Related Experiment Video
Updated: Aug 16, 2026
Quantitation of Protein Expression and Co-localization Using Multiplexed Immuno-histochemical Staining and Multispectral Imaging
Published on: April 8, 2016
PCaSFUA-Net: Spatial frequency collaboration and uncertainty-aware fusion for multimodal prostate cancer segmentation
Chunyu Li1, Mengxing Huang2, Yuchun Li3
1State Key Laboratory of Marine Resource Utilization in South China Sea, School of Information and Communication Engineering, Hainan University, Haikou 570288, China; School of Artificial Intelligence, Hainan Normal University, Haikou, 571158, China.
This study introduces PCaSFUA-Net, a novel deep learning model for accurate prostate cancer segmentation using multiparametric MRI. The network enhances multimodal fusion, improving segmentation of challenging prostate tumors.
Area of Science:
- Medical imaging analysis
- Artificial intelligence in oncology
- Prostate cancer diagnostics
Background:
- Accurate prostate cancer segmentation is crucial for assessing disease progression and prognosis.
- Multiparametric MRI (mpMRI) enhances lesion segmentation accuracy compared to single-modality MRI.
- Existing methods face challenges with high-frequency details, modality uncertainty, and heterogeneous data.
Purpose of the Study:
- To develop an advanced multimodal prostate cancer segmentation network (PCaSFUA-Net).
- To improve segmentation accuracy by addressing limitations in detail preservation and multimodal fusion.
- To leverage a pretrained medical vision foundation model for efficient task adaptation.
Main Methods:
- Proposed PCaSFUA-Net, a multimodal segmentation network utilizing a pretrained medical vision foundation model.
- Integrated a unified multimodal enhancement and fusion (MEF) framework with a coarse-to-fine (C2F) strategy.
- Employed a spatial-frequency collaborative (SFC) module for enhanced intra-modality representations and a modality uncertainty-aware fusion (MUAF) module for robust cross-modal fusion.
Main Results:
- PCaSFUA-Net demonstrated superior performance in prostate cancer segmentation across multiple evaluation metrics.
- The network effectively preserved high-frequency details and handled modality uncertainty.
- Consistent outperformance was observed on public and private datasets compared to existing methods.
Conclusions:
- PCaSFUA-Net offers a significant advancement in multimodal prostate cancer segmentation.
- The proposed MEF framework and C2F strategy effectively enhance fusion and refinement.
- The model shows promise for improved clinical assessment and prognostic evaluation of prostate cancer.
