Development of a Hybrid Model of Digital Filter and Bicubic Spline Methods in Medical Image Preprocessing

In this paper, we have developed and presented a hybrid algorithm that combines the digital filters and splines, and we have also explored its practical applicability. Digital filters are often referred as edge preserving filters, have minimal influence on noise while preserving the sharp edges and patterns on signal and images. Simultaneously, splines offer high accuracy in handling the complex data and excel in filtering and smoothing input images. This hybrid technique, which integrates the most effective elements of both approaches, enables enhanced noise reduction and preserves the crucial information in biomedical image processing. In this research, we also presented the theoretical background of the hybrid filtering and derived a mathematical model to support it. The hybrid technique "Bilateral + Bicubic", which combines Gaussian filtering and bicubic interpolation to provide best filtering and segmentation result. To test the proposed techniques, we utilized the benchmarked skin cancer dataset, ISIC 2024, and included 1024 dermatographic images. Our proposed hybrid method "Bilateral +Bicubic", provides the filtering with PSNR value of 36.2 dB, and the MSE is 0.00023, best among all. To validate our method we have also compared our techniques with several other techniques, such as Gaussian + Bicubic, and Median + Bicubic methods. Our proposed approach provides not only the best structural similarity but also provides the best filtering result and also perform the best segmentation result. The effectiveness of the proposed hybrid algorithm is discussed in terms of its applicability to enhance diagnostic outcomes in the actual medical image processing.