2 The effect of dry needling in chronic stroke with a complex network approach: a case study 3 Abstract 4 Background: Dry Needling (DN) has been demonstrated to be effective in improving 5 sensorimotor function and spasticity in patients with chronic stroke. Electroencephalogram 6 (EEG) has been used to analyze if DN has effects on the central nervous system of patients with 7 stroke. There are no studies on how DN works in patients with chronic stroke based on EEG 8 analysis using complex networks. Objective: The aim of this study was to assess how DN works 9 when it is applied in a patient with stroke, using the graph theory. Methods: One session of DN 10 was applied to the spastic brachialis muscle of a 62-year-old man with right hemiplegia after 11 stroke. EEG was used to analyze the effects of DN following metrics that measure the 12 topological configuration: 1) network density, 2) clustering coefficient, 3) average shortest path 13 length, 4) betweenness centrality, and 5) small-worldness. Measurements were taken before and 14 during DN. Results: An improvement of the brain activity was observed in this patient with 15 stroke after the application of DN, which led to variations of local parameters of the brain 16 network in the delta, theta and alpha bands, and inclined towards those of the healthy control 17 bands. Conclusions: This case study showed the positive effects of DN on brain network of a 18 patient with chronic stroke. 19 Keywords: EEG. Stroke. Dry needling. Brain. Graph theory. Network. 1 20 21 22 23 24 25 26 27 28 29 1 30 31 Introduction Stroke is one of the leading causes of disability with up to 70% of stroke patients 32 experiencing moderate to severe dysfunction post-stroke which places a heavy physical and 33 mental burden on patients and their families. Early post-stroke rehabilitation can improve 34 recovery, minimize functional disability, and reduce the potential costs of long-term care [1]. 35 Rehabilitation protocols in stroke patients usually combine different physiotherapy 36 approaches with different medical treatments, such as oral antispastic drugs or botulinum toxin 37 type A (BTX-A) infiltration to decrease spasticity and improve motor function. Moreover, in 38 recent years, other non-pharmacological treatments such as dry needling (DN) have been 39 demonstrated to be effective in improving sensorimotor function and spasticity in patients with 40 chronic stroke [2] as well as demonstrating to be a cost-effective treatment [3]. In the case of 41 DN, recent publications suggest that it may have central effects [4-6]. 42 Electroencephalogram (EEG) has been used to analyze if interventions in stroke patients 43 have effects on the central nervous system, as it provides continuous, real-time, non-invasive 44 measurement of brain function, which offers new insights into the pathophysiology of the brain 45 after a stroke [7-10]. Studies of brain network organization have adopted techniques used to 46 quantitative analyze complex networks, largely based on graph theory, which provide a powerful 47 way of quantifying the brain’s structural and functional systems [11]. The low cost and 48 availability of EEG, the simplicity, and the extraordinary sensitivity and specificity make this 49 approach suitable for assessing the efficacy of therapeutic interventions [12]. 50 However, to our knowledge, there are no studies on how DN works in patients with 51 chronic stroke based on EEG analysis using complex networks. Therefore, the objective of this 52 case study was to assess the effects of DN on brain network when it is applied in a patient post- 53 stroke, using the graph theory. 