posted on 2022-03-29, 01:46authored byAbdullah-Al Nahid
Machine-Learning (ML) techniques bring a new paradigm which has generated a revolutionary momentum and wrought changes in every day of our modern lives, ranging from autonomous lifestyles to decision-making scenarios. Among the different branches of ML activities, the medical field is notable, covering the field of detection and monitoring as well as the present status of diseases. Among the different medical diseases cancer is a serious threat. In particular, breast cancer is always a serious threat to women. Proper identification and then proper management and monitoring help the patient to recover from the disease, or at least help them to lead a better life. Proper identification and the current status of cancer largely depend on biomedical image analysis, a complex area of understanding. The analysis of these images requires special knowledge. The autonomous finding of Benign and Malignant information based on the images and making a Computer-Aided Diagnosis (CAD) system provide both the patient and the doctor with a second layer of confidence and allow them to make a more reliable decision. For the autonomous identification and detection of cancer, digital ML techniques have provided a revolutionary improvement. The recent development of the Deep Neural Network (DNN) and the logic-based algorithm make it possible to detect the target form the image more reliably. In this thesis we have investigated the performance of the DNN-based biomedical image classiffer as well as the Extreme Gradient Boosting (XGBoost)-based image classiffierfor the autonomous CAD system.
History
Table of Contents
1. Introduction -- 2. Involvement of machine Learning for breast-Cancer image classiffica-tion : a survey -- 3. Histopathological breast-cancer Image classiffication by Deep Neural Network Techniques guided by local clustering -- 4. Histopathological breast-image classiffication with image enhancement by convolutional neural network -- 5. Local and global feature utilisation for breast-image classiffication by convolutional neural network --6. Frequency-domain information along with LSTM and GRU methods for histopathological breast-image classiffication -- 7. Histopathological breast-image classiffication using local and frequency domains by convolutional neuralnNetwork -- 8. Histopathological breast-cancer Image classiffication with restricted Boltzmann machine along with back propagation -- 9. Histopathological breast-cancer Image classiffication with feature prioritisation -- 10. Conclusion and future work -- 11. List of abbreviations -- References.
Notes
Bibliography: pages 285-317
Thesis by publication.
Awarding Institution
Macquarie University
Degree Type
Thesis PhD
Degree
PhD, Macquarie University, Faculty of Science and Engineering, School of Engineering