This constructed model can help guide clinicians direct personalised therapy strategies in a more comprehensible manner.
Radiometrics refers to the use of high-throughput radiological features from medical images to guide clinical decision-making. These obtained radiological features can reflect certain biological information of tumours that cannot be directly obtained by traditional image interpretation. Therefore, machine learning-based approaches that use in-depth data mining will allow us to gain more knowledge about tumour heterogeneity. However, there has not yet been a model to accurately predict the efficacy of radiotherapy in brain cancer patients in clinical practice.
Recently, a research team led by Professor Li Hai and Wang Hongzhi from the Hefei Institutes of Physical Sciences of the Chinese Academy of Sciences (CAS) recently constructed an interpretable radiomic model to predict radiotherapy treatment response in patients with brain tumours.
From 228 patients, 960 features were extracted from the magnetic resonance imaging (MRI) images of each tumour before radiotherapy. Then, they used machine learning to model the radiation. Finally, Shapley Additive exPlanations (SHAP) was used to assess the treatment response of whole-brain radiotherapy.
According to Wang Yixin, the model performed well and the prediction results of the external validation group also show that the model has generalisation to a certain extent.
At the same time, the SHAP method can achieve interpretability and visualisation of the model, avoiding the “black box” effect of traditional machine learning algorithm. This means that the team’s constructed radiomic model with the SHAP method is better able to aid clinicians in directing personalised whole-brain radiotherapy strategies in a more comprehensible manner. [APBN]
Source: Wang et al. (2022). The radiomic-clinical model using the SHAP method for assessing the treatment response of whole-brain radiotherapy: a multicentric study. European Radiology, 1-11.