54 55 Material and methods 56 Patient description and assessment 57 The patient was a 62-year-old man with chronic ischemic stroke as diagnosed by the 58 neurologist. His stroke duration was two years since onset. He had right hemiplegia. The affected 2 59 limb was in the Brunnstrom Recovery Stage 3 (marked spasticity with basic limb synergies 60 performed voluntarily). Written consent was obtained before starting the measurements. Initial 61 assessment was carried out with EEG, collecting data during 5 minutes in a resting mode in 62 seated position with eyes closed, before and during DN. Data were compared with those of a 63 healthy matched control (61-year-old). EEG data were recorded with 18 electrodes on the scalp 64 at a sampling rate of 256 Hz. For recording, we used the 10-20 system and the electrodes Fp1, 65 Fp2, F7, F3, Fz, F4, F8, T3, C3, Cz, C4, T4, P3, Pz, P4, T5, T6, O1. 66 67 68 Treatment To perform the intervention, the patient was in supine position on his back. DN was 69 performed in the approximate motor point of the brachialis muscle [13] for 60 seconds using a 70 50mm x 0.3 mm disposable sterile needle; DongBang AcuPrime Ltd, Korea. The needle was 71 manipulated fast in and fast out, with a frequency of approximately 1Hz. The brachialis muscle 72 was selected as it is shown the most limiting factor of the elbow extension in spasticity flexion 73 pattern of elbow and the relevant target for spasticity treatment post stroke [13]. A point from 74 distal 30% and 2 cm medial to a reference line connecting humerus lateral epicondyle to the 75 coracoid process was needled for brachialis muscle [14]. 76 77 78 Data analysis Graph theory allows studying a complex real-world system by defining a network (or 79 graph) as composed by a set of nodes (vertices) and the links (edges) between them, which 80 models such system. Network structures exist in a wide range of different areas, such as 81 technological and transportation infrastructures, social phenomena, biological and neural 82 systems. Each network structure presents specific topological features which characterize a 83 network's connectivity, interactions, and dynamic processes [15]. Therefore, a complex 84 network's analysis relies on using measurements that can express its most relevant topological 85 features to enable characterization of its complex statistical properties [16]. In the case of brain 86 function studies, structural and functional brain networks can be defined from anatomical 87 representations of the brain or from EEG electrodes, while links, depending on the data set, 88 refers to anatomical, functional, or effective connections [17]. In this study, we considered a 89 functional brain network using the electrodes as nodes of the graph and edges defined by 3 90 analyzing the generalized partial directed coherence (GPDC) [18] between the signals of each 91 pair of electrodes. EEG data was extracted in European Data Format and was processed in 92 MATLAB with EEGLAB toolbox to generate the brain functional network and the association 93 matrices. The analyzes were performed in python with NetworkX and SciPy to assess the effects 94 of DN in this patient with stroke. Data below 45 Hz were considered for analysis. 95 Prior to data analysis, aberrant waves such as blink and electromyography signals were 96 removed. GPDC relates each pair of electrodes by assigning a normalized signal coherence value 97 between 0 and 1, thus nodes’ association matrix defines a weighted link between all the nodes of 98 the network. We converted the matrices from weighted to non-weighted directed by establishing 99 a cut-off threshold of 0.15, removing the links between nodes with signal coherence below that 100 threshold. Finally, 2700 unweighted directional matrices for each dataset were obtained 101 corresponding to the processing of the EEG signals carried out on delta (below 4 Hz), theta 102 (between 4 Hz to 8 Hz), alpha (between 8 to 13 Hz), beta (between 13 to 30 Hz), and gamma 103 (between 30 and 45 Hz) bands. 104 For the assessment of the effects of DN in this patient with stroke, we used the following 105 metrics that measure the topological configuration: 1) network density, 2) clustering coefficient, 106 3) average shortest path length, 4) betweenness centrality, and 5) small-worldness. Table 1 107 summarizes these key metrics. 