Dermoscopy imaging has been a routine examination approach for skin lesion diagnosis. Accurate segmentation is the first step for automatic dermoscopy image assessment. The main challenges for skin lesion segmentation are numerous variations in viewpoint and scale of skin lesion region. To handle these challenges, we propose a novel skin lesion segmentation framework via a very deep residual neural network based on dermoscopic images. The deep residual neural network and generic multi-path Deep RefineNet are combined to improve the segmentation performance. The deep representation of all available layers is aggregated to form the global feature maps using skip connection. Also, the chained residual pooling is leveraged to capture diverse appearance features based on the context. Finally, we apply the conditional random field (CRF) to smooth segmentation maps. Our proposed method shows superiority over state-of-the-art approaches based on the public skin lesion challenge dataset.