Innovative hybrid approach for enhancing multispectral transmission images in early breast cancer screening

In early breast cancer screening, multispectral transmission imaging is an emerging approach. In this research, denoising and image registration are focused on improving the quality of multispectral transmission images and enhancing the accuracy of image registration. The Root Mean Square Error (RMSE) and Registration Time (RT) can be reduced, and the coefficient correlation (CC), Mutual Information (MI), Normalized Cross Correlation (NCC), as well as Normalized Mutual Information(NMI) of an image, can be improved by frame accumulation. However, during acquiring a multispectral transmission image, the sequence of frames is shifted because of human respiration, camera jitter, and significant light absorption. This leads the frame accumulation technique to the loss of accuracy. In this paper, we have suggested bilateral with unsharp masking to denoise the multispectral transmission images, improve the MI and CC, NCC, NMI, and RT, and reduce the RMSE to prevent this consequence. This proposed algorithm improves the CC, MI, NCC, and NMI and reduces the RMSE and RT in the experimental work. To validate the suggested technique, this study processed multispectral transmission images for early breast cancer screening with 600, 620, 670, and 760 nm wavelengths. This study revealed that the proposed method is much faster, more accurate, and better at preserving the frame-accumulated image’s RMSE, CC, MI, NCC, NMI, and RT. In addition to reliably and quickly registering the image sequence with enhanced image quality, the method presented in this research establishes the groundwork for early screening of breast cancer and abnormality. The accumulated metrics allow for a comprehensive assessment of the pipeline’s performance across the image dataset. This quantitative evaluation supports the validation of the proposed methodology.