Quantum oncology

Bruno F. E. Matarèse, Arnie D. Purushotham

Siddhi Ramesh, Teague Tomesh, Samantha J. Riesenfeld +2 more

Jiaran Li, Li Li, Ning Liu +3 more

Manish Kakar

In the field of oncology, Artificial intelligence (AI) and deep learning (DL) are an essential component of decision-support. However, traditional narrow-AI models have significant limitations in clinics with respect to narrow, task-specificity (TS), high data requirement and interpretability. Moreover, the absence of common clinical criteria and the lack of doctors in the co-development of explainable AI (XAI) have undermined the implementation causing conflict with general data protection regulation (GDPR), trust and ethical integration requirements. The application of AI/DL tools in clinics are often constrained by TS, heavy dependency on hyperparameter tuning and large data volume. This review fills in these gaps by creating a cohesive framework that links problem-driven clinical demands with emerging technologies, specifically, the convergence of Generalist Medical AI (GMAI) and Quantum Oncology (QO). Although GMAI's use foundation models, have high computational requirements and have inherent complexity in their validation pipelines, they are designed to be based on self-supervised learning from multimodal data to address a range of downstream clinical tasks. This review aims to critically discuss the potential of quantum computing (QC) to augment GMAI for more efficient data processing, medical imaging, drug discovery, and genomic analysis, owing to the inherent strengths of quantum superposition and entanglement that surpass the capabilities of classical AI/DL systems. Structural and technical trade-offs of this paradigm change are also discussed. We also give recommendations for safe bedside translation by facilitating a common assessment through clinician in the loop design, the CLAIM checklist, and the framework FUTURE-AI. Finally, this analysis outlines oncology and quantum convergence into quantum oncology (QO). This enables scalable, sustainable, and precision oncology while respecting ethics and privacy.

H. B. SHARMA

Modern radiation oncology depends heavily on computational algorithms for inverse treatment planning, Monte Carlo dose simulations, and real-time image guidance. Classical computing architectures face critical scaling bottlenecks as clinical techniques advance toward ultra-high-resolution volumetric modulated arc therapy (VMAT) and adaptive radiotherapy. This paper explores the transformation of radiotherapy by integration of quantum computing for handling complex planning problems. Quantum computing in radiation oncology promises drastic reduction in planning time by leveraging quantum algorithms for optimization and dose estimation. Preliminary.

Sakshi Kumar, Mrittunjoy Guha Majumdar

Quantum machine learning (QML) leverages quantum phenomena such as superposition, entanglement, and interference to address computational challenges that are difficult for classical systems. In radiation oncology, where treatment planning, image guidance, adaptive replanning, and multi-modal data integration require high-dimensional optimisation and complex pattern recognition, QML may offer advantages in computational efficiency and representational capacity. This review systematically maps quantum algorithms to the clinical radiation oncology workflow. We outline foundational concepts in quantum computing, including complexity insights from the BQP class and practical constraints of the noisy intermediate-scale quantum (NISQ) era. The radiation oncology pipeline-from consultation and simulation to treatment planning, delivery, and follow-up-is analysed to identify computational bottlenecks where quantum methods could provide benefit. Current developments in quantum-enabled diagnostic modelling, quantum-inspired clinical decision support for precision radiotherapy, quantum reinforcement learning for adaptive treatment policies, and emerging concepts such as quantum digital twins are reviewed. Key challenges-including hardware limitations, barren plateaus, data encoding constraints, limited datasets, regulatory considerations, and the gap between theoretical speedups and clinical implementation-are critically examined. Finally, a three-horizon roadmap is proposed outlining near-term proof-of-concept research, mid-term integration with treatment planning systems, and long-term prospects for fault-tolerant quantum simulation in oncology.