Artificial Intelligence in Cancer Diagnosis and Prognosis, Volume 3: Brain and prostate cancer

Within this third volume dealing with breast and bladder cancer, the editors and authors will detail the latest research related to the application of AI to cancer diagnosis and prognosis and summarize its advantages. It's the editors and authors intention to explore how AI assists in these activities, specifically with regard to its unprecedented accuracy, which is even higher than that of general statistical applications in oncology. Ways will also be demonstrated as to how these methods in AI are advancing the field. There have been thousands of papers written between 1995 and 2019 related to AI for cancer diagnosis and prognosis. However, to this date (and unknown to the Editors) there has not yet been published a comprehensive overview of the latest findings pertaining to these AI technologies, within a single book project(s). Therefore, the purpose of this three volume work and particularly for this third volume dealing with brain and bladder cancer, is to present a compendium of these findings related to these two pervasive cancers. Within this coverage it's our hope that scientists, researchers and clinicians can successfully incorporate these techniques into other significant cancers such as pancreatic, esophageal, leukemia, melanoma, etc. PRELIMS.pdf Preface Acknowledgements Editor biographies Ayman El-Baz Jasjit S Suri List of contributors Outline placeholder Adeel Ahmed Abbasi Nahla B Abdel-Hamid H Arafat Ali Sarah M Ayyad Samir Kumar Bandyopadhyay Gustavo M Callico Daniel U Campos-Delgado Inés Alejandro Cruz-Guerrero Dimitrios E Diamantis Shawni Dutta Mohamed Abou El-Ghar Moumen El-Melegy Himar Fabelo Davide Fontanarosa Matthew Foote Mohamed Ghazal Preetam Ghosh Vishal Goyalis Cheng-Yeh Hsieh Lal Hussain Jiwoong Jason Jeong Rishabh Kapoor Ali Keles Ayturk Keles Labib M Labib Rui Li Tian Liu Zecheng Liu Chung-Ming Lo Yeh-Chi Lo Ali Mahmoud Hui Mao Aldo Rodrigo Mejia-Rodríguez Akash Mehta Serafeim Moustakidis Charis Ntakolia Samuel Ortega Jatinder R Palta Elpiniki I Papageorgiou Nikolaos Papandrianos Ben Perrett Mark Pinkham Prabhakar Ramachandran Venkatakrishnan Seshadri Ahmed Shalaby Mohamed Shehata Ren-Dih Sheu William C Sleeman IV Sriram Srinivasan Richard Stock James Tam Zhen Tian Tzu-Chi Tseng Jia Wei Lei Yang Xiaofeng Yang Wenguang Yuan Yading Yuan CH001.pdf Chapter 1 Artificial Intelligence in prostate cancer treatment with image-guided radiation therapy 1.1 Introduction 1.1.1 External radiation therapy for prostate cancer 1.1.2 Brachytherapy for prostate cancer: radioactive seed implants 1.2 Deep contouring: automated multiple organ segmentation using dilated U-Net with generalized Jaccard distance 1.2.1 Introduction 1.2.2 Methodology 1.2.3 Experiments 1.2.4 Summary 1.3 Deep planning: fully 3D-knowledge-based treatment planning 1.3.1 Introduction 1.3.2 Methodology 1.3.3 Experiments 1.3.4 Summary 1.4 Conclusions References CH002.pdf Chapter 2 Artificial-intelligence-based diagnosis of brain tumor diseases 2.1 Introduction 2.2 Related works 2.3 Current methods used to collect images 2.3.1 Ultrasound (USG) 2.3.2 Projection radiography (x-rays) 2.3.3 Computed tomography 2.3.4 Magnetic resonance imaging 2.3.5 Positron emission tomography 2.4 Background 2.4.1 Artificial intelligence and machine learning 2.4.2 Performance evaluation metrics 2.5 Datasets of brain tumors 2.6 Proposed methodologies for disease detection 2.6.1 Brain tumor detection methodology 2.7 Experimental results 2.8 Conclusions References CH003.pdf Chapter 3 Multisite brain tumor segmentation using a unified generative adversarial network 3.1 Introduction 3.2 UGAN 3.2.1 Method overview 3.2.2 Loss function 3.3 Experiments 3.3.1 Datasets 3.3.2 Training settings 3.3.3 Segmentation performances 3.4 Conclusions References and further reading CH004.pdf Chapter 4 Role of artificial intelligence in automatic segmentation of brain metastases for radiotherapy 4.1 Introduction 4.1.1 Brain metastasis treatment options 4.2 Manual segmentation of tumors 4.2.1 Limitations of manual segmentation 4.3 Automatic segmentation 4.3.1 Automatic segmentation techniques 4.3.2 U-Net 4.3.3 Identification of small lesions 4.3.4 Post-treatment volumetric assessment 4.3.5 Post-treatment response prediction 4.3.6 Post-treatment radionecrosis 4.4 Summary References and further reading CH005.pdf Chapter 5 Applications of artificial intelligence in the fields of brain and prostate cancer Abbreviations 5.1 Introduction 5.2 AI applications in brain cancer 5.2.1 Brain tumor segmentation 5.2.2 Survival prognosis 5.2.3 Surgical performance 5.3 AI applications in prostate cancer 5.3.1 Analyzing histopathological images 5.3.2 PCa segmentation 5.3.3 Robotic surgery 5.3.4 PCa treatment 5.4 Conclusions Acknowledgments References CH006.pdf Chapter 6 AI-based non-deep learning and deep learning techniques used to accurately