108 Table 1. Network metrics Parameter Abbreviation Density 𝐷𝐷 Average shortest path length Local clustering coefficient Global clustering coefficient Betweenness centrality 109 Small-worldness 𝐿𝐿 𝐢𝐢𝑖𝑖 𝐢𝐢 𝐡𝐡𝐢𝐢 𝑆𝑆𝑆𝑆 4 Formula 2π‘šπ‘š 𝐷𝐷 = 𝑛𝑛(𝑛𝑛 βˆ’ 1) 1 𝐿𝐿 = οΏ½ 𝑑𝑑𝑖𝑖𝑖𝑖 𝑛𝑛(𝑛𝑛 βˆ’ 1) 𝑖𝑖≠𝑗𝑗 1 𝐢𝐢𝑖𝑖 = οΏ½ π‘Žπ‘Žπ‘–π‘–π‘–π‘– π‘Žπ‘Žπ‘—π‘—π‘—π‘— π‘Žπ‘Žπ‘–π‘–π‘–π‘– π‘˜π‘˜π‘–π‘– (π‘˜π‘˜π‘–π‘– βˆ’ 1) (1) (2) (3) 𝑗𝑗,π‘˜π‘˜ (4) 𝑖𝑖 (5) 1 𝐢𝐢 = οΏ½ 𝐢𝐢𝑖𝑖 𝑛𝑛 𝐢𝐢𝐡𝐡 (𝑖𝑖) = οΏ½ π‘–π‘–β‰ π‘—π‘—β‰ π‘˜π‘˜ πœŽπœŽπ‘–π‘–π‘–π‘– (𝑖𝑖) πœŽπœŽπ‘˜π‘˜π‘˜π‘˜ 𝑆𝑆𝑆𝑆 = 𝐢𝐢/𝐿𝐿 (6) 110 Network density (𝐷𝐷, formula 1) was obtained by the ratio of the number of network edges 112 to the maximum possible number of network edges [19], with 𝑛𝑛 as the number of nodes and π‘šπ‘š 113 cohesive and denser the network, and the lower the number, the less cohesive the network. 111 the number of edges. Values of 𝐷𝐷 may range from 0 to 1. The closer 𝐷𝐷 is to one, the more 114 Average shortest path length (𝐿𝐿, formula 2) is defined as the average number of steps in 115 the shortest paths for all node pairs in a network [20], where 𝑑𝑑𝑖𝑖𝑖𝑖 indicates the distance between 116 117 node 𝑖𝑖 and node 𝑗𝑗 (the number of edges in the shortest path between both nodes). Local clustering coefficient (𝐢𝐢𝑖𝑖 , formula 3) calculates the local cohesion of a node 𝑖𝑖 with 119 its neighbors [20], being π‘˜π‘˜π‘—π‘— the number of nodes directly connected with node 𝑖𝑖, and π‘Žπ‘Žπ‘₯π‘₯π‘₯π‘₯ terms 120 otherwise). Global clustering coefficient (𝐢𝐢, formula 4), also called Network average clustering 118 121 122 indicate connections between pair of nodes (π‘₯π‘₯, 𝑦𝑦) (they equal 1 if nodes are connected and 0 coefficient, provides an overall measure of the cohesion of the nodes in the whole network. Centralities measure the relative importance of a node in a network, such as connecting 123 directly or being available to others as well as being an intermediary between others. 124 Infrastructural analysis to determine the characteristics of a network is derived from the concept 125 of centrality, which is measured by a variety of criteria. In this analysis, we used betweenness 126 centrality. Betweenness centrality (𝐡𝐡𝐢𝐢 , formula 5) determines which particular node is most 127 128 129 130 131 132 among the nodes in the network [21]. In the formula, πœŽπœŽπ‘˜π‘˜π‘˜π‘˜ is the number of shortest paths from node π‘˜π‘˜ to node 𝑗𝑗, and πœŽπœŽπ‘˜π‘˜π‘˜π‘˜ (𝑖𝑖) is the number of those paths that pass through node 𝑖𝑖. The measure of network small-worldness (𝑆𝑆𝑆𝑆, formula 6) is defined as the ratio between C and L [22,23]. The 𝑆𝑆𝑆𝑆 coefficient is used to describe the balance between the local connectedness and the global integration of a network. When 𝑆𝑆𝑆𝑆 is larger than 1, a network is said to have small-worldness properties. Small-worldness organization mixes short path length 133 and high clustering. 134 Results 135 The results of the EEG analysis using complex dynamic networks are presented in figures 136 1 to 5, which show the variations of the metrics defined in Table 1 for each of the EEG bands. 137 Each figure present 3 curves showing the changes in the network metrics considered from its 138 maximum value to its minimum value. The green and blue lines show the changes in the network 5 139 parameter considered for the patient before and during the DN treatment respectively, and the 140 red line shows the parameter for the healthy subject as control. 141 Figure 1 shows the changes in network density, variations in the delta, theta, and alpha 142 bands of the stroke case that are significantly different than those of the healthy control. As 143 shown in figure 1, the values of the delta, theta and alpha bands in the stroke case are becoming 144 closer to the healthy control during DN application. In the case of beta and gamma bands, the 145 values are similar for both the stroke patient and the healthy control, although some changes also 146 occurred in these bands during DN. 147 148 6 149 150 151 152 153 Figure 1. Variations in network density show delta (a), theta (b), and alpha (c) bands that becoming more similar to those of the healthy control during DN application. The beta (d) and gamma (e) bands show no significant differences between the stroke case and the healthy control. 