The study protocol was approved by the Ethics Committee of Nippon Dental University School of Life Dentistry (NDU-T2013-30) and conformed to the guidelines of the Declaration of Helsinki. Informed consent was obtained from all participants before their inclusion in the study. All participants were given an honorarium (approximately 5000 JPY) after completion of the study. Because few studies have reported the cerebral activation pattern due to DS, the sample size was determined in a pilot study. Using the G*power program (ver.3.1.9.2) [12], with an effect size of 1.2, and t tests, we determined that this study required at least 12 participants per group (α = 0.05, β = 0.20).

Thirty-four right-handed Japanese individuals (21 women and 13 men; age, 19–49 years; average age, 31.2 years [standard deviation (SD) = 9.1]) were recruited from the surrounding community. Their mean educational achievement level was 14.1 years (SD = 2.1). All candidates had undergone dental treatment, had normal vision and hearing, and met all magnetic resonance imaging (MRI) inclusion criteria (no cardiac pacemaker, metallic implants, history of vascular surgery, claustrophobia, or tattoo). Candidates were carefully screened by a clinical expert (MK) using a standardized neuropsychiatric interview process [13]. No participant had a history of psychiatric disorders, significant physical illnesses, head injury, neurological disorders, or substance abuse. Routine MRI was performed to rule out anatomic cerebral abnormalities, and no participant took any medication prior to fMRI. All participants were right-handed according to the Edinburgh Handedness Inventory (EHI) [14]. We defined a right-handed participant as one with an EHI score > 50 according to a previous study [15].

The DF level was assessed using the self-reported Dental Fear Survey (DFS) [16]. A Japanese version of the questionnaire was used to verify the validity and reliability of the test [17]. This questionnaire assesses anxiety-provoking situations associated with dental treatment. It consists of 20 questions scored from 1 to 5, summed to give a total score between 20 and 100. The mean score for the Japanese population has been estimated at 37.4 (SD = 14.1) [17]. Based on an overall score > 52 (Japanese mean score + 1 SD), the participants were categorized into the DF and control groups. Twelve participants (nine women and three men; mean age, 32.3 years; SD = 11.2) with scores ranging from 52 to 82 points were included in the DF group. The remaining 22 participants (12 women and 10 men; mean age, 30.6 years; SD = 7.9) with scores < 52 were included in the control group. Depression and dental anxiety levels were evaluated using the Self-Rating Depression Scale (SDS) [18] and Dental Anxiety Scale (DAS), respectively [19].

A passive-listening fMRI experiment was conducted using a previously published protocol [20, 21]. In a single session, two types of stimuli were presented: DS and neutral sounds (NS). DS included the sounds of dental drilling using a high-speed dental handpiece, vacuum suction, dental drilling using a low-speed dental engine, hand scaling, ultrasound scaling, and a saliva ejector. The sounds of the French horn or pure tone (2000 Hz), which are not associated with dental treatment, were used as NS. In each listening session, six sounds of that category were presented for 3 s, and the interval between sounds was 2 s, making the duration of each block 28 s. Before each sound category, no sound was presented from the headphones for 20 s (rest condition). Each set lasted 96 s and consisted of two sound conditions (28 s × 2) and two rest conditions (20 s × 2). In the NS condition, the French horn and pure tone were presented alternately. The two stimuli (DS and NS) were presented pseudo-randomly in each set of experiments. Each session consisted of four sets, with a total scanning time of 6.4 min (96 s × 4). (Fig. 1).

fMRI protocol. A single fMRI session consisted of listening to dental sounds (DS) and neutral sounds (NS). The names indicate the categories of sounds presented. The sequence of the DS–NS or NS–DS blocks was repeated four times, and the duration of each session was 6.4 min

All auditory stimuli were presented using Media Studio Pro (version 6.0, Ulead Systems, Inc., Taiwan), and participants listened to the stimuli through headphones connected to an air conductance sound-delivery system (Commancer X6, MRI Audio System, Resonance Technology Inc., Los Angeles, California, USA) [22]. The average sound pressure at the stimulus amplitude was maintained at 80 dB, which is equivalent to the sound pressure of dental drilling using a high-speed dental handpiece during dental treatment.

After completing the session, the participants rated their feelings during exposure to the sounds (six DS and two NS) on a visual analog scale (VAS; 0–100 mm). The following emotional dimensions were assessed: valence, fear, and pain (valence: 0 [very negative], 50 [neutral], and 100 [very positive]; fear and pain: 0 [not at all] and 100 [very strong]) [23].

Imaging was performed using a Philips 3.0 Tesla MRI system. Functional images of 395 volumes were acquired using T2*-weighted gradient echo-planar imaging (EPI) sequences that are sensitive to blood oxygenation level-dependent (BOLD) contrast. Each volume consisted of 35 transaxial contiguous slices with a slice thickness of 4 mm, covering almost the entire brain (flip angle = 72.5°; echo time = 23 ms; repetition time = 1.6 s; matrix = 52 × 30 × 64; field of view = 208 × 120 × 256) [22].

Data were analyzed using the statistical parametric-mapping software SPM8 (Wellcome Department of Cognitive Neurology, London, United Kingdom), which was run in MATLAB (MathWorks, Natick, Massachusetts, USA). All functional EPI images from each session were realigned to the first volume to correct for the participant’s motion. Images were spatially normalized to the standard space defined by the Montreal Neurological Institute template. In this study, slice thickness was set at 4 mm. For the normalization analysis conducted in SPM8, we utilized a voxel size of 2 × 2 × 2. Following normalization, all scans had a final spatial resolution of 2 × 2 × 2 mm3. Functional images were smoothed using a 3-D isotropic Gaussian kernel with a full width at half maximum of 8 mm. A high-pass filter was applied to the fMRI time series to remove low-frequency noise and enhance the temporal signal-to-noise ratio. Hemodynamic changes during each condition were analyzed using a general linear model combined with boxcar functions convoluted with hemodynamic response functions. Voxel-by-voxel statistical parametric maps were constructed for each t statistic. The t values were transformed into a unit normal distribution to obtain z scores.

Models of the two stimuli (DS and NS) were created using a blocked design for fMRI experiments. First, to investigate the effect of DF on cerebral activation during auditory processing, cerebral activation during the two stimuli was analyzed. Next, to clarify cerebral activation during exposure to DS, cerebral activation in the DS minus NS contrast was examined based on previous studies [20, 21].

Group analysis (2nd-level analysis in SPM8) was performed on the data of 22 participants in the control group and 12 participants in the DF group using a random-effects model on a voxel-by-voxel basis. Two trials (DS and NS) were presented for each explanatory variable. Each explanatory variable was convoluted with a standard hemodynamic response function taken from SPM8 to account for the hemodynamic response lag. First, we analyzed the one-sample t test for cerebral activation of the control and DF groups, respectively. Second, we evaluated the difference in the mean cerebral activation between the DF and control groups using a two-sample t test, and t statistics were calculated to compare the two trials. Additionally, we analyzed the one-sample t test for cerebral activation in all 34 participants. Five regions of interests (ROIs) were identified from the peaked activations in this analysis. Cerebral activation at the ROIs was investigated in each group. For the main effect of task, ROIs were defined as sphere voxels of 10 mm radius from the coordinates of the peak voxel of activation.

Demographic data for the DF and control groups were compared using Student’s t test and Fisher’s exact test. A two-way analysis of variance was used to analyze the subjective ratings between the two groups for DS and NS. Statistical significance was set at P < 0.05, and all analyses were conducted using IBM SPSS Statistics 21 software (IBM Japan, Tokyo, Japan).