We hypothesized that computational modeling might place electrodes in positions different to those of conventional tDCS, thus increasing the electric field in the target region. We simulated conventional and optimized tDCSs for 21 stroke patients using a within-subject design, and compared the results. Optimized tDCS generated higher-intensity electric fields at targets than did conventional tDCS. Electrode placement for optimized tDCS varied greatly among patients, and differed from the conventional placement. In patients with cortical lesions or low initial FMA-UE scores, the electrode positions of optimized tDCS were remote from those of conventional tDCS. The results confirm our hypothesis; tDCS optimized via computational modeling may improve stroke motor rehabilitation to a greater extent than does conventional tDCS.
Computational modeling using a healthy head model showed that conventional tDCS delivered a robust, average electric field near the hand motor area [11]. However, application of a conventional montage to the stroke head model significantly reduced the electric field in the hand motor region and triggered unexpected current flow caused by the lesion [8, 16]. To overcome the limitations of conventional tDCS, we designed an optimized montage that produced the maximum electric field in the hand knob. The positions of the optimized tDCS electrodes differed from those of conventional electrodes, and the inter-subject variance was large. This result is in agreement with a recent study showing that the optimal anode position targeting hand knob is primarily C3 in healthy individuals, but a different position in stroke patients [9]. However, in the study by van der Crujan et al., the optimal electrode positions were not various because the candidate electrodes were limited to the 80 electrodes of a 10/10 system. On the other hand, in our current study, the optimal electrode position was determined using a more sophisticated 1-cm grid, revealing the diversity of individual, optimal electrode positions. Our results suggest that standard, single electrode configurations do not ensure effective stimulation of stroke patients, and that personalized montages that consider the brain lesions and anatomical differences can be a useful strategy for stroke rehabilitation via tDCS.
We found that the differences in electrode positions between conventional and optimized tDCS were greater in those with cortical than non-cortical lesions, suggesting that the former lesions significantly affect electric field distribution and increase complexity when computationally modeling optimal montage selection. In patients with low initial FMA-UE scores, the optimized electrodes lay further away from conventional electrodes. Strokes that damage the corticospinal tract are accompanied by reductions in the thickness and surface area of the associated cerebral cortex [24, 25]. Cortical atrophy and the resulting increase in local CSF thickness greatly reduce tDCS induced-electric field [8]. In patients with low FMA-UE scores, the corticospinal tract may be seriously damaged and the associated cortical structures changed, affecting the optimal tDCS electrode configuration by altering the electric field around the hand motor area. Therefore, when treating stroke patients with cortical lesions and/or severely compromised, initial upper limb function, use of an optimal montage determined via computational modeling will enhance the effectiveness of tDCS.
Dmochoswki et al. developed an optimal tDCS montage using a volume conduction model and high-definition (HD) electrodes, increasing the electric field strength by an average of 63% compared to the conventional approach [16]. In our present study, the electric field magnitude increased by the optimized tDCS averaged about 20%. This lower increase may be due to our use of large electrodes. A large electrode generates wide but rather weak electric fields; HD electrodes create concentrated strong fields. The electric field strength required to produce a physiological effect is unknown, but one study found that a weak electric field of 0.2 V/m induced small but coherent changes in neuronal timing and rate changes, by affecting dynamic network activity [26]. During in vivo human intracranial recordings, a 1-mA tDCS produced a maximal electric field of 0.4 to 0.5 V/m in the cortical brain, close to the value predicted by computational modeling [6, 7]. One meta-analysis found that a 1-mA tDCS significantly increased corticospinal excitability and that tDCS efficacy depended on current density [27]. These reports indicate that electric field increases of less than 0.1 V/m may affect the tDCS outcomes. Whether the optimized tDCS developed here really improves upper extremity function will be revealed by a randomized clinical trial that is currently in progress [19].
Our study had several limitations. Optimized tDCS generated higher electric fields in target regions than did conventional tDCS in most patients. However, for patient S20, there was no difference. Coincidentally, the conventional tDCS montage may have been among the electrode configurations that generated the highest electric fields. If so, using HD electrodes may be an alternative to create stronger electric fields. Another issue is stroke lesion conductivity; this remains unclear. Several studies used the CSF value, but this may be inaccurate [8, 9]. In this study, we manually segmented stroke lesions distinct from liquefaction lesions on MRI. The conductivity of the lesion was specified as 0.8089, which is the average of reference values for several known brain lesions [21]. It is not known how changes in lesion conductivity affect optimization; further studies are needed.