![]() | Only 14 pages are availabe for public view |
Abstract Carcinoma is the globe’s second-most prevalent contributor to demise. Cancer can be detected early, which lowers mortality rates and increases the likelihood of successful therapy. The cancerous tumor known as hepatocellular carcinoma (HCC) targets the liver and can be fatal. The incidence of liver cancer patients is still rising despite technological breakthroughs in prevention, diagnosis, and therapy. Even if some of the liver’s parts are not in good shape, the liver can still function normally. Therefore, it might be challenging to identify the early signs of liver cancer. The objective of this research is to develop a method for cancer detection using artificial intelligence. This is done by using CNN model for identification HCC from scents in biological data. Based on electronic nose measurements, odors of biological substances. The study comprised Egyptian men and women who were referred to the Alexandria Main University Hospital for diagnosis and/or treatment. All the subjects had a comprehensive medical examination, as well as demographic information and electronic nose measurements. The E-Nose employed in this investigation was a movable PEN3 (Airsense Analytics GmbH, Schwerin, Germany) with ten nonspecific metal-oxide detectors. All experiment measurements and sensor array patterns were stored in files for subsequent investigation. The data was divided into two categories depending on their health state. Controlled health participants were among the participants who showed no indications or pathologies or have any health issues, including diseases. The information came from three biological samples: exhaled breath, blood, and urine. Data is preprocessed before being split into training and testing datasets for One Dimensional Convolutional Neural Network training and evaluation. The process of producing additional information from existing information so as to unnaturally raise the quantity of data is known as data augmentation. Network architecture comprised of five convolutional layers, one max-pooling layer, one fully connected layer and two dense layers. The proposed network has best accuracies 0.9993, 0.9990, 0.9996 for exhaled breath, blood and urine dataset respectively with AUC of 1.00 for all datasets. Conclusion and Recommendations 87 6. CONCLUSION AND RECOMMENDATIONS Recent advances in both biology and computer sciences have prompted researchers to pay more attention to the significance of computational tools in cancer research. Artificial intelligence and machine learning approaches have gained considerable interest due to their shown benefits over traditional cancer detection and treatment methods. Deep learning is an artificial intelligence technique that uses algorithms to estimate functions between input and output vectors. Because they do not require intrusive procedures, machine learning systems may be used to diagnose cancer more rapidly and accurately than traditional methods. Using electronic nose measurements processing, we developed a CNN network for effectively identification of Hepatocellular carcinoma, which obtained accuracy of up to 0.999, furthermore high AUC of 1.00 obtained using exhaled breath, blood or urine datasets, based on the scents of distinct biological fluids like we used in this study, exhaled breath, blood and urine. To the best of knowledge, this is the first research that uses this type of data to construct a Deep learning model that can distinguish between HCC and healthy people. We can summaries the study’s primary strengths as: • PCA cluster diagrams for HCC patients and healthy controls, measuring, exhaled breath, blood and urine with variances of 99.2, 96.6 and 96.5%, respectively. While with 100, 85.2, and 99.9% accuracy, the CNN network can differentiate between patients with HCC and healthy controls, utilizing exhaled breath, blood and urine data in scoring time less than a second. • When contrasting CNN and PCA findings, we notice CNN’s strong performance and high accuracy. |