A retrospective chart review was performed at Konkuk University Medical Center. From March 2008 to June 2012, a total of 155 stroke patients underwent a video-fluoroscopic swallowing study (VFSS). Only those with middle cerebral artery (MCA) lesions were included in the current study. Exclusion criteria were (1) subarachnoid hemorrhage; (2) previous stroke or swallowing disorders; (3) concomitant lesions in the brainstem or cerebellum; (4) other diseases that lead to dysphagia, such as multiple sclerosis, oropharyngeal structural damage, dementia [12], or neurodegenerative disorders [13,14]; (5) decreased consciousness, poor cooperation for VFSS, or inability to maintain a sitting position for VFSS; (6) contraindication for MRI; or (7) refusal of DTI. The final sample included 31 MCA stroke patients who had undergone DTI within 6 weeks of a stroke. The Korean version of the Western Aphasia Battery [15] was used to detect aphasia, and the line bisection test [16] and Albert test [17] were used to detect hemispatial neglect. The study was approved by the Institutional Review Board at Konkuk University Medical Center (No. KUH1180010).

VFSS, clinical dysphagia scale (CDS), and diet level were assessed on admission. VFSS was performed by an experienced rehabilitation physician, assisted by an occupational therapist and a radiologic technician in the fluoroscopic laboratory. Subjects were seated upright in a wheelchair, and the fluoroscopic X-ray tube was focused in a lateral plane and the placement was verified. Each subject swallowed two 5 mL and then two 10 mL dysphagia diet boluses (mixture of Solotop-HD, barium sulfate powder for suspension). Dysphagia diet level was based on the study by Han et al. [18], which suggested that dysphagia diet should be divided into four classes based on the level of viscosity with reference values of 1,000, 100, and 10 cP by a viscometer and 1, 3, and 4 cm by the line spread test. In the present study, we used the line spread test for dysphagia diet for VFSS. Dysphagia diet levels were defined as follows. Level 1, <1 cm by the line spread test, 3 g of thickener (Toromi-up; Nisshin OilliO, Tokyo, Japan) added to 100 mL of water; level 2, 1–3 cm by the line spread test, 2 g of thickener (Toromi-up) added to 100 mL of water; level 3, 3–4 cm by the line spread test, 1 g of thickener (Toromi-up) added to 100 mL of water; level 4, >4 cm by the line spread test, 0.7 g of thickener (Toromi-up) added to 100 mL of water; level 5, water. Lip closure, bolus formation, mastication, and oral transit time were measured in the oral phase. Normal oral transit time was defined as less than 1 second. In the pharyngeal phase, laryngeal elevation, cricopharyngeal dysfunction, oronasal regurgitation, residues in the valleculum and pyriform sinus, pharyngeal transit time, and aspiration status were observed. Residues in the valleculum and pyriform sinus less than 10% were considered normal; normal pharyngeal transit time was defined as less than 1 second. Aspiration was defined when the bolus passed through the vocal folds. Furthermore, in the esophageal phase, the esophageal relaxation status was examined. The results of the initial examination were then reviewed, and dysphagia was confirmed if abnormal findings were apparent during any one of the oral, pharyngeal, or esophageal phases. Subjects were divided into two groups according to the VFSS findings (dysphagia group and non-dysphagia group).

The CDS (Appendix 1) was used to quantify the clinical severity of dysphagia. The CDS has been confirmed as a quantitative clinical tool that reflects the VFSS findings well, and it can be adapted to various patients with dysphagia irrespective of the causal disorders [19].

All imaging was performed on a 3.0-Tesla MRI system (Signa HDxt; GE Medical System, Milwaukee, WI, USA) with a standard 8-channel phase array head coil. For each of the 15 noncollinear and noncoplanar diffusion-sensitizing gradients, approximately 70 contiguous slices parallel to the anterior commissure-posterior commissure line were acquired. Imaging parameters were as follows: matrix, 120×120 matrix; field of view, 240×240 mm², TE, 84 ms; TR, 16,000 ms; b=800 mm²/s; and slice thickness, 2 mm. In addition to DTI, conventional T2-FLAIR images were obtained.

Raw image data were transferred in the DICOM format. All DTI images were corrected for eddy current-induced image distortions using FSL software (The Analysis Group, FMRIB, Oxford, UK; http://www.fmrib.ox.ac.uk/fsl). DTI analyses were performed by an experienced rehabilitation medicine physician using the DTI studio (https://www.mristudio.org). All parameters were measured twice, and results were averaged. To reconstruct the corticospinal tract (CST), the first ROI was drawn free-hand in the posterior limb of the internal capsule, through which motor fibers descend; the second ROI was drawn in the basis pontis of the lower pons, using the "AND" operation [20,21]. The CST volume was calculated by the number of voxels in the reconstructed CST.

