@@ -14,10 +14,16 @@ import CorticalSites4 from '../Pictures/Research/CorticalSites4.jpg';
14
14
import CorticalSites5 from '../Pictures/Research/CorticalSites5.jpg' ;
15
15
16
16
function Main ( ) {
17
- const [ index , setIndex ] = useState ( 0 ) ;
17
+ const [ index1 , setIndex1 ] = useState ( 0 ) ;
18
18
19
- const handleSelect = ( selectedIndex ) => {
20
- setIndex ( selectedIndex ) ;
19
+ const handleSelect1 = ( selectedIndex ) => {
20
+ setIndex1 ( selectedIndex ) ;
21
+ } ;
22
+
23
+ const [ index2 , setIndex2 ] = useState ( 0 ) ;
24
+
25
+ const handleSelect2 = ( selectedIndex ) => {
26
+ setIndex2 ( selectedIndex ) ;
21
27
} ;
22
28
23
29
return (
@@ -69,7 +75,11 @@ function Main() {
69
75
</ Button >
70
76
</ Col >
71
77
< Col >
72
- < Carousel interval = { null } activeIndex = { index } onSelect = { handleSelect } >
78
+ < Carousel
79
+ interval = { null }
80
+ activeIndex = { index1 }
81
+ onSelect = { handleSelect1 }
82
+ >
73
83
< Carousel . Item >
74
84
< Image fluid src = { CorticalSites1 } />
75
85
</ Carousel . Item >
@@ -86,7 +96,7 @@ function Main() {
86
96
< Image fluid src = { CorticalSites5 } />
87
97
</ Carousel . Item >
88
98
</ Carousel >
89
- { index === 0 ? (
99
+ { index1 === 0 && (
90
100
< p >
91
101
"A DES was used either intraoperatively (depicted) or in the
92
102
epilepsy monitoring unit to identify sites critical to language
@@ -113,24 +123,119 @@ function Main() {
113
123
interquartile range. We used these metrics to train machine
114
124
learning classifiers to predict which nodes would be critical to
115
125
language and speech. Example data (C–E) are provided from a single
116
- participant (n = 1) for each visualization. Source data are
126
+ participant (n = 1) for each visualization. Source data are
117
127
provided as a Source Data file."
118
128
</ p >
119
- ) : (
129
+ ) }
130
+ { index1 === 1 && (
120
131
< p >
121
- "An example of performance of a bivariate smoothing model,
122
- dependently on the number of data-points included in 2D moving
123
- average (window size), for ERC containing 20 channels (K=20)
124
- recorded during naming of ambiguous objects. Top panel shows
125
- results in patient #8. Top-left: the difference between the ERC
126
- values and the values of 2D moving average. Top-middle; confidence
127
- interval. Top-right: the criterion for model selection. X and Y
128
- axes represent window size by distances from the center-point of
129
- the window of 2D moving average, in time-points and
130
- frequency-points accordingly. Colorscale (min-max) at the right.
131
- Bottom panel shows the criterion for model selection averaged over
132
- all patients (bottom-left) and their projections on time-plane
133
- (bottom-middle), and on frequency-plane (bottom-right)."
132
+ "PC participation coefficient, S strength, CC clustering
133
+ coefficient, LEff local efficiency, EC eigenvector centrality. A
134
+ Diagram illustrating coassignment. Two yellow-outlined coassigned
135
+ nodes are found within the same community (dark blue fill); two
136
+ blue-outlined nodes are found in two different communities
137
+ (magenta and orange fill)—i.e., not coassigned. B Diagram
138
+ demonstrating graph metrics. The large magenta node in the top
139
+ panel has a high PC—it connects across all communities in this
140
+ network. The same node has a low clustering coefficient (its
141
+ neighbors are not themselves connected, denoted by dashed arrows)
142
+ and low local efficiency (long path lengths between its
143
+ neighbors). In the bottom panel, the large dark blue node has high
144
+ strength, i.e., a high sum of connection weights. The large orange
145
+ node has higher eigenvector centrality than the smaller orange
146
+ node; both have the same number of connections, but the larger
147
+ node’s connections themselves have more connections. C Intuition
148
+ for three node types. Connector nodes connect across communities
149
+ (high PC), while their neighbors do not connect as closely to each
150
+ other (low CC, LEff). Global hubs connect to many nodes across the
151
+ network (high PC, high S, likely high EC). Local hubs connect
152
+ densely in their neighborhood (low PC, high CC/LEff)."
153
+ </ p >
154
+ ) }
155
+ { index1 === 2 && (
156
+ < p >
157
+ "PC participation coefficient, S strength, CC clustering
158
+ coefficient, LEff local efficiency, EC eigenvector centrality. A
159
+ Composite of all participants’ electrodes colocalized on a single
160
+ template brain. Speech arrest nodes (yellow fill) were primarily
161
+ located in ventral premotor regions, but also in ventrolateral
162
+ prefrontal and ventral temporal regions. Language error nodes
163
+ (blue fill) were widely distributed in perisylvian regions. B
164
+ Three example participant brain reconstructions. Node color
165
+ (filled) represents community assignment, and node size is
166
+ proportional to its participation coefficient. The outline color
167
+ indicates critical nodes (blue—LE node, yellow—SA node). C
168
+ Corresponding three network diagrams. The electrode position is
169
+ spring-weighted (stronger connections draw electrodes closer
170
+ together). Fill color indicates community, and if present, outline
171
+ color indicates critical node type (LE vs. SA) D Corresponding
172
+ network metrics for the three example patients. Metrics for all
173
+ nodes (electrodes) for each of the three participants (n = 1 per
174
+ graph) are plotted. Here, colored circles represent critical
175
+ nodes; gray circles represent other nodes. Boxes demonstrate
176
+ median and interquartile range, and whiskers demonstrate
177
+ non-outlier maxima/minima. Source data are provided as a Source
178
+ Data file."
