30 January 2025 in IEEE Transactions on Circuits and Systems for Video Technology
Ultra-high field magnetic resonance imaging (MRI), such as 7-Tesla (7T) MRI, provides significantly enhanced tissue contrast and anatomical details compared to 3T MRI. However, 7T MRI scanners are more costly and less accessible in clinical settings than 3T scanners. In this paper, we propose a wavelet-based frequency attention network (WFANet) and a semi-supervised method named dual domain consistency learning (DDCL), and combine them to form a WFANet-DDCL framework for 7T MRI synthesis. WFANet leverages the frequency sensitivity of the proposed wavelet-based frequency attention encoder (WFAE) along with the large receptive field of dilated convolution. WFAE is proposed as an independent module to capture multi-scale frequency attention via the proposed wavelet-based frequency attention (WFA) mechanism. WFAE can be integrated into any backbone network as a plug-and-play component and improve network performance. To tackle the challenge of limited paired data for network training, DDCL is proposed to take advantage of both paired and unpaired data. Frequency domain perturbation is proposed and combined with Gaussian noise to regularize the supervised learning process in dual domains, better avoiding overfitting. Extensive experimental results demonstrate that WFANet-DDCL can achieve comparable performance to state-of-the-art supervised methods even using 66% of all paired data.
2 March 2026 in Cancer Letters
3 April 2026 in IEEE Transactions on Multimedia
Accurate prognosis prediction in hepatocellular carcinoma (HCC) remains a critical challenge. This paper introduces the Multimodal Framework for HCC Prognosis Prediction Integrating Microscopic Hyperspectral Imaging (MHPF-HSI), which leverages hyperspectral images (HSIs), clinical data, and pathology reports from our custom multimodal dataset to significantly enhance prognosis prediction accuracy. Our investigation reveals that no existing open dataset includes prognosis data integrating HSIs, pathology reports, and clinical information. Therefore, we collaborated with multiple hospitals to create this custom dataset. Our novel Multiscale Spectral Attention Network with Spectral Selection innovatively decouples spatial and spectral features at multiple scales, effectively minimizing noise and redundancy to improve data quality. Furthermore, our threshold-free spectral selection method dynamically identifies and extracts crucial spectral information without human intervention, ensuring precise feature utilization. The integration of these features through our Multimodal Residual Feature Fusion Module, coupled with the Balanced Modality Training strategy, robustly addresses overfitting and improves model reliability across modalities. Experimental results demonstrate that our framework enhances prognosis prediction accuracy by 2% to 5% over conventional RGB imaging methods. Extensive validations confirm the MHPF-HSI framework's superior performance, establishing a new benchmark in HCC prognosis.
22 May 2026 in Optics and Lasers in Engineering