TY - JOUR
T1 - Cancer Detection and Classification Using CNN Model
AU - Okae, Percy
AU - Addo, Theophilus
AU - Owusu-Afari, Joseph Boateng
AU - Bondzie, Gifty
N1 - Publisher Copyright:
© 2024 Seventh Sense Research Group®.
PY - 2024/12
Y1 - 2024/12
N2 - The research utilizes the CNN model to develop the machine learning mode due to its image extraction performance. The system was developed to identify and categorize eight (8) different kinds of cancers, namely lymphoma, oral cancer, brain cancer, breast cancer, cervical cancer, kidney cancer, lung and colon cancer, and leukemia. The multi cancer image dataset from Kaggle was utilized to train and test the models. The dataset contained eight (8) types of cancers grouped into different classes. For each class, 2000 images were used for training and 500 for testing. Pre-processing techniques were applied to normalize and standardize the images to ensure the correct format and resolution. Nine (9) CNN models were trained, with eight responsible for classifying each cancer type while the remaining model detects the cancer type. The system was designed to perform two levels of classification for each image. The first level is the detection of the type of cancer, and the second level is the classification of the cancer type. Generally, the manual examination of cancer diagnoses is error-prone, and this work sought to automate the process as best as possible by investigating the performance of the CNN model on selected types of cancer. The results demonstrated the effectiveness of the developed system in accurately detecting and classifying the eight types of cancers and the potential to alleviate the errors faced with the manual examination. All the models obtained accuracies above 90%.
AB - The research utilizes the CNN model to develop the machine learning mode due to its image extraction performance. The system was developed to identify and categorize eight (8) different kinds of cancers, namely lymphoma, oral cancer, brain cancer, breast cancer, cervical cancer, kidney cancer, lung and colon cancer, and leukemia. The multi cancer image dataset from Kaggle was utilized to train and test the models. The dataset contained eight (8) types of cancers grouped into different classes. For each class, 2000 images were used for training and 500 for testing. Pre-processing techniques were applied to normalize and standardize the images to ensure the correct format and resolution. Nine (9) CNN models were trained, with eight responsible for classifying each cancer type while the remaining model detects the cancer type. The system was designed to perform two levels of classification for each image. The first level is the detection of the type of cancer, and the second level is the classification of the cancer type. Generally, the manual examination of cancer diagnoses is error-prone, and this work sought to automate the process as best as possible by investigating the performance of the CNN model on selected types of cancer. The results demonstrated the effectiveness of the developed system in accurately detecting and classifying the eight types of cancers and the potential to alleviate the errors faced with the manual examination. All the models obtained accuracies above 90%.
KW - Cancer detection and classification
KW - Convolutional Neural Network (CNN)
KW - Machine learning
KW - Magnetic Resonance Imaging (MRI)
KW - Mobile application
KW - Web application
UR - http://www.scopus.com/inward/record.url?scp=85213283715&partnerID=8YFLogxK
U2 - 10.14445/22315381/IJETT-V72I12P104
DO - 10.14445/22315381/IJETT-V72I12P104
M3 - Article
AN - SCOPUS:85213283715
SN - 2349-0918
VL - 72
SP - 42
EP - 54
JO - International Journal of Engineering Trends and Technology
JF - International Journal of Engineering Trends and Technology
IS - 12
ER -