Artificial Intelligence in Radiology: Transforming Cancer Detection and Diagnosis

Abstract

Artificial intelligence (AI) is increasingly integral to radiological oncology, where it supports complex image interpretation, quantitative tumor analysis, and clinical decision-making. This review synthesizes state-of-the-art developments in AI applications across major cancer types: breast, lung, prostate, brain, gastrointestinal, and metastatic disease, focusing on deep learning, radiomics, and radiogenomics frameworks. These technologies have demonstrated substantial improvements in lesion detection, segmentation accuracy, risk prediction, and molecular phenotype inference, with performance metrics that approach or surpass those of experienced radiologists. The review also explores AI integration with diverse imaging modalities, including computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography/computed tomography (PET/CT), and digital mammography, while examining AI's impact on workflow triage, report standardization, and radiologist efficiency. Despite these advancements, key challenges persist, including limited model generalizability across populations and institutions, data silos, regulatory uncertainty, and the need for explainable AI outputs in clinical contexts. Emphasis is placed on enabling strategies such as federated learning, multicenter data harmonization, post-deployment monitoring, and integration into picture archiving and communication system/radiology information system (PACS/RIS) infrastructure. The article provides a critical evaluation of the current landscape while outlining strategic directions for safe, equitable, and effective implementation. Rather than replacing radiologists, AI is emerging as a collaborative partner shaping a future of data-driven, personalized oncologic imaging that aligns with precision medicine goals.