**5. Conclusion**

At modern radiotherapy, the main aim is enhancing treatment quality by maximizing target localization and dose delivery accuracy onto tumor volume while minimizing the dose received by normal nearby tissues. Reaching to this aim can be problematic and difficult for thoracic tumors where these tumors move mainly due to respiration. Therefore, while tumor motion is an issue, target localization cannot be done carefully and an over-under dose my deliver onto tumor volume that will not be the prescribed dose simulated at treatment planning process. In order to compensate the effect of tumor motion error during therapeutic beam irradiation, several strategies have been implemented or under developing. Three major strategies are as follow: breath-holding technique as old method, respiratory-gated radiotherapy as current clinical available method, and real-time tumor tracking radiotherapy as under developing technique. In the latter case, the irradiation beam is continuously repositioned dynamically to trace breast tumor motion in real time. For both latter cases, the key component for reaching to our aim is to discover the information of tumor position versus time. To do this, some additional monitoring systems are required to track tumor motion as real time ranging from continuous X-ray imagers to the use of external markers or surrogates radiotherapy. In this chapter, we introduced readers with tumor motion as a challenging issue during radiotherapy and presenting external surrogates based radiotherapy as clinical implemented method at several radiotherapy centers or hospitals in the worldwide. In this work, we utilized a typical fuzzy logic-based correlation model to predict tumor motion due to the robustness and simplicity of this model that has been proved at our recent works. This method is still under assessment to minimize available uncertainty errors or to remove possible drawbacks. We had several comprehensive studies on different aspect of this strategy by introducing different prediction models for real-time tumor tracking, their mathematical structures, and the properties of motion data set as inputs of the prediction models [34–38].
