Scholarlink

In wound healing assays, accurate segmentation is essential to quantifying cell migration. Manual methods are labor-intensive and subject to individual interpretation. This study presents a robust, automated image analysis framework that integrates entropy- based thresholding techniques in conjunction with level set-based contour analyses to segment scratch wound regions in microscopy images. PC3 Prostate cancer cells were treated with Indole-3-carbinol (I3C) and Docetaxel, and the method demonstrated consistent accuracy across all treatment conditions. Shanbhag entropy-driven thresholding achieved Dice scores up to 0.98 as well, particularly excelling in treated samples. Level set models, including Local Binary Fitting (LBF) and Distance Regularized Level Set Evolution (DRLSE), provided robustness against intensity variation, with DSCs ranging from 0.99 to 0.97. Experimental results show drug-induced inhibition of wound closure, validating both the segmentation pipeline and its application to pharmacological screening. The framework offers a scalable solution for high-throughput wound healing analysis.

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

Artificial Intelligence-Assisted Image Segmentation in Scratch Wound Healing Assays Using Threshold and Level Set Methods

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Viviana Benfante,
  • Muhammad Ali,
  • Farhan Khan,
  • Sebastiano Piana,
  • Alessandro Sperandeo,
  • Anthony Yezzi,
  • Albert Comelli

摘要

In wound healing assays, accurate segmentation is essential to quantifying cell migration. Manual methods are labor-intensive and subject to individual interpretation. This study presents a robust, automated image analysis framework that integrates entropy- based thresholding techniques in conjunction with level set-based contour analyses to segment scratch wound regions in microscopy images. PC3 Prostate cancer cells were treated with Indole-3-carbinol (I3C) and Docetaxel, and the method demonstrated consistent accuracy across all treatment conditions. Shanbhag entropy-driven thresholding achieved Dice scores up to 0.98 as well, particularly excelling in treated samples. Level set models, including Local Binary Fitting (LBF) and Distance Regularized Level Set Evolution (DRLSE), provided robustness against intensity variation, with DSCs ranging from 0.99 to 0.97. Experimental results show drug-induced inhibition of wound closure, validating both the segmentation pipeline and its application to pharmacological screening. The framework offers a scalable solution for high-throughput wound healing analysis.