In industrial computed tomography for defect detection in 18650 lithium-ion batteries, streak-like artifacts caused by sparse-view projection sampling severely hinder the accurate identification of subtle structural defects. This paper proposes a hybrid-domain CT reconstruction algorithm with projection inpainting based on a joint CNN-Transformer architecture. The proposed method integrates projection-domain restoration with image-domain optimization to construct an end-to-end dual-domain collaborative reconstruction framework. Specifically, a hybrid-domain consistent restoration module is designed to leverage filtered back-projection priors to guide the convolutional network in performing an initial completion of sparse-view projection data. In addition, a multi-level Transformer structure is introduced to model global correlations across projection views, thereby accurately correcting projection deviations caused by sparse sampling and noise. The overall framework enables an accurate mapping from undersampled sinograms to high-fidelity CT images. Simulation and experimental results demonstrate that the proposed method effectively suppresses artifacts under sparse-view projection sampling condition and outperforms present methods in key metrics such as RMSE, PSNR, and SSIM. In particular, it shows superior performance in edge preservation and structural recovery for pixel-level defects, highlighting its strong potential for high-performance industrial CT defect inspection.
Zihao Liu, Chenglong Wang, Zhengxin Li et al.· International Conference on...· 0 citations
Accurate segmentation of polyps and skin lesions is pivotal for clinical diagnosis, yet existing methods struggle with low contrast, ambiguous boundaries, and cross-domain distribution discrepancies. Discriminative networks and most diffusion-based segmentation approaches predict standalone binary masks, leaving the visual priors of large-scale pretrained generative models largely unexploited. We propose InstEditSeg, a unified generative framework that reformulates medical segmentation as an instruction-driven image editing problem. Instead of emitting a mask, the model renders a color-coded overlay on the original image, conditioned on a textual instruction, so that the edited output aligns with the natural image distribution learned by latent diffusion models and mitigates the domain gap between natural and medical imagery. To recover fine anatomical structures, we introduce DINOv3 as an auxiliary visual encoder and a DINO Feature Guidance Block that builds a multi-scale feature pyramid. The pyramid is fused into the diffusion U-Net by channel concatenation and zero-initialized convolution so that hierarchical discriminative priors can be injected without perturbing the pretrained weights. A dual-branch classifier-free guidance strategy requiring only two forward passes per denoising step reduces inference cost. On polyp and skin lesion benchmarks the framework achieves accuracy competitive with strong discriminative baselines, and it further demonstrates concrete advantages of the generative formulation: notably better cross-domain generalization on unseen data, more complete multi-lesion segmentation, instruction-conditioned task control, and sampling flexibility. We also analyze the strengths and limitations of the paradigm, including its color sensitivity and unsupported attribute-conditioned selection. Code is available at: https://github.com/wincharm001/InstEditSeg.
Zihao Liu, Zhe Zhu, Xuzi Shi· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
