Vox-UDA: Voxel-wise Unsupervised Domain Adaptation for Cryo-Electron Subtomogram Segmentation with Denoised Pseudo Labeling

TL;DR

Vox-UDA tackles the lack of labeled data and domain shift in cryo-ET subtomogram segmentation by performing voxel-wise unsupervised domain adaptation. It combines a Noise Generation Module that injects target-like noise into source data and a Denoised Pseudo-Labeling strategy built on a teacher-student framework with an Improved Bilateral Filter to produce high-quality pseudo-labels for the target domain. The method achieves state-of-the-art results on simulated and real cryo-ET datasets, often surpassing fully supervised baselines, and is validated through extensive ablations and visualizations. This voxel-level UDA approach enhances robustness to noise and enables scalable segmentation in 3D biomedical imaging.

Abstract

Cryo-Electron Tomography (cryo-ET) is a 3D imaging technology facilitating the study of macromolecular structures at near-atomic resolution. Recent volumetric segmentation approaches on cryo-ET images have drawn widespread interest in biological sector. However, existing methods heavily rely on manually labeled data, which requires highly professional skills, thereby hindering the adoption of fully-supervised approaches for cryo-ET images. Some unsupervised domain adaptation (UDA) approaches have been designed to enhance the segmentation network performance using unlabeled data. However, applying these methods directly to cryo-ET images segmentation tasks remains challenging due to two main issues: 1) the source data, usually obtained through simulation, contain a certain level of noise, while the target data, directly collected from raw-data from real-world scenario, have unpredictable noise levels. 2) the source data used for training typically consists of known macromoleculars, while the target domain data are often unknown, causing the model's segmenter to be biased towards these known macromolecules, leading to a domain shift problem. To address these challenges, in this work, we introduce the first voxel-wise unsupervised domain adaptation approach, termed Vox-UDA, specifically for cryo-ET subtomogram segmentation. Vox-UDA incorporates a noise generation module to simulate target-like noises in the source dataset for cross-noise level adaptation. Additionally, we propose a denoised pseudo-labeling strategy based on improved Bilateral Filter to alleviate the domain shift problem. Experimental results on both simulated and real cryo-ET subtomogram datasets demonstrate the superiority of our proposed approach compared to state-of-the-art UDA methods.

Figures (7)

• Figure 1: Some examples of the subtomograms and their corresponding segmentation masks. This figure shows: (a) simulated 3D cryo-ET subtomogram; (b) grey-scale ground truth segmentation mask; (c) binary segmentation mask after pre-processd (b), we set a threshold (300 in this paper) to turn the grey-scale mask into a binary one; (d) and (e) are real 3D cryo-ET subtomogram and its binary mask, respectively.
• Figure 2: Overview of our proposed Vox-UDA framework. IBF denotes the improved Bilateral Filter, which is detailed in Fig \ref{['fig:DPL']}. We use different colors to represent different flows. Best viewed in color.
• Figure 3: Proposed improved Bilateral Filter. (a) Both domain filtering and range filtering are applied to an sub-figure extracted from the input target subtomogram with size $3 \times 3 \times 3$. (b) Deploying Laplace transform to obtain the gradient changes used in range filtering.
• Figure 4: Visualization of subtomogram segmentation results using 1bxn, 1f1b and 1yg6 as the source datasets. We use UCSF Chimera pettersen2004ucsf for 3D cryo-ET visualization.
• Figure 5: Additonal visualization of subtomogram segmentation results using 2byu, 2h12 and 21db as source dataset.