Scholarlink

<p>Accurate and efficient liver segmentation from computed tomography (CT) images remains a critical challenging task due to the organ’s irregular shape, variable intensity, and lies close to surrounding organs with similar appearance. In this study, we propose SLIC-Former, a superpixel-guided transformer framework for automatic liver segmentation in abdominal CT scans. The method replaces fixed square image patches with adaptive superpixels generated by the Simple Linear Iterative Clustering (SLIC) algorithm, so that each token follows anatomical boundaries and reduces redundant computation. Quantitative evaluation on the Liver Tumor Segmentation (LiTS) dataset demonstrated strong segmentation performance, achieving a Dice coefficient of 0.93, an IoU of 0.87, and a VOE of 13.5%, indicating high overlap with expert annotations. Qualitative results confirm that the model produces smooth and coherent liver masks. Overall, SLIC-Former offers an accurate and computationally efficient tool for automatic liver segmentation and provides a promising basis for future extensions to more organs, larger datasets, and clinical decision-support systems.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

SLIC-Former: a superpixel-guided transformer framework for automatic liver segmentation in CT images

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月10日

查看全文
  • Sarah F. Elqersh,
  • Amira Y. Haikal,
  • Mahmoud M. Saafan,
  • Noha A. Sakr

摘要

Accurate and efficient liver segmentation from computed tomography (CT) images remains a critical challenging task due to the organ’s irregular shape, variable intensity, and lies close to surrounding organs with similar appearance. In this study, we propose SLIC-Former, a superpixel-guided transformer framework for automatic liver segmentation in abdominal CT scans. The method replaces fixed square image patches with adaptive superpixels generated by the Simple Linear Iterative Clustering (SLIC) algorithm, so that each token follows anatomical boundaries and reduces redundant computation. Quantitative evaluation on the Liver Tumor Segmentation (LiTS) dataset demonstrated strong segmentation performance, achieving a Dice coefficient of 0.93, an IoU of 0.87, and a VOE of 13.5%, indicating high overlap with expert annotations. Qualitative results confirm that the model produces smooth and coherent liver masks. Overall, SLIC-Former offers an accurate and computationally efficient tool for automatic liver segmentation and provides a promising basis for future extensions to more organs, larger datasets, and clinical decision-support systems.