A decision aid for radiotherapy dose selection in prostate cancer based on non-linear case based reasoning

A. Cox, N. Mishra, I. Sayers, S. Petrovic, S. Sundar

Research output: Contribution to journalArticlepeer-review

10 Citations (Scopus)


Introduction: External beam radiotherapy is a widely used curative treatment modality for prostate cancer. A dose response relationship exists between total radiation dose and tumour control rates. Randomised clinical trials exploring various dose levels (64 vs 74 Gy, 70 vs 78 Gy) have confirmed the better efficacy of higher doses of radiation. But this better efficacy comes at a price of higher normal tissue toxicity (e.g. rectal toxicity) due to low therapeutic ratio. In day-to-day clinical practice, experienced clinicians decide on an appropriate dose (& Phase1+2dose) for each individual patient evaluating the trade off between increased efficacy and higher toxicity rates of radiation dose escalation. We developed a decision making software tool, which would make this process formal, explicit and more importantly consistent across the entire patient population. Methods: A case based reasoning (CBR) is developed to capture the expertise and experience of oncologists in treating previous patients. The proposed retrieval process makes a trade-off between the benefit and risk of the radiation. Importance (weights) of different clinical parameters in the dose planning is highly subjective and is generally fixed in the system with the input from the oncologist based on their past experience. In this research, the weights corresponding to each feature used in the retrieval process are updated automatically each time after generating a treatment plan for a new patient using a group based simulated annealing approach. Results: The developed approach is analysed on the real data sets collected from patients. Extensive experiments show that in most of the cases, the dose plan suggested by the proposed method is coherent with the dose plan prescribed by an experienced oncologist or even better. Conclusion: Software decision-making tools can aid clinicians make consistent decisions using past experience. A working model of the software will be demonstrated at the conference.
Original languageEnglish
Pages (from-to)S19-S20
Number of pages2
JournalClinical Oncology
Issue number3
Publication statusPublished - Apr 2011

Cite this