Neuroimaging studies in adults report activations of mirror neuron system (MNS) during a task of imitation learning^1-3 and during rehabilitation treatment of motor functions after stroke^4,5. This activation is also observed in stroke patients, showing a motor improvement following intensive training based on action observation⁶, suggesting that MNS activation constitutes a powerful mechanism for recovery of motor deficits after stroke. To our knowledge , only one fMRI study on action observation of a simple hand movement was conducted in brain damaged children⁷. We aimed to study the (re)-organization of MNS using as visual stimulus a goal-directed hand action in children with congenital unilateral cerebral palsy (UCP). 12 UCP children (age=11±3y) were studied: 4 with early malformative (type I), 4 with prenatal (type II) and 4 with connatal (type III) brain lesions^8,9, all clinically tested with Melbourne Unilateral Upper-Limb assessement¹⁰ (MUUL). Data were compared with adult and children controls previously described^11-13. Written informed consent, approved by IRB, was obtained for all subjects. MRI data were acquired using a 1.5T GE scanner, including a whole brain 3D FSPGR (isotropic voxel =1x1x1mm3) and a fMRI session with two series of 6’36’’ duration (GRE-EPI, TR/TE=3000/50, voxel=3.5x3.5x5mm3, 4 dummy scans). Each series was block-design structured with 16 blocks of 24s alternating between BASELINE and TASK conditions. Four TASK conditions were presented using videoclips of hand actions observed from a first person perspective: simple (S) or complex (C) actions performed by the dominant (d) or no-dominant (nd) hand (Sd,Snd,Cd,Cnd). Each functional series included 2 TASK blocks for each of the 4 hand actions, in a random order¹². The BASELINE condition consisted in the still picture of the resting hands belonging to TASK videos. Data analysis was performed using BrainVoyager. Functional data were preprocessed (mean intensity adjustment, slice scan time correction, 3D motion correction and high-pass temporal filtering), aligned to the volumetric images and transformed into Talairach space. BOLD responses were analyzed using a GLM considering the regressor of interest (box-car function for each block convolved with hemodynamic response) and six spurious movement-regressors in order to generate statistical maps on the group (Random Effect Analysis, p<0.05 Bonferroni-corrected) and on each subject (Fixed effect Analysis, p<0.001 uncorrected). For each subject, a laterality index was calculated by the ratio of number of activated voxels in the dominant (Nd) and non-dominant (Nnd) hemispheres as follows: λ=(Nd–Nnd)/(Nd+Nnd). In UCP, we considered as non-dominant the hemisphere contralateral to the plegic limb.The group of UCP children showed activations in areas belonging to the action observation network¹², as the age-matched controls (Figure 1). With respect to them, UCP children present an activated network more lateralized to the dominant hemisphere with differences related to the type, extension and location of the lesion. Regarding the single-subject lateralization, UCP children with early malformative (I) lesions have λ=0.31±0.12, those with prenatal (II) ones have λ=0.01±0.16, whilst UCP with connatal (III) lesions show a large variability (λ=0.02±0.49, range=-0.62÷0.56). Figure 2 shows the relation between λ values and the MUUL scores of all patients. The linear function (y=A+Bx) that best fits the relationship has A=1.15±0.58 and B=-0.013±0.007 (R²=0.17, p=0.10). Considering only type I and II, the best linear fit becomes highly significant with A=1.22±0.20 and B=-0.014±0.003 (R²=0.79, p=0.002).

This study describes the action observation network in UCP children, using a goal-directed hand action stimulus. Previous works with the same paradigm have demonstrated that the network activated in the healthy children is more bilateral than that of adults group. However, children with UCP present differences in the (re)-organization of this system with respect to the age-matched control group: in general they appear more lateralized in the dominant hemisphere as the adults. Moreover, the specific pattern of lateralization of each patient seems related to the type and extension of lesion. In particular, the early malformative lesions are associated to activations more lateralized to the dominant hemisphere, presenting λ values similar to adult ones; the prenatal forms are more bilateral with λ values similar to age-matched controls; whilst the connatal forms are very variable, with the reorganization depending especially on the extension of the lesion. Finally, we found a negative correlation of λ values with the severity of the hand impairment assessed by MUUL.These preliminary findings demonstrate that our paradigm might be useful to explore the mirror neuron function also in UCP children. Further investigation will be conducted in order to describe in details the re-organization process and to understand whether it could be used as an indicator of the response to therapy in subjects with brain damage and motor disorder.