Radiotherapy is a well-established modality for cancer treatment, employed for more than 100 years. Conventional radiotherapy aims at eliminating all cancer cells by irradiating the whole tumor, which will result in damage to healthy adjacent tissues. An unconventional approach to radiotherapy is to partially irradiate the tumor, thereby reducing normal tissue damage. However, by utilizing classical cancer theory, where it is assumed that the tumor only consists of non-interacting malignant cells, partial radiation is not expected to result in any therapeutic benefit . But new studies are suggesting that an anti-tumor immune response may be induced by partial irradiation, where the immune system is working with radiation to kill malignant cancer cells. In order to achieve a mechanistic understanding of these effects and guide new experimental designs, it is necessary to develop a mathematical prediction model. In the current study a model has been developed, encompassing the interplay between immune cells and cancer cells, with or without irradiation. In the model the tumor comprises four cell types: active and inactive CD8+ T cells (immune cells), viable cancer cells and doomed cells (destroyed cells). Each day the cell interactions are calculated, describing the tumor evolution. The main quantity of interest was the tumor volume. To test the accuracy of the model, experimental data of radiation-induced anti-tumor immune response from previous publications are sampled. The model was fitted to these data using a novel optimization method, which resulted in promising fits and low root mean squared (RMSE) values. Additionally, predictive capabilities of the model were tested by using the estimated model parameters to predict scenarios using higher dose and different tumor irradiation fraction. These predictions were validated by experimental data. The predictions slightly deviate from the observations, and more validation data is needed to conclude whether the model possess predictive capabilities. To achieve partial radiation a particular technique called GRID therapy may be implemented, where the dose e.g. can be delivered in cylinders evenly spread across the tumor. Clinical practice of GRID therapy is limited and has always been employed in combination with conventional radiotherapy. This is likely due to a lack of knowledge for the underlying biological mechanisms. Instead of using cylinders, we suggest partial irradiation GRID therapy by prescribing dose to spheres evenly distributed inside the tumor. Several methodologies for placing the spheres evenly distributed is developed and implemented in Varian Eclipse treatment planning system for a patient with a large bulky lung tumor (largest diameter 10 cm). In the treatment plans protons are employed over X-rays in order to maximize dose delivery precision and to decrease normal tissue damage. Protons has an increased dose deposition toward the end of the track, characterized by the Bragg peak. Limitations of our GRID therapy lies with the lateral spread of the Bragg peak. Therefore, an analysis of the proton beamlet is conducted in the treatment planning software, where the size of the Bragg peak in the lateral direction is σ ~ 0.50 mm (90 MeV) and increase with energy of the imparted protons. Knowing the limitations of the treatment delivery, different methods of GRID therapy by prescribing dose into evenly distributed spheres are compared. However, a perfect dose distribution is impossible due to limitations in proton physics. The scoring of treatment plans is defined by the dose between prescription spheres (valley dose) for a given dose in the spheres (peak dose), where a lower valley dose is preferable. Face-centered cubic (FCC) was the best method with the lowest valley dose, 8.4 Gy for a 15 Gy peak dose. To investigate GRID plan parameters that result in the most promising anti-tumor immune response, the immune model is combined with the results from GRID treatment plans. By implementing valley dose between prescriptions spheres into the immune model, a tumor outcome probability space is simulated. For a GRID treatment plan using the FCC method the optimal parameters were 10 Gy prescribed dose and 0.2 tumor volume irradiation fraction with treatments twice a week. The newly obtained treatment parameters were then implemented in the treatment planning software and compared to a conventional intensity modulated proton therapy plan. GRID therapy reduced the normal tissue doses to all organs at risk in the tumor vicinity for the same tumor effect. However, further experimental validation and careful clinical testing is needed before any firm conclusions can be made. The current study set out to rejuvenate GRID therapy and bring into our proton-based approach with prescription spheres closer to clinical applications. By combining GRID treatment plans with the immune model, the framework may potentially be used as treatment support system. By predicting the extent of tumor progression and selecting the best treatment parameters for GRID therapy, the future goal is to pick the best treatment plan for the individual patient so that more can be cured with less side-effects.