AI-Powered Deep Learning Model Accurately Counts Cell Types in Whole Slide Images

Improved methods for counting cell types in pathology images using deep learning approaches are much needed. Current techniques based on segmentation and regression face challenges such as the necessity for precise pixel-level annotations, difficulties in handling overlapping nuclei or obscured regions, and insufficient information on individual cell type locations. Moreover, probabilistic models tend to yield uncertain predictions and can lead to overconfident predictions. Researchers have now developed an advanced deep learning model to predict and count various cell types in the tumor microenvironment, which is expected to enhance the accuracy and efficiency of cancer diagnostics and treatment planning.

Identifying the different cell types in the tumor microenvironment can offer valuable insights into the tumor's histology and underlying biology. Precise and reliable cell type counting is also crucial for research and clinical applications. In addition, cell counts can be used to study the distribution of different cell types in the tumor microenvironment and its correlation with patient outcomes. In clinical settings, cell counts can help monitor therapy response and track disease progression. Researchers from the University of Eastern Finland (Kuopio, Finland) have proposed a new evidential multi-task deep learning approach, called CT-EMT, to overcome the limitations of current methods for cell type counting in whole slide tumor images. This approach formulates cell type density estimation and cell type counting as regression tasks, and nuclei segmentation as a pixel-level classification task.

The proposed cell type segmentation and counting approach has outperformed state-of-the-art HoVer-Net and StarDist models, with relative improvements of 21% and 12% in terms of mean panoptic quality. The developed model can deliver persuasive interpretations of diverse cell types and can be applied to various computational pathology tasks, such as tumor grading, prognosis, and treatment planning. This work will pave the way for the creation of more accurate and robust digital pathology tools that can support pathologists and clinicians in diagnosing and treating cancer patients.

“The UEF Cancer AI research team aims to explore the potential of using deep learning technology in cancer and health data analysis,” said senior researcher Hamid Behravan of the University of Eastern Finland. “Our study will involve the development and evaluation of cutting-edge deep learning algorithms for analyzing cancer and various types of health-related data, including medical images, genomic data, and electronic health records. We believe that this approach has the potential to significantly improve the accuracy and efficiency of breast cancer diagnosis and treatment planning, as well as to facilitate the discovery of new insights and patterns in cancer data. We hope that our research will contribute to the advancement of precision medicine and the development of more effective and personalized approaches to breast cancer prevention and prognosis.”

Related Links:
University of Eastern Finland

Visit expo >

Gold Member

Pharmacogenetics Panel

VeriDose Core Panel v2.0

Verification Panels for Assay Development & QC
Seroconversion Panels

New

TRAb Immunoassay

Chorus TRAb

New

Fixed Speed Tube Rocker

GTR-FS