One of the most predominant cancers is liver cancer, which is a major cause of death worldwide. Because liver cancer is asymptomatic and there are currently no sensitive or specific diagnostic tests available, early detection of liver cancer is extremely important for successful therapy. The paper presents An unconventional approach for detecting liver cancer based on particle swarm optimization (PSO). The following measures are a part of the suggested PSO-based strategy. A considerable number of datasets are first segmented using U-net. The segmented volumes are then sent to CNN, where the PSO is used to optimize the CNN parameters, such as the number of convolutional layers, filter size, number of filters, pool size, number of hidden layers, and depth of the filter. The resultant CNN is then used to get the feature vector to train a classification model. We can achieve a more comprehensive and Precise identification of liver cancer, which is crucial for early diagnosis and timely treatment. The proposed method gives an accuracy of 96.67 % with PSO and 93.33% with Genetic Algorithm prove the efficiency of the method. Conclusion: In addition, there is a possibility to investigate different optimization algorithms other than PSO and Genetic Algorithm. This can potentially lead to improved accuracy and faster convergence during the optimization process.