154 155 156 Figure 2 on the average shortest path length shows that the variations in the stroke case are smaller than those of the normal case, with significant differences observed in the delta, theta 7 157 and alpha bands. In the beta and gamma bands, there are no significant differences between the 158 stroke case and the healthy control. In the delta, theta, and alpha bands, during DN application, 159 the values of stroke case variations are becoming closer to normal. In the beta and gamma bands, 160 although there is no difference between the stroke case and healthy control, performing DN 161 increased the average shortest path length parameter values in these bands. 162 163 164 165 Figure 2. Variations in average shortest path show delta (a), theta (b), and alpha (c) bands becoming more similar to the healthy control during DN application. The beta (d) and gamma (e) bands show no significant differences between the stroke case and the healthy control. 8 166 Figure 3 shows the global clustering coefficient. In delta, theta, and alpha bands, there are 167 significant differences between the stroke and the healthy control. During DN application, the 168 global clustering coefficient parameter variations in the stroke case are becoming closer to 169 normal. In beta and gamma bands, there are not much difference between the stroke case and the 170 healthy control. In fact, DN caused changes in the structure of the stroke patient’s brain network. 171 The variations in Figure 3 are very similar to the density variations as shown in Figure 1. 9 172 173 174 175 176 Figure 3. Variations in global clustering coefficient show delta (a), theta (b), and alpha (c) bands becoming more similar to the healthy control during DN application. The beta (d) and gamma (e) bands show no significant differences between the stroke case and the healthy control. Figure 4 shows the betweenness centrality changes. The values of these variations in all 177 bands show a lower value for the stroke case compared to those of the healthy control. In all 178 bands, the betweenness centrality variations in the stroke case are becoming closer to normal 179 during DN. In the delta, theta, and alpha bands, there are significant differences in maximum 10 180 values of the variations in betweenness centrality between the stroke case (maximum value less 181 than 0.1) and the healthy control (maximum value larger than 0.2). The betweenness centrality is 182 the only parameter that shows differences in all bands between the stroke case and the healthy 183 control. 184 185 186 187 Figure 4. Variations in betweenness centrality show delta (a), theta (b), alpha (c), beta (d), and gamma (e) bands becoming more similar to the healthy control during DN application. 188 11 189 Figure 5 shows the small-worldness parameter. There were significant differences in 190 small-worldness parameter variations in the patient with stroke and the healthy control in delta, 191 theta, and alpha bands. In these bands, DN caused changes in the structure of the brain network 192 in the stroke case towards normal. There were no differences in the beta and gamma bands 193 between the stroke case and healthy control. In fact, DN caused changes in the structure of the 194 stroke patient’s brain network. 195 196 197 198 Figure 5. Variations in small-worldness show delta (a), theta (b), and alpha (c) bands becoming more similar to the healthy control during DN application. The beta (d) and gamma (e) bands show no significant differences between the stroke case and the healthy control. 12 199 200 Discussion Stroke affects the whole brain and its network characteristics and therefore can be 201 considered as a network disease. Many studies have been performed on the brain networks as 202 well as on the effects of rehabilitation on patients with stroke. Network assessment to predict the 203 treatment effects and to individualize rehabilitation is a promising approach to enhance the 204 specific treatment effects and overall outcome after stroke [12]. 