FA and ADC values were measured for manually-identified and atlas-based 11 ROIs (Fig. 1). These ROIs were defined after a literature review to identify the areas that have been implicated in swallowing function. We chose to include the areas whose function is unclear or whose involvement in swallowing has been questioned, in addition to the areas that have already been implicated in swallowing function. Polygonal ROIs were drawn on the slice of the color map or FA map, according to the previously defined criteria.

To standardize and unify the boundaries of the ROIs, two raters devised the training sessions. The ROIs were drawn after obtaining consensus about the boundaries of ROIs using a few selected images. The first author (rater A) manually investigated the ROIs, and 10 cases were randomly selected to determine the inter-rater reliability of the measurements. An experienced neuroradiologist, with an extensive background in neuroanatomy, participated as the other rater (rater B) and conducted the same drawing process while also being blinded to the diagnosis, gender, age and severity of stroke.

Intraclass correlation coefficients were calculated for the primary somatosensory cortex (PSC), primary motor cortex (PMC), supplementary motor cortex (SMC), anterior cingulate cortex (ACC), orbitofrontal cortex (OFC), parieto-occipital cortex (POC), insular cortex (IC), posterior limb of the internal capsule (PLIC), thalamus, and basal ganglia, from a subset of 10 cases. Although the sample size was small, there was good inter-rater reliability. For the ROIs, the intraclass correlation coefficients were 0.973 (95% confidence interval [CI], 0.90–0.99) for the PSC, 0.96 (95% CI, 0.84–0.99) for the PMC, 0.65 (95% CI, 0.71–0.90) for the SMC, 0.90 (95% CI, 0.67–0.98) for the ACC, 0.58 (95% CI, 0.29–0.88) for the OFC, 0.93 (95% CI, 0.74–0.98) for the POC, 0.89 (95% CI, 0.62–0.97) for the IC, 0.93 (95% CI, 0.75–0.98) for the PLIC, 0.67 (95% CI, 0.12–0.91) for the thalamus, and 0.75 (95% CI, 0.28–0.93) for the basal ganglia. The measurements obtained by rater A were used as the final measured value.

DTI parameters were compared between the two groups and were assessed for correlation with clinical dysphagia parameters.

For comparison between the dysphagia and non-dysphagia groups, clinical factors, including age, duration of VFSS, duration of DTI, initial CDS score, and initial MI, were analyzed using the nonparametric Mann-Whitney test. Sex, lesion side, aphasia, and neglect were analyzed by Pearson chi-square test. Mean FA and ADC values were compared between the two groups for each of the 11 ROIs. The critical value for statistical significance was set at p=0.05. Data analyses were performed using SPSS ver. 17.0 (SPSS Inc., Chicago, IL, USA).

A total of 31 patients with first episode MCA infarction were included (19 males, 12 females; mean age, 61.1±9.42 years). The dysphagia group included 5 women and 12 men, and the non-dysphagia group included 7 women and 7 men. Overall, the mean ages for the dysphagia and non-dysphagia groups were 62.88±9.53 years and 58.93±9.21 years, respectively.

Among the 31 patients, 17 were diagnosed with dysphagia using VFSS. Right-sided lesions were observed in 13 patients (76%) of the dysphagia group and 6 patients (43%) of the non-dysphagia group. No group differences were observed with respect to the number of days from stroke onset to DTI, number of days from stroke onset to VFSS, Mini-Mental Status Examination, or frequency of aphasia or neglect (Table 1).

The mean TVs on the ipsilesional side were 837.29±1,108.88 in the dysphagia group and 838.21±825.84 in the non-dysphagia group. Mean TVs on the contralesional side were 3,052.71±1,272.24 in the dysphagia group and 2,467.29±879.53 in the non-dysphagia group (Table 2). There were significant intergroup differences in the following four DTI values: FA values in the contra-lesional primary motor cortex (0.41±0.04 in the dysphagia group vs. 0.45±0.03 in the non-dysphagia group); ADC values in the bilateral posterior limbs of the internal capsule (ipsilesional side, 0.71±0.15 in the dysphagia group vs. 0.59±0.13 in the non-dysphagia group; contra-lesional side, 0.64±0.05 in the dysphagia group vs. 0.60±0.05 in the non-dysphagia group; all p<0.05) (Table 3).