predict prostate cancer 6.1 Introduction 6.2 Study data 6.2.1 Dataset 6.3 AI-based non-deep-learning prediction methods 6.3.1 Handcrafted features 6.3.2 Classification algorithms 6.4 AI-based deep learning prediction methods 6.4.1 Convolutional neural network (CNN) overview 6.4.2 CNN methods 6.4.3 CNN layers 6.4.4 Training/testing data formulation 6.4.5 Performance evaluation measures 6.4.6 Receiver operating characteristic curve 6.5 Results and discussion 6.6 Conclusions and future recommendations References CH007.pdf Chapter 7 Intelligent brain tumor classification using deep convolutional neural networks with transfer learning 7.1 Introduction 7.2 Materials and methods 7.2.1 MR images 7.2.2 Image analysis 7.2.3 Transfer learning 7.2.4 Data augmentation 7.2.5 Results 7.2.6 Discussion 7.3 Conclusions References CH008.pdf Chapter 8 Big data applications in radiation oncology: challenges and opportunities 8.1 Introduction 8.2 Methods for structure set standardization 8.2.1 Overview 8.2.2 DICOM structure set standardization methods 8.2.3 Results 8.3 The use of natural language processing with medical texts 8.3.1 NLP feature extraction and models 8.3.2 NLP implementation results 8.3.3 Challenges for NLP in understanding free text 8.4 Standardization through structured templates 8.4.1 Manual data extraction 8.4.2 Analytic dashboard 8.4.3 Limitations of automated data extraction 8.4.4 Health Information Gateway Exchange (HINGE) 8.5 Future directions in data standardization and aggregation 8.5.1 Retrospective data 8.5.2 Transfer learning 8.5.3 Federated learning 8.6 Conclusions References CH009.pdf Chapter 9 A hybrid approach to the hyperspectral classification of in vivo brain tissue: linear unmixing with spatial coherence and machine learning 9.1 Introduction 9.2 Intraoperative HS acquisition system and HS dataset 9.2.1 Data preprocessing 9.3 Processing framework based on linear unmixing with spatial coherence and machine learning 9.3.1 Abundances estimation 9.3.2 End-members estimation 9.3.3 Internal abundances estimation 9.3.4 Machine learning for classification 9.4 Hybrid classification methodology 9.5 Experimental results and discussion 9.5.1 Evaluation of the hybrid classification methodology 9.5.2 Comparison with other related works 9.5.3 Limitations 9.6 Conclusions References CH010.pdf Chapter 10 Application and post-hoc explainability of deep convolutional neural networks for bone cancer metastasis classification in prostate patients 10.1 Introduction 10.2 Computer-aided diagnosis (CAD) system 10.2.1 Study population 10.2.2 Explainable deep learning pipeline for diagnosis 10.3 Results 10.3.1 Bone metastasis classification results 10.3.2 Post-hoc explainability results 10.4 Discussion 10.5 Conclusions References CH011.pdf Chapter 11 Prostate cancer detection using histopathology image analysis 11.1 Introduction 11.2 Histopathological images 11.3 Handcrafted feature-based CAD 11.4 Deep learning-based CAD 11.5 Conclusions Acknowledgments References CH012.pdf Chapter 12 Machine learning of gliomas in 3D dynamic contrast enhanced MRI: automatic segmentation and classification 12.1 Introduction 12.2 Segmentation and classification methods 12.2.1 Segmentation method 12.2.2 Automatic classification system 12.3 Results 12.3.1 Segmentation results 12.3.2 Classification results 12.4 Discussion 12.4.1 Comparison of segmentation methods 12.4.2 Correlation thresholds and feature lists 12.4.3 Classification results using positive features 12.4.4 Significant radiomics features 12.4.5 Limitations 12.5 Conclusions References Within this third volume dealing with brain and prostate cancer, the editors and authors detail the latest research related to the application of artificial intelligence (AI) to cancer diagnosis and prognosis and summarize its advantages. It is the intention of the editors and authors to explore how AI assists in these activities, specifically with regard to its unprecedented accuracy, which is even higher than that of general statistical applications in oncology. Ways will also be demonstrated as to how these methods in AI are advancing the field. There have been thousands of papers written between 1995 and 2019 related to AI for cancer diagnosis and prognosis. However, to date (to the best of our knowledge) there has not yet been published a comprehensive overview of the latest findings pertaining to these AI technologies, within a single book project. Therefore, the purpose of this three-volume work, and particularly for this third volume dealing with brain and prostate cancer, is to present a compendium of these findings related to these two pervasive cancers. Within this coverage it is our hope that scientists, researchers and clinicians can successfully incorporate these techniques into other significant cancers such as pancreatic, esophageal, leukemia, melanoma, etc. Part of IPEM-IOP Series in Physics and Engineering in Medicine and Biology