Advancements in deep learning for cancer detection in histopathological images: A comprehensive review of techniques and applications

Cancer is a leading cause of mortality worldwide and remains a major global health challenge. Medical imaging techniques such as computed tomography, magnetic resonance imaging, ultrasonography, and histopathology have critical roles in cancer diagnosis. Histopathological examination, typically performed using biopsy samples, remains the gold standard for cancer detection; however, traditional diagnostic approaches rely heavily on pathologists’ expertise and are often time-consuming and resource intensive. Recent advances in computer vision and deep learning (DL) have improved the automation, accuracy, and consistency of histopathological cancer detection. Therefore, this review investigates the application of DL techniques for cancer detection using histopathological images. First, the process of histopathological image acquisition for cancer analysis is discussed. Second, commonly used pre-processing techniques for converting raw images into suitable analytical formats are reviewed. Third, DL-based classification methods for histopathological cancer images are compared and evaluated. Performance metrics used to assess model quality are also examined. Finally, current challenges and future research directions are discussed. This review may guide future research in DL-based histopathological cancer detection. However, several limitations remain. Existing studies rely predominantly on publicly available datasets, which may introduce bias and limit generalizability across diverse clinical settings. Additionally, variations in staining protocols, imaging conditions, and limited cross-dataset validation may affect model robustness. These limitations highlight the need for standardized benchmarks and more clinically diverse datasets in future research.