Together with the radiation oncology department of the Amsterdam UMC, location AMC, Peter Bosman of CWI’s Life Sciences and Health group has been granted a project by the Dutch Cancer Society (KWF Kankerbestrijding) for the research, development, and nation-wide validation of new treatment planning software for internal radiation treatment (i.e., brachytherapy) of cervical cancer. The project, entitled “Fast, accurate, and insightful brachytherapy planning for cervical cancer through artificial intelligence” is co-funded by industry partner Elekta.
Brachytherapy
In this new project, with AMC as lead institute (project leader: Dr. Tanja Alderliesten), the focus will be on the extension of previous and still ongoing research on brachytherapy (a type of radiation to treat cancer) planning for prostate cancer to cervical cancer and to clinical translation. With approximately 750 newly diagnosed patients yearly, cervical cancer is the 11th most frequent cancer among women in the Netherlands and the 3rd most frequent cancer among women aged 15-45yrs. Brachytherapy has been proven to have the essential potential needed, to provide the high radiation dose required to control cervical cancer. Moreover, it is the only method capable of doing so without extreme risks of side effects. Although the achieved local control of current brachytherapy practice is good, patients do still suffer from acute (e.g., diarrhea, bladder inflammation) and long-term side effects (e.g., changes in vaginal anatomy and function that potentially negatively affect the patient’s sexual function). For an important part, this is because limitations of currently available clinical treatment planning software make it non-trivial to design treatment plans that achieve the crucially important preferred trade-off between covering the cervical clinical target volume and sparing the surrounding organs at risk.
Designing brachytherapy plans for cervical cancer brings its own unique challenges that require further problem-specific algorithmic innovation. For example determining how many catheters should be used in combination with an intracavitary applicator to realize the maximum achievable result for individual patients. The new project will furthermore go beyond algorithmic innovation by also performing a nation-wide study to evaluate the new planning software in a to-be-developed user-friendly prototype, aiming to facilitate broad clinical uptake.
Evolutionary algorithms
In 2015, Peter Bosman was granted a similar project together with AMC in NWO’s IPPSI-TA programme. In that project, with CWI as the lead institute, the focus has been on developing new –predominantly evolutionary – algorithms, to innovate brachytherapy planning software. On the clinical side, the focus has been on prostate cancer. Results from this project have been very successful, enabling automatically finding better plans in terms of clinical evaluation criteria in a fraction of the time compared to the current mostly-manual clinical approach at AMC. Moreover, because the new approach allows finding multiple high-quality plans that represent possible trade-offs between covering the clinical target volume and sparing the surrounding organs at risk, the new approach was also found to be uniquely insightful. These results have led to clinicians at AMC preferring automatically constructed plans over plans coming from current clinical practice at a rate of nearly 100%. Moreover, the latest algorithmic innovation was recently awarded the ESTRO-Elekta Brachytherapy Award at the European SocieTy for Radiation & Oncology (ESTRO 38) conference in Milan, Italy.
Tanja Alderliesten, Project leader on behalf of Amsterdam UMC