179
+ { ' ' }
180
+ </ p >
181
+ ) }
182
+ { index1 === 3 && (
183
+ < p >
184
+ "PC participation coefficient, S strength, CC clustering
185
+ coefficient, LEff local efficiency, EC eigenvector centrality. *p
186
+ < 0.05. **p < 0.01. ***p < 0.001 (FDR-corrected). A
187
+ Histogram of the number of communities per participant (n = 16). B
188
+ Coassignment percentages vs. chance. Coassignment is calculated as
189
+ the mean % of critical, LE, or SA node pairs per participant
190
+ sharing a community. Empiric chance was calculated based on 1000
191
+ random shuffles of community assignment per participant, presented
192
+ as mean coassignment% per participant with bars indicating
193
+ standard error of mean (n = 16 for Critical, n = 15 for LE and
194
+ SA). Critical nodes, language error nodes, and speech arrest nodes
195
+ were significantly more likely to coassign in the same communities
196
+ than chance (p < 0.001 for all, one-tailed estimate against
197
+ empiric chance). Language error and speech arrest nodes were not
198
+ more likely to be found in the same community as each other
199
+ compared to chance (35.2 vs. 30.4%, p = 0.112, one-tailed estimate
200
+ against empiric chance). C Network metrics for critical vs. all
201
+ other nodes (150 critical nodes, 1084 non-critical nodes).
202
+ Critical nodes have higher PC and lower CC, LEff, and EC than
203
+ other nodes. D Network metrics for LE, SA, and other nodes (92
204
+ language error nodes, 52 speech arrest nodes, 1084 non-critical
205
+ nodes). LE nodes have markedly higher PC than SA and other nodes.
206
+ C, D Metrics were z-scored for each subject prior to pooling all
207
+ nodes together. All nodes are plotted in light gray; mean values
208
+ per participant in larger, bolder colors. Boxes indicate the
209
+ median and IQR, and notch indicates the standard error of the
210
+ median. Statistical testing is based on a two-sided two-sample
211
+ t-test on z-scored metrics across all pooled nodes with FDR
212
+ correction. For additional details, refer to Table 1. Source data
213
+ are provided as a Source Data file."
214
+ </ p >
215
+ ) }
216
+ { index1 === 4 && (
217
+ < p >
218
+ "For within-participant classification, participants with at least
219
+ four nodes of the relevant class were included; for critical
220
+ nodes, LE nodes, and SA nodes, n = 15, 10, and 8, respectively.
221
+ For across-participant classification, participants with at least
222
+ one node of the relevant class were included—for critical nodes,
223
+ LE nodes, and SA nodes, n = 16, 13, and 13, respectively. A–D Each
224
+ dot represents average classification balanced accuracy or
225
+ sensitivity for a single participant. Box plots show median and
226
+ IQR across participants and are derived from a single value per
227
+ participant. Whiskers indicate a non-outlier maximum range. True
228
+ balanced accuracy and sensitivity were compared against empirical
229
+ chance calculated by label-shuffling. The average chance
230
+ classification accuracy per participant is represented by the
231
+ chance box plots for SVN and KNN (one value per participant). Data
232
+ for SVM, KNN, and chance for SVM and KNN are presented in
233
+ different colors as indicated by the legend. E, F ROC curves
234
+ presented for SVM (solid lines) and KNN (dashed lines)
235
+ classifiers, when classifying SA (orange), LE (magenta), and
236
+ critical (dark blue) nodes separately, as indicated by the legend.
237
+ For further details, refer to Tables 2, 3. Source data are
238
+ provided as a Source Data file."
134
239
</ p >
135
240
) }
136
241
</ Col >
@@ -183,15 +288,19 @@ function Main() {
183
288
</ Button >
184
289
</ Col >
185
290
< Col >
186
- < Carousel interval = { null } activeIndex = { index } onSelect = { handleSelect } >
291
+ < Carousel
292
+ interval = { null }
293
+ activeIndex = { index2 }
294
+ onSelect = { handleSelect2 }
295
+ >
187
296
< Carousel . Item >
188
297
< Image fluid src = { ERC_Naming } />
189
298
</ Carousel . Item >
190
299
< Carousel . Item >
191
300
< Image fluid src = { ERC_Naming2 } />
192
301
</ Carousel . Item >
193
302
</ Carousel >
194
- { index === 0 ? (
303
+ { index2 === 0 ? (
195
304
< p >
196
305
"Results of event-related causality (ERC) estimated with 2D moving
197
306
average of window size 7x7 time-frequency points, averaged across
0 commit comments