An open-source software tool may help improve how digital pathology images are analyzed and integrated with laboratory data, according to a study in Nature Methods.
Whole-slide images (WSIs) are widely used in pathology to assess tissue structure, cellular detail, and disease-related changes. However, analyzing these large images often requires multiple software platforms, which can be difficult to use together and are not always compatible with molecular data such as gene expression results.
LazySlide was developed to address these challenges by providing a single, accessible framework for WSI analysis. Built within the scverse ecosystem, the platform allows histopathology images to be analyzed alongside genomic and transcriptomic data in a unified workflow.
The system includes tools for key steps in digital pathology analysis, including tissue segmentation, cell detection, image tiling, and feature extraction using deep learning models. These processes enable identification of tissue regions, measurement of cellular composition, and quantification of morphological patterns relevant to diagnosis.
At its core is a data structure called WSIData, which allows direct access to multiple slide formats without the need to convert or duplicate files. This design supports efficient handling of large image datasets while maintaining compatibility with other analytical tools.
A key feature is the ability to link image-derived data with molecular information. By combining histology with RNA sequencing data, the system can identify relationships between tissue appearance and underlying biological pathways. Integrating imaging and molecular data improved separation of disease states compared with molecular data alone.
The platform also includes tools that allow users to search images using descriptive terms. For example, users can query for features such as “lymphocyte,” and the system identifies matching regions within the slide. This approach may support more efficient review and annotation of complex cases.
In addition, LazySlide supports classification of tissue types without task-specific training, using image–text models. In testing, it correctly identified organ types from WSIs using simple text prompts.
Benchmarking indicated that the platform required fewer steps to complete standard workflows and performed tissue segmentation more quickly than some existing tools.
The study highlights the potential value of tools that can combine histopathology with molecular data in a single workflow. This may support more comprehensive tissue characterization and help link morphological findings with underlying biology.
Limitations include the need for further validation in clinical settings and integration with existing laboratory systems. As with other computational tools, standardization and reproducibility will be important for routine use.
Overall, LazySlide provides a unified approach to analyzing digital pathology images, with potential to support both research and future diagnostic workflows.
Newsletters
Receive the latest pathologist news, personalities, education, and career development – weekly to your inbox.
