Artificial Intelligence (AI) presents a promising solution to mitigate the rising pressure on healthcare workers by optimizing time-consuming and laborious tasks. The technical and data-heavy nature of radiotherapy (RT) lends itself well to AI solutions. Despite the suitability of AI for RT, a gap persists between AI development and its adoption. A challenge addressed in the literature is the black-box characteristic of AI. XAI aims at minimizing AI opacity, thereby increasing AI interpretability for humans. To better understand the concepts related to XAI, a theoretical model was developed and evaluated through a national survey among RT workers. In the theoretical model, the interplay between XAI, perceived risk, perceived control, trust, and actual usage of AI is conceptualized. We test our model using a structural equation modeling approach. The model contributes to a better understanding of the adoption of AI in RT.