Scholarlink

Purpose <p>The paper presents a high-precision hierarchical registration method to accurately align image coordinates across Whole Slide Images of histopathological slides. The proposed technique was designed to achieve robust and pixel-precise stain- and scanner-independent colocalization. It addresses well-known challenges of histopathological imaging and differences arising from various staining protocols and digitization processes.</p> Methods <p>Our method leverages the Elastix registration framework to achieve exceptionally precise colocalization of cell nuclei and other similarly sized tissue structures. By utilizing the pyramidal data structure of Whole Slide Images, we developed a hierarchical, multi-stage registration algorithm in which the transformation is gradually refined from a macroscopic to a microscopic scale.</p> Results <p>Unlike other work in the field, our approach focuses on the colocalization of tissue structures rather than the alignment of the image data. The algorithm achieved sub-micrometer accuracy in colocalization, outperforming state-of-the-art solutions. We propose two distinct registration strategies to minimize the computation time, considering the spatial distribution of the coordinates.</p> Conclusion <p>This algorithm is designed exclusively to compute point cloud transformations within Whole Slide Images, achieving a high accuracy at the expense of computation time. Consequently, it should be considered a highly specialized solution tailored to a specific subset of the registration problems occurring in digital pathology.</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

A high-precision hierarchical registration approach for stain- and scanner-independent colocalization on whole slide images in histopathology

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年5月23日

查看全文
  • Tom Bisson,
  • Michael Franz,
  • Tim-Rasmus Kiehl,
  • Peter Boor,
  • Peter Hufnagl,
  • Norman Zerbe

摘要

Purpose

The paper presents a high-precision hierarchical registration method to accurately align image coordinates across Whole Slide Images of histopathological slides. The proposed technique was designed to achieve robust and pixel-precise stain- and scanner-independent colocalization. It addresses well-known challenges of histopathological imaging and differences arising from various staining protocols and digitization processes.

Methods

Our method leverages the Elastix registration framework to achieve exceptionally precise colocalization of cell nuclei and other similarly sized tissue structures. By utilizing the pyramidal data structure of Whole Slide Images, we developed a hierarchical, multi-stage registration algorithm in which the transformation is gradually refined from a macroscopic to a microscopic scale.

Results

Unlike other work in the field, our approach focuses on the colocalization of tissue structures rather than the alignment of the image data. The algorithm achieved sub-micrometer accuracy in colocalization, outperforming state-of-the-art solutions. We propose two distinct registration strategies to minimize the computation time, considering the spatial distribution of the coordinates.

Conclusion

This algorithm is designed exclusively to compute point cloud transformations within Whole Slide Images, achieving a high accuracy at the expense of computation time. Consequently, it should be considered a highly specialized solution tailored to a specific subset of the registration problems occurring in digital pathology.