Medical Image Denoising Using the Cuckoo Optimization Algorithm

Medical images acquired using imaging modalities such as Magnetic Resonance Imaging (MRI), Computed Tomography (CT), and ultrasound constitute essential sources of information for clinical diagnosis, treatment planning, and quantitative biomedical analysis. Nevertheless, these images are inevitably affected by different types of noise and acquisition-related degradations, which can obscure anatomical structures, distort intensity information, and adversely affect subsequent image-processing and diagnostic tasks. Consequently, effective noise reduction while preserving diagnostically important structures remains a fundamental challenge in medical image processing. Among conventional denoising techniques, the bilateral filter has attracted considerable attention because of its ability to suppress noise while preserving edges by jointly incorporating spatial proximity and intensity similarity into the filtering process. Despite these advantages, the performance of the bilateral filter is highly dependent on the appropriate selection of its parameters and filtering window size. Fixed or manually selected parameters may not provide satisfactory performance across images characterized by different noise distributions and intensities. To overcome this limitation, this study proposes a modified and adaptive bilateral filtering framework in which the filter parameters and weighting coefficients are automatically optimized using the Cuckoo Optimization Algorithm (COA). The proposed hybrid approach exploits the global optimization capability of COA, inspired by the parasitic breeding behavior of cuckoos, to efficiently search the parameter space and identify an appropriate combination of bilateral-filter coefficients. An objective fitness function based on image-quality measures is employed to guide the optimization process toward improved denoising performance. The proposed framework is evaluated experimentally on benchmark medical image datasets and compared with conventional and alternative filtering approaches. Both quantitative and qualitative assessments demonstrate that the proposed method provides superior noise suppression while maintaining important anatomical structures, fine textures, and sharp image boundaries. In particular, the method exhibits robust performance under high-density noise conditions, where excessive smoothing can substantially degrade clinically relevant information. The results demonstrate the effectiveness of integrating metaheuristic optimization with edge-preserving filtering and indicate that the proposed framework can improve the quality and interpretability of medical images for subsequent computer-aided analysis and clinical applications.

کلیدواژه ها:

Medical image denoising ، bilateral filter ، Cuckoo Optimization Algorithm (COA) ، Parameter optimization ، Edge Preservation ، Image quality enhancement