205 We found that variations in network density, global clustering coefficient, and small- 206 worldness were similar. The values of the parameters for the stroke case were higher than the 207 values of the healthy control in the delta, theta, and alpha bands. During DN, the values of the 208 stroke case parameters became closer to the values of the healthy control. The variations in 209 average shortest path and betweenness centrality were the only variations where the healthy 210 control had smaller values than the stroke case. In the case of the average shortest path 211 variations, differences between the stroke case and the healthy control were seen only in delta, 212 theta, and alpha bands, whereas the differences in betweenness centrality variations were in all 213 bands. In all parameters except betweenness centrality, the beta and gamma bands had no 214 differences between the stroke case and the healthy control. However, there were differences in 215 all bands in the betweenness centrality parameter between the patient with stroke and the healthy 216 control. 217 Among rehabilitation methods, DN impact positively on spasticity, pain, and range of 218 motion in patients with stroke [24]. Our results in this patient with stroke showed that DN causes 219 structural changes in the brain network, which is in line with other studies that have used and 220 analyzed the EEG [29] and fMRI [6]. Calvo et al [25] showed that after the application of DN 221 [DNHS technique] based on the measurement of quantitative EEG activity and EEG 222 concordance, improvements in the regional brain activity occured. Mohammadpour et al [6] by 223 using fMRI, showed that DN had a positive effect on the stroke patient's recovery. Absence of 224 resting-state network rearrangement in beta and gamma bands is consistent with previous data 225 [26, 27] considering that measurements with EEG were performed with closed eyes. A possible 226 explanation could be that coherence in higher bands may be more involved in active (either 227 motor or cognitive) tasks [28-30] and therefore, it might be better to study the effects of DN with 228 open eyes for beta and gamma bands. However, we clearly observed the positive effects of DN 13 229 in the delta, theta and alpha bands, which suggest performing DN in resting state could improve 230 the structure of the brain network in this patient with stroke. There is a need for further study on 231 the effects of DN on the structure of the brain network in patients with stroke using the EEG with 232 eyes open. In the case of other interventions in stroke patients such as focal vibration, the authors 233 reported binding power occurred in some central electrodes after focal vibration [31]. In fact, 234 focal vibration as a rehabilitation method causes a cortical reorganization of the somatosensory 235 representational maps. However, DN rebuilds the structure of the brain as shown in this patient 236 with stroke, which has not been observed, to the best of our knowledge, in any other study. 237 The delta, theta and alpha bands are related to low frequency bands in EEG. With bregard 238 to the large differences between the brain network of the stroke case and the healthy control on 239 these bands before DN and significant improvements occurring during DN, we may 240 conceptualize that the DN normalized the structure of the brain network of this patient with 241 stroke in low frequency bands. Changes in resting state network were mainly detected in EEG 242 low frequency bands, while no network rearrangement was found in beta and gamma bands 243 except for betweenness centrality parameter which is consistent with previous findings [32]. 