Convolution Neural Network Approaches for Cancer Cell Image Classification

• Published in: Biotechnology and bioprocess engineering Vol. 28; no. 5; pp. 707 - 719
• Main Authors: Kim, Chaeyoung, Shin, Sungtae, Jeong, Sehoon
• Format: Journal Article
• Language: English
• Published: Seoul The Korean Society for Biotechnology and Bioengineering 01.10.2023; Springer Nature B.V; 한국생물공학회
• Subjects: Artificial neural networks; Biotechnology; Breast cancer; breast neoplasms; Cancer; Cervical cancer; Chemistry; Chemistry and Materials Science; Classification; Colon cancer; Colorectal cancer; colorectal neoplasms; data collection; Deep learning; image analysis; Image classification; Industrial and Production Engineering; Lung cancer; lung neoplasms; Machine learning; Medical imaging; Medical research; neoplasm cells; Neural networks; Review Paper; Transfer learning; uterine cervical neoplasms; 생물공학; deep learning; binary class classification; cancer cell classification; convolutional neural network; multi-class classification
• DOI: 10.1007/s12257-023-0164-7
• ISSN: 1226-8372

Recently, research incorporating the benefits of deep learning in the application of cancer cell classification and analysis has been actively conducted. In this paper, we investigated examples of binary-class classification and multi-class classification of cancer cell image data of commonly occurring types of cancer worldwide, such as cervical cancer, breast cancer, lung cancer, and colon cancer, using convolutional neural networks (CNNs) models. For instance, some studies explored the utilization of transfer learning, leveraging a pre-trained CNN model is used as a starting point for additional training on a specific cancer cell dataset. Cancer cells have irregular and abnormal growth, making accurate classification challenging. The application of deep learning techniques, such as CNN, for cancer cell classification has been able to solve these complex analysis problems and enable fast cancer cell classification results, leading to early detection of cancer. Indeed, most of the studies in this paper achieved high performance using CNN models, and this approach enables faster and more accurate confirmation of cancer cell classification results, leading to early detection of cancer. This shows the current trend of applying deep learning in the application of cancer cell classification and demonstrates the significant potential of deep learning to contribute to cancer research. Overall, we provide an overview of the current trend of applying deep learning in the field of cancer cell classification and expect that deep learning will open the way for more effective cancer diagnosis and treatment in the future.