Multi-scan Partitioning Visual Mamba for Breast Cancer Segmentation

• Item type: SIG Technical Reports(1)
• 公開日: 2025-01-14
• タイトル: Multi-scan Partitioning Visual Mamba for Breast Cancer Segmentation
• タイトル / 言語: en
• タイトル: Multi-scan Partitioning Visual Mamba for Breast Cancer Segmentation
• 言語: eng
• 資源タイプ / 資源タイプ識別子: http://purl.org/coar/resource_type/c_18gh
• 資源タイプ: technical report
• 著者所属: Waseda University
• 著者所属: Waseda University
• 著者所属(英): en
• 著者所属(英): Waseda University
• 著者所属(英): en
• 著者所属(英): Waseda University
• 著者名: Fengze, Li; Seiichiro, Kamata
• 著者名(英): Fengze, Li; Seiichiro, Kamata
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: Breast cancer is the most commonly diagnosed cancer in women, making it a leading cause of cancer-related deaths worldwide. Accurate diagnosis often relies on Whole-Slide Images (WSIs), a key pathological imaging modality that provides detailed visual information for identifying tumor regions. To facilitate this, advanced image segmentation methods have been developed, with convolutional neural networks (CNNs) and Transformers being the most prevalent approaches. However, CNNs often fail to capture long-range dependencies effectively, while Transformer-based methods suffer from high computational complexity. Recently, Mamba, an emerging method based on the state space model, has gained attention for its ability to address these challenges by capturing long-range dependencies with linear computational complexity. Building on this innovation, we propose the Multi-Scan Partitioning Mamba UNet for WSI segmentation. This approach integrates a novel module, the Multi-Scan Partition State Space (MPSS) module, which we introduce in detail. Extensive experiments validate the efficacy of the proposed method in processing WSIs, demonstrating its superiority over existing benchmarks.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: Breast cancer is the most commonly diagnosed cancer in women, making it a leading cause of cancer-related deaths worldwide. Accurate diagnosis often relies on Whole-Slide Images (WSIs), a key pathological imaging modality that provides detailed visual information for identifying tumor regions. To facilitate this, advanced image segmentation methods have been developed, with convolutional neural networks (CNNs) and Transformers being the most prevalent approaches. However, CNNs often fail to capture long-range dependencies effectively, while Transformer-based methods suffer from high computational complexity. Recently, Mamba, an emerging method based on the state space model, has gained attention for its ability to address these challenges by capturing long-range dependencies with linear computational complexity. Building on this innovation, we propose the Multi-Scan Partitioning Mamba UNet for WSI segmentation. This approach integrates a novel module, the Multi-Scan Partition State Space (MPSS) module, which we introduce in detail. Extensive experiments validate the efficacy of the proposed method in processing WSIs, demonstrating its superiority over existing benchmarks.
• 書誌レコードID / 収録物識別子タイプ: NCID
• 書誌レコードID / 収録物識別子: AA11131797
• 書誌情報: 研究報告コンピュータビジョンとイメージメディア（CVIM）; 巻 2025-CVIM-240, 号 1, p. 1-7, 発行日 2025-01-14
• ISSN / 収録物識別子タイプ: ISSN
• ISSN / 収録物識別子: 2188-8701
• Notice: SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc.
• 出版者 / 言語: ja
• 出版者: 情報処理学会