Artificial Intelligence in Bone Cancer Detection: A Comprehensive Analysis of Imaging-based Approaches

Abstract

Bone cancer diagnosis remains a major clinical challenge due to the complexity of tumor characteristics and the limitations of traditional diagnostic methods, which are often time-intensive and dependent on specialist expertise. Artificial intelligence (AI) has gained significant momentum as a tool to enhance diagnostic accuracy, reduce subjectivity and support early detection. This review consolidates recent developments in AI-based bone cancer detection and classification, encompassing machine learning, deep learning, hybrid learning, transfer learning and ensemble approaches applied to X-ray, CT, MRI and histopathological imaging. The reviewed studies report promising diagnostic accuracy, often exceeding 90%, in tasks such as detection, localization, and subtype classification. The AI models have exhibited diagnostic efficiency comparable to radiologists, particularly in terms of speed and reproducibility. However, limitations such as small and homogeneous datasets, absence of external validation, single-modality dependence, and high computational demands restrict clinical generalizability. By systematically synthesizing findings across diverse methodologies, this review offers a comprehensive perspective on the current progress, challenges, and future directions of AI-driven bone cancer diagnosis, highlighting the need for scalable, interpretable, and resource-efficient frameworks to advance clinical adoption and improve patient outcomes.

Ähnliche Arbeiten