244 Regarding high frequency bands of beta and gamma, there were no significant changes in 245 the brain network structure in the studied parameters. Nevertheless, DN changed the structure of 246 the brain network in this patient with chronic stroke. There were differences between the stroke 247 case and the healthy control in high frequency bands and betweenness centrality parameter in the 248 brain network. Considering the betweenness parameter, DN caused improvements even in high 249 frequency bands such that they became similar to those of the healthy control network. The 250 greatest difference between the stroke case and the healthy control was in the betweenness 251 centrality parameter. Given that betweenness centrality plays an intermediate role in the network, 252 it seems that DN could work through the creation of a network that maximizes the betweenness 253 centrality in the nodes. This indicates that the betweenness centrality is an important parameter 254 in the brain as it was modified in all bands in this patient with stroke. 255 Small-worldness organization of the brain networks [32-35] along with other measures 256 from graph theory have been used to quantify the changes in brain connectivity and functional 257 recovery in patients after stroke [36-39]. The results on the delta, theta, and alpha bands showed 258 that the small-worldness variations in the patient with stroke were greater than those in the 14 259 healthy control, which is consistent with the previous a reports for the theta band [40]. 260 Accordingly, Caliandro et al. [32] found an increased segregation and a decreased integration in 261 ΞΈ-band network consistent with a previous fMRI study [41]. 262 This study has strength and limitations. The main strength is the innovation associated 263 with using graph theory of complex network approach in the clinical context used for analysis of 264 the effects of DN on the structure of the brain network. The main limitation is that this study was 265 carried out only in one patient and therefore further study is needed to investigate whether DN is 266 effective in improving the brain networks in stroke patients towards normal and whether there 267 might be a cause-effect relationship. Future research should also examine how the brain network 268 structure differs from normal in patients with chronic stroke. In this patient with stroke, the 269 changes were not evaluated after the end of DN application. As well, clinical measures 270 particularly muscle spasticity level and motor function were not assessed. Therefore, studies with 271 larger sample sizes with rigorous design and follow-up should be carried out to investigate the 272 effects of DN on the structure of the brain network and evaluate the associations with functional 273 changes in patients with stroke. 274 In conclusion, this case study showed the positive effects of DN on brain network with 275 the delta, theta and alpha bands becoming closer to the normal in density, average shortest path, 276 global clustering coefficient, and small-worldness parameters in a patient with stroke. Further 277 investigations in the context a well-design clinical trials are warranted. 278 279 280 281 282 283 284 285 286 287 15 288 References 289 290 291 [1] Duncan PW, Zorowitz R, Bates B, et al. Management of Adult Stroke Rehabilitation Care: a clinical practice guideline. Stroke. 2005;36(9):e100-e143. doi:10.1161/01.STR.0000180861.54180.FF 292 293 294 295 [2] Calvo S, BrandΓ­n-de la Cruz N, JimΓ©nez-SΓ‘nchez C, Bravo-Esteban E, Herrero P. Effects of dry needling on function, hypertonia and quality of life in chronic stroke: a randomized clinical trial [published online ahead of print, 2021 Dec 13]. Acupunct Med. 2021;9645284211056347. doi:10.1177/09645284211056347 296 297 298 [3] FernΓ‘ndez-Sanchis D, BrandΓ­n-de la Cruz N, JimΓ©nez-SΓ‘nchez C, Gil-Calvo M, Herrero P, Calvo S. Cost-Effectiveness of Upper Extremity Dry Needling in Chronic Stroke. Healthcare (Basel). 2022;10(1):160. Published 2022 Jan 14. doi:10.3390/healthcare10010160 299 300 301 [4] Ansari NN, Naghdi S, Fakhari Z, Radinmehr H, Hasson S. Dry needling for the treatment of poststroke muscle spasticity: a prospective case report. NeuroRehabilitation. 