Enhancement of microtubules in EM tomography

The interpretation of the EM tomography of microtubules is challenging due to the low SNR and low contrast of the volume data. Therefore, image enhancement is crucial for the subsequent segmentation and structural analysis of microtubules. In this paper, we propose a model based 3D image enhancement approach by combining transform domain technique and spatial domain techniques in three consecutive steps. The enhancement starts with an anisotropic invariant wavelet transform to effectively enhance the elongated features, followed by a 3D shape filter via eigen analysis to capture the local geometric properties of the tubular structure. The enhancement ends with a coherence enhancing diffusion to complete the interruptions along the microtubules. The contribution of this work is that we have tailored and improved each of the above techniques to exploit the unique geometric and photometric properties of microtubules. Experimental results indicate that our proposed approach has excellent performance in noise removal and enhancement of the tomography volume.