Aims: This study aimed to detect and classify the cavitation phenomenon during Sonothrombolysis therapy through the use of artificial intelligence, where the region of interest was the heart. It is especially important to have a feedback mechanism to control the intensity and type of cavitation of microbubbles to avoid harm to the patient. Methods: The signals were generated using the Kwave toolbox available for Matlab, where features of the acoustic medium can be settled, including non-linearities, attenuations, and the matrix array topology. The stimuli at a certain point in the medium were analyzed as an ultrasonic signal source of three types: stable cavitation, inertial cavitation, or a combination of the cases mentioned. After the simulation of those signals, an automatic and uncomplicated classifier method was proposed, based on the Continuous Wavelet Transform tool and Convolutional Neural Network (CNN) approach. The method made use of a pre-trained CNN, called AlexNet, using a database of 1,340 waves for training, testing, and validation. The evaluation of the statistics included both the detection using broad and narrow bands, the noise level applied, and the database size. Results: For the case of a database compounded by a random noise level between 2 and 5% (dynamic range of the signals), and narrow-band receivers, the results of the study indicated that the technique achieved state-of-the-art values of around 97%, 88%, and 93% for sensitivity, precision, and F1-score, respectively. Conclusion: The considerable degree of accuracy demonstrated that the use of artificial intelligence can be an approach to explore the detection of cavitation for therapies using ultrasound signals.