The aim of the proposed work is to improve low-dose lung CT (LDCT) screening using the processing of weighted intensity averaging over large-scale neighborhoods (WIA-LN). Both current and voltage reductions were considered for LDCT imaging. In the WIA-LN method, the processed pixel intensities are calculated by weighted averaging intensities among a large neighboring region. The weights are determined by the inter-similarity of the surrounding textures. A compute unified device architecture based parallelization was applied to accelerate the implementation. To evaluate the effectiveness of the proposed processing, low-dose lung CT images were obtained under both 75 % reduced tube current and 33.3 % reduced tube voltage condition respectively from a 16 detector rows Siemens CT. The standard routine standard-dose CT images were also collected as the reference images. In addition to clinical data from patients, an anthropomorphic lung phantom was also used in the study. Visual comparison and statistical qualitative analysis of image quality scores on the datasets are made in validation. Compared to the original LDCT images, improved visual and qualitative performance can be observed for the processed images. Statistically significant improvement of noise/artifacts suppression and nodule structure enhancement are achieved by using the proposed method (P < 0.05). The proposed method is capable of providing LDCT images under significantly reduced tube current and voltage settings in low-dose condition. Quality of the processed images was assessed by radiology specialists. Parallelization based algorithm optimization was also performed to increase the clinical applicability of the proposed processing.