2015;36(1):61-65. doi:10.3233/NRE-141192 302 303 304 [5] Fakhari Z, Ansari NN, Naghdi S, Mansouri K, Radinmehr H. A single group, pretest-posttest clinical trial for the effects of dry needling on wrist flexors spasticity after stroke. NeuroRehabilitation. 2017;40(3):325-336. doi:10.3233/NRE-161420 305 306 307 [6] Mohammadpour F, Ali Oghabian M, Nakhostin Ansari N, Naghdi S, Dommerholt J. Effects of dry needling on post-stroke brain activity and muscle spasticity of the upper limb: a case report. Acupunct Med. 2021;39(1):69-71. doi:10.1177/0964528420920294 308 309 310 [7] Cohen BA, Bravo-Fernandez EJ, Sances A Jr. Automated electroencephalographic analysis as a prognostic indicator in stroke. Med Biol Eng Comput. 1977;15(4):431-437. doi:10.1007/BF02457998 311 312 313 [8] Cohen BA, Bravo-Fernandez EJ, Sances A Jr. Quantification of computer analyzed serial EEGs from stroke patients. Electroencephalogr Clin Neurophysiol. 1976;41(4):379-386. doi:10.1016/0013-4694(76)90100-0 314 315 316 [9] JuhΓ‘sz C, Kamondi A, Szirmai I. Spectral EEG analysis following hemispheric stroke: evidences of transhemispheric diaschisis. Acta Neurol Scand. 1997;96(6):397-400. doi:10.1111/j.1600-0404.1997.tb00305.x 317 318 319 [10] Luu P, Tucker DM, Englander R, Lockfeld A, Lutsep H, Oken B. Localizing acute strokerelated EEG changes: assessing the effects of spatial undersampling. J Clin Neurophysiol. 2001;18(4):302-317. doi:10.1097/00004691-200107000-00002 320 321 322 [11] Bullmore E, Sporns O. Complex brain networks: graph theoretical analysis of structural and functional systems [published correction appears in Nat Rev Neurosci. 2009 Apr;10(4):312]. Nat Rev Neurosci. 2009;10(3):186-198. doi:10.1038/nrn2575 16 323 324 325 [12] Guggisberg AG, Koch PJ, Hummel FC, Buetefisch CM. Brain networks and their relevance for stroke rehabilitation. Clin Neurophysiol. 2019;130(7):1098-1124. doi:10.1016/j.clinph.2019.04.004 326 327 328 [13] Genet F, Schnitzler A, Droz-Bartholet F, et al. Successive motor nerve blocks to identify the muscles causing a spasticity pattern: example of the arm flexion pattern. J Anat. 2017;230(1):106-116. doi:10.1111/joa.12538 329 330 331 [14] Park BK, Shin YB, Ko HY, Park JH, Baek SY. Anatomic motor point localization of the biceps brachii and brachialis muscles. J Korean Med Sci. 2007;22(3):459-462. doi:10.3346/jkms.2007.22.3.459 332 333 334 [15] Barrat A, BarthΓ©lemy M, Pastor-Satorras R, Vespignani A. The architecture of complex weighted networks. Proc Natl Acad Sci U S A. 2004;101(11):3747-3752. doi:10.1073/pnas.0400087101 335 336 [16] Costa LD, Rodrigues FA, Travieso G, Villas Boas PR. Characterization of complex networks: A survey of measurements. Advances in physics. 2007;56(1):167-242. 337 338 [17] Friston KJ. Functional and effective connectivity in neuroimaging: a synthesis. Human brain mapping. 1994;2(1‐2):56-78. 339 340 [18] Baccala LA, Sameshima K, Takahashi DY. Generalized partial directed coherence. In2007 15th International conference on digital signal processing 2007 Jul 1 (pp. 163-166). IEEE. 341 342 [19] Barabasi AL, Albert R. Emergence of scaling in random networks. Science. 1999;286(5439):509-512. doi:10.1126/science.286.5439.509 343 344 345 [20] van Wijk BCM, Stam CJ, Daffertshofer A. (2010). Comparing brain networks of different size and connectivity density using graph theory. PLoS ONE. 2010; 5(10): e13701–e13701. doi:https://doi.org/10.1371/journal.pone.0013701 346 347 348 349 [21] Makarov VV, Zhuravlev MO, Runnova AE, Protasov P, Maksimenko VA, Frolov NS, Pisarchik AN, Hramov AE. Betweenness centrality in multiplex brain network during mental task evaluation. Physical Review E. 2018; 98(6): 062413. doi: https://doi.org/10.1103/PhysRevE.98.062413 350 351 352 [22] Onnela JP, SaramΓ€ki J, KertΓ©sz J, Kaski K. Intensity and coherence of motifs in weighted complex networks. Phys Rev E Stat Nonlin Soft Matter Phys. 2005;71(6 Pt 2):065103. doi:10.1103/PhysRevE.71.065103 353 354 [23] Rubinov M, Sporns O. Complex network measures of brain connectivity: uses and interpretations. Neuroimage. 