Proton therapy is a modality of external radiotherapy that has the potential to provide state-of-the-art dose conformality in the tumor area, since protons have a limited range and deposit most of their energy at the end of their path, in the so-called Bragg peak region. Therefore, it can reduce possible adverse effects on surrounding organs at risk. However, uncertainties in the exact location of the proton Bragg peak inside the patient prevent this technique from achieving full clinical potential. In this context, in vivo verification of the range of protons in patients is key to reduce uncertainty margins. Protoacoustic range verification employs acoustic pressure waves generated by protons due to the radio-induced thermoacoustic effect to reconstruct the dose deposited in a patient during proton therapy. Nevertheless, dose image reconstruction implies a high computational cost which makes difficult to implement this technique in real time during treatment. In this work, we propose to use the a priori knowledge of the shape of the proton dose distribution to create a dictionary with the expected ultrasonic signals at predetermined detector locations. Using this dictionary, the reconstruction of the dose deposited is performed by matching pre-calculated dictionary acoustic signals with data acquired online during treatment. The dictionary method was evaluated on a single-field proton plan for a prostate cancer patient. Dose calculation was performed with the open-source treatment planning system matRad, while acoustic wave propagation was carried out with k-Wave. We studied the ability of the proposed dictionary method to detect range variations caused by anamotical variations, patient position misalignment and errors in CT Hounsfiled Units conversion to relative stopping power. Our results show that the dictionary-based protoacoustic method was able to identify the changes in range originated by all the alterations introduced, with an average accuracy of 1.1 mm. This procedure could be used for in vivo verification, since the in-house developed algorithm takes approximately 100 ms to identify the most probable Bragg peak position.