Digital pathology, deep learning, and cancer: a narrative review

Area of Science:

• Digital pathology and deep learning represent advancements in cancer diagnostics and treatment.
• These technologies have the potential to improve cancer detection rates and patient outcomes.

Background:

• Cancer remains a significant global health challenge, driving the need for innovative diagnostic tools.
• Digital pathology and deep learning are emerging as transformative forces in healthcare, particularly in oncology.

Purpose of the Study:

• This review evaluates the current state of deep learning applications in digital pathology.
• It analyzes factors affecting the development and clinical implementation of these technologies.

Main Methods:

• A comprehensive literature search was conducted across multiple scientific databases.
• 36 relevant articles published between 2014 and 2023 on whole slide imaging and deep learning were selected for analysis.

Main Results:

• Most reviewed studies focus on the laboratory development of deep learning models, with limited emphasis on clinical translation.
• A performance gap often exists between in-vitro model metrics and real-world clinical utility.

Conclusions:

• Bridging the gap between deep learning model development and clinical practice is crucial for enhancing cancer care.
• Future research should focus on clinical trials and longitudinal studies to validate real-world effectiveness and promote equitable adoption.