Skip to main content
Fig. 3 | Brain Informatics

Fig. 3

From: Inferring the temporal evolution of synaptic weights from dynamic functional connectivity

Fig. 3

Performance of static FC measures in inferring the presence of synapses. A Example distribution of static FC values measured using XCov. The 90th, 95th and 99th percentile were highlighted by dashed vertical lines (in green, red and blue, respectively). B Connectivity matrix obtained by considering only synaptic weights above a given percentile of the static FC distribution. The color of synapses represents the percentile (matched with panel A) for which they were included in the connectivity matrix. C Precision–recall (PR) curves computed from 180 min of simulated activity for TE, HOTE, XCov and XCorr. Each point is one percentile of the distribution of static FC values across pairs, going from the 1st (bottom right) to the 99th (top left) percentile. D AUPR trend with simulation length (length ranges from 5 to 180 min). E Comparison of precision in identifying connected pairs with simulation lengths, with a 90th (1000 pairs) and 95th (500 pairs) percentile threshold on the static FC distributions. F Fraction of pairs belonging to each group of synapses, from 180 min simulation and with a 90th percentile threshold on the FC distributions GT denotes ground truth. All error bars in the figure are SEM across repetitions of the simulation (N = 5)

Back to article page