.079	0.114	0.147
		S.D.	0.036	0.039	0.044
Temporal lobe	T4	Average	0.125	0.152	0.197
		S.D.	0.060	0.071	0.070

). It was determined that the heat discomfort felt by the subjects increased as the temperature increased from 24 °C to 32 °C by 4 °C intervals and the gamma waves of EEG were relatively activated, compared to delta, theta, alpha, and beta waves. The prefrontal lobe (Fp2) is activated upon stress [27], and the frontal lobe (F8) is a significant role in the processing of emotion [25,29]. Considering that the RGP value of the temporal lobe (T4) did not show a significant difference (p < 0.05) in response to the presence or absence of work, as shown in Table 4, the brain stress did not occur in the temporal lobe (T4) based on RGP value. Therefore, it was determined that the right prefrontal lobe (Fp2) and right frontal lobe (F8) can be effective measurement locations to evaluate brain stress in response to the temperature change. The RGP values derived from other electrodes did not show a significant difference in response to the temperature changes at the 5% significance level. Figure 14 shows the change of RGP values as time series data in the prefrontal lobe (Fp2), frontal lobe (F8), and temporal lobe (T4) when the temperature changed. Three-dimensional mapping for RGP in response to the temperature changes is shown in Figure 15.

3.2.3. EWI Values in Response to the Temperature Changes

One-way ANOVA was performed to confirm whether there were significant differences in the levels of EWI in response to the ambient temperature changes in the active state (

Table 10. ANOVA results based on EWI at each electrode in the acting state in response to ambient temperature changes (p < 0.05).

EEG Electrode Position		Sum of Squares	Degree of Freedom	Mean Square	F-Value	p-Value
Frontal lobe	F8	3.644	2	1.822	10.542	0.000
Temporal lobe	T4	5.389	2	2.694	4.965	0.011

).

The results revealed that the EWI measured in the F8 location of the frontal lobe and the T4 location of the temporal lobe were significantly different (p < 0.05 for both) in response to temperature changes. As the temperature increased based on 24 °C, the EWI value showed a tendency to increase.

As the temperature increased by 4 °C intervals from 24 °C to 32 °C, the EWI values measured in the frontal lobe (F8) increased by 31.02% and 24.31%, respectively (

Table 11. EWI values in response to ambient temperature changes in the active state.

EEG Electrode Position			Ambient Temperature (°C)
			24	28	32
Frontal lobe	F8	Average	1.041	1.364	1.696
		S.D.	0.413	0.334	0.486
Temporal lobe	T4	Average	1.444	1.839	2.240
		S.D.	0.635	0.707	0.851

). The power spectrum of gamma waves (24 °C: 0.079, 28 °C: 0.116, and 32 °C: 0.142) and beta waves (24 °C: 0.102, 28 °C: 0.133, and 32 °C: 0.151) increased as the temperature increased (Figure 16). The power spectrum of theta waves (24 °C: 0.107, 28 °C: 0.113, and 32 °C: 0.103) and alpha waves (24 °C: 0.067, 28 °C: 0.070, and 32 °C: 0.070) decreased or was maintained as the temperature increased from 28 °C to 32 °C.

In the case of the temporal lobe (T4), the EWI values increased to 27.34% and 21.82, respectively. The power spectrum of gamma waves (24 °C: 0.125, 28 °C: 0.152, and 32 °C: 0.197) and beta waves (24 °C: 0.147, 28 °C: 0.160, and 32 °C: 0.201) increased as the temperature increased. However, the power spectrum of theta waves (24 °C: 0.108, 28 °C: 0.094, and 32 °C: 0.091) decreased conversely. In addition, the power spectrum of alpha waves (24 °C: 0.080, 28 °C: 0.076, and 32 °C: 0.087) tended to decrease when the temperature increased from 24 °C to 28 °C.

It was determined that that the heat discomfort felt by the subjects increased as the temperature increased from 24 °C to 32 °C by 4 °C intervals, and the activity of beta and gamma waves gradually increased, while the activity of theta and alpha waves decreased or was maintained. Beta waves increase in a state of tension or stress from environmental changes [20,42,43], and gamma waves are mainly generated by mental stress caused by excitement, intense anxiety, and high cognitive load [44]. The EWI values derived from other electrodes did not show a significant difference in response to the temperature changes at the 5% significance level. Figure 17 shows the change of EWI values as time series data in the frontal lobe (F8) and temporal lobe (T4) when the temperature changed. Three-dimensional mapping for EWI in response to the temperature changes is shown in Figure 18.

4. Discussion

This study measured and analyzed the EEG of 17 adult male agricultural workers (20–80 years old) who use various agricultural machines in performing agricultural work to analyze brain stress in response to the ambient temperature changes.

The results indicated not only that brain stress occurred in the active state after inactive state, but also that the cerebral regions where SEF95%, RGP, and EWI increased as the temperature rose were the Fp2 location of the prefrontal lobe and the F8 location of the frontal lobe. These locations were in the right hemisphere. This is presumed to be because the regions of the cerebral cortex in the right hemisphere tend to be activated when processing negative emotions [29].

Yoon et al. (2005) reported that exposure to a stress factor could cause a physical reaction with the involvement of various brain regions, including the prefrontal lobe, forebrain, hypothalamus, septo-hippocampal system, and amygdala [49]. The frontal lobe analyzes and makes judgments on information received from the external environment, with a significant role in the processing of emotion [25,29,50,51], and the prefrontal lobe is responsible for executive control and problem-solving and is activated upon stress [27]. Workers who use agricultural machines are exposed to high temperatures of ≥24 °C [52]. Therefore, it is judged that as the temperature rose above 24 °C, the stress felt by subjects increased, and EEG in a high-frequency bandwidth at the Fp2 location of the prefrontal lobe was activated. It is also possible that brain stress is generated in the process of studying and responding to external stimuli in order to operate the machines. In addition, the increased level of SEF95% with increased temperature in our study could be attributed to increased displeasure felt by the subjects as the temperature increased above 24 °C, in reference to a previous study, where the reported threshold of thermal comfort was 24 °C [32].

5. Conclusions

The findings of this study may provide basic data in determining the brain stress in agricultural workers during activity through EEG analyses and assessing the variations in brain stress in the temperature range of 24 °C–32 °C. Furthermore, the use of SEF95% and RGP as an index of brain stress demonstrated that the right prefrontal lobe (Fp2) and the right frontal lobe (F8) could be the most appropriate brain regions for effective measurements. While the stress factor in this study was limited to temperature range of 24 °C–32 °C. Therefore, the temperature range will need to increase in a follow-up study. In addition, more varied factors of brain stress, such as noise and vibration, will be examined to conduct a more comprehensive analysis of brain stress. Also, in order to improve the reliability of the future research, it is judged that the experiment should be conducted by increasing the number of subjects with various genders, ages, heights, weights, etc.