Analyses of biomedical images often rely on accurate segmentation of structures of interest. Traditional segmentation methods based on thresholding, watershed, fast marching, and level set perform well in high-contrast images containing structures of similar intensities. However, such methods can under-segment or miss entirely low-intensity objects on noisy backgrounds. Machine learning segmentation methods promise superior performance but require large training datasets of labeled images which are difficult to create, particularly in 3D. Here, we propose an algorithm based on the Local Binary Fitting level set method and its application specifically designed to improve the segmentation accuracy for low-contrast structures even with significant noise levels present. The proposed algorithm, the Normalized Local Binary Fitting level set method, shows promise in enhancing the segmentation of low-contrast structures in biomedical images, addressing the limitations of traditional segmentation methods, and offering an alternative to machine learning approaches that require extensive training datasets.
|