2010;52(3):1059-1069. doi:10.1016/j.neuroimage.2009.10.003 355 356 357 358 [24] Valencia-ChuliΓ‘n R, Heredia-Rizo AM, Moral-Munoz JA, Lucena-Anton D, Luque-Moreno C. Dry needling for the management of spasticity, pain, and range of movement in adults after stroke: A systematic review. Complement Ther Med. 2020;52:102515. doi:10.1016/j.ctim.2020.102515 17 359 360 361 [25] Calvo S, Navarro J, Herrero P, Del Moral R, De Diego C, MarijuΓ‘n PC. Electroencephalographic changes after application of dry needling [DNHSΒ© technique] in two patients with chronic stroke. Myopain. 2015;23(3-4):112-7. 362 363 364 [26] Dubovik S, Pignat JM, Ptak R, et al. The behavioral significance of coherent resting-state oscillations after stroke. Neuroimage. 2012;61(1):249-257. doi:10.1016/j.neuroimage.2012.03.024 365 366 367 [27] Westlake KP, Hinkley LB, Bucci M, et al. Resting state Ξ±-band functional connectivity and recovery after stroke [published correction appears in Exp Neurol. 2012 Dec;238(2):100]. Exp Neurol. 2012;237(1):160-169. doi:10.1016/j.expneurol.2012.06.020 368 369 370 [28] Gerloff C, Bushara K, Sailer A, et al. Multimodal imaging of brain reorganization in motor areas of the contralesional hemisphere of well recovered patients after capsular stroke. Brain. 2006;129(Pt 3):791-808. doi:10.1093/brain/awh713 371 372 373 [29] de Vico Fallani F, Astolfi L, Cincotti F, et al. Evaluation of the brain network organization from EEG signals: a preliminary evidence in stroke patient. Anat Rec (Hoboken). 2009;292(12):2023-2031. doi:10.1002/ar.20965 374 375 [30] Jin SH, Lin P, Hallett M. Reorganization of brain functional small-world networks during finger movements. Hum Brain Mapp. 2012;33(4):861-872. doi:10.1002/hbm.21253 376 377 378 [31] Li W, Li C, Xu Q, Ji L. Effects of Focal Vibration over Upper Limb Muscles on the Activation of Sensorimotor Cortex Network: An EEG Study. J Healthc Eng. 2019;2019:9167028. Published 2019 May 27. doi:10.1155/2019/9167028 379 380 381 [32] Caliandro P, Vecchio F, Miraglia F, et al. Small-World Characteristics of Cortical Connectivity Changes in Acute Stroke. Neurorehabil Neural Repair. 2017;31(1):81-94. doi:10.1177/1545968316662525 382 383 [33] Wang L, Yu C, Chen H, et al. Dynamic functional reorganization of the motor execution network after stroke. Brain. 2010;133(Pt 4):1224-1238. doi:10.1093/brain/awq043 384 385 386 387 [34] Vecchio F, Caliandro P, Reale G, et al. Acute cerebellar stroke and middle cerebral artery stroke exert distinctive modifications on functional cortical connectivity: A comparative study via EEG graph theory. Clin Neurophysiol. 2019;130(6):997-1007. doi:10.1016/j.clinph.2019.03.017 388 389 390 [35] Vecchio F, Tomino C, Miraglia F, et al. Cortical connectivity from EEG data in acute stroke: A study via graph theory as a potential biomarker for functional recovery. Int J Psychophysiol. 2019;146:133-138. doi:10.1016/j.ijpsycho.2019.09.012 391 392 [36] Jiang L, Xu H, Yu C. Brain connectivity plasticity in the motor network after ischemic stroke. Neural Plast. 2013;2013:924192. doi:10.1155/2013/924192 393 394 395 [37] Rehme AK, Grefkes C. Cerebral network disorders after stroke: evidence from imagingbased connectivity analyses of active and resting brain states in humans. J Physiol. 2013;591(1):17-31. doi:10.1113/jphysiol.2012.243469 18 396 397 398 [38] Wu J, Quinlan EB, Dodakian L, et al. Connectivity measures are robust biomarkers of cortical function and plasticity after stroke. Brain. 2015;138(Pt 8):2359-2369. doi:10.1093/brain/awv156 399 400 [39] Van Kaam RC, van Putten MJAM, Vermeer SE, Hofmeijer J. Contralesional Brain Activity in Acute Ischemic Stroke. Cerebrovasc Dis. 2018;45(1-2):85-92. doi:10.1159/000486535 401 402 403 [40] Fanciullacci C, Panarese A, Spina V, et al. Connectivity Measures Differentiate Cortical and Subcortical Sub-Acute Ischemic Stroke Patients. Front Hum Neurosci. 2021;15:669915. Published 2021 Jul 1. doi:10.3389/fnhum.2021.669915 404 405 406 [41] Adhikari MH, Hacker CD, Siegel JS, et al. Decreased integration and information capacity in stroke measured by whole brain models of resting state activity. Brain. 2017;140(4):10681085. doi:10.1093/brain/awx021 19