Ovarian cancer is one of the deadly diseases which causes death among women, during lactation or pregnancy. Though there are many advancements and many symptoms, it is difficult to distinguish between malignant and benign. Sonography remains the primary and most prominent imaging process for ovarian cancer prediction. The most common method used to predict ovarian cancer is computed imaging. Magnetic resonance imaging stays as the second imaging method to find the problems, primarily in the pelvis. Furthermore, most of the previous models lack to give clear accuracy while detecting the disease. Further, there is an immense need to implement stable ultrasound criteria. Predicting ovarian cancer using the ultrasound imaging method is safe, it is the easiest way and standard when compared with other imaging equipment. This paper presents an Artificial Intelligence (AI) based an optimized method to predict ovarian cancer at an early stage. To overcome all the obstacles in finding the cancer, whether it is harmful or harmless, a grey wolf optimization-based convolutional neural network algorithm is suggested to overcome the restrictions. A large number of computerized ultrasound images or datasets are processed in limited duration and the output will give a more exact or higher rate of cancer recognition.