Current Applications of Deep Learning in Cancer Diagnostics

Cover
Half Title
Title Page
Copyright Page
Table of Contents
List of Figures
List of Tables
Introduction
Contributors
Chapter 1: Contemporary Trends in the Early Detection and Diagnosis of Human Cancers Using Deep Learning Techniques
	1.1 Introduction
	1.2 Deep Learning Architectures Commonly Used for Cancer Diagnosis
		1.2.1 Artificial Neural Networks (ANNs)
		1.2.2 Convolutional Neural Networks (CNNs)
	1.3 Use of Deep Learning in Cancer Diagnosis
	1.4 Results and Discussion
	1.5 Conclusion
	References
Chapter 2: Cancer Data Pre-Processing Techniques
	2.1 Introduction
	2.2 Cancer Types
		2.2.1 Cervical Cancer
		2.2.2 Liver Cancer
		2.2.3 Breast Cancer
		2.2.4 Lung Cancer
		2.2.5 Colorectal Cancer
		2.2.6 Oral Cancer
	2.3 Data Collection Modes
		2.3.1 Magnetic Resonance Imaging (MRI) Data
		2.3.2 Computed Tomography (CT) Scan Image Data
		2.3.3 X-ray Image Data
		2.3.4 Ultrasound Image Data
		2.3.5 Gene Expression Data
		2.3.6 Text Data
	2.4 Common Pre-Processing Techniques Applicable for Cancer Data
		2.4.1 MRI Data
			2.4.1.1 Intensity Inhomogeneity Correction
			2.4.1.2 Registration
			2.4.1.3 Segmentation
			2.4.1.4 Slice Timing Correction
			2.4.1.5 Motion Correction
			2.4.1.6 Nuisance Variable Removal
			2.4.1.7 Filtering
			2.4.1.8 Spatial Smoothing
		2.4.2 CT Scan Image Data
			2.4.2.1 Denoising
			2.4.2.2 Interpolation
			2.4.2.3 Registration
			2.4.2.4 Normalization
		2.4.3 X-ray Image Data
			2.4.3.1 Adaptive Contrast Enhancement
			2.4.3.2 Region Localization
		2.4.4 Ultrasound Image Data
			2.4.4.1 Deblurring
			2.4.4.2 Resolution Enhancement
			2.4.4.3 Denoising
		2.4.5 Gene Expression Data
			2.4.5.1 Scale Transformations
			2.4.5.2 Management of Missing Values
			2.4.5.3 Replicate Handling
		2.4.6 Text Data
			2.4.6.1 Data Cleaning
			2.4.6.2 Data Reduction
			2.4.6.3 Normalization
			2.4.6.4 Discretization and Concept Hierarchy Generation
	2.5 Conclusions
	References
Chapter 3: A Survey on Deep Learning Techniques for Breast, Leukemia, and Cervical Cancer Prediction
	3.1 Introduction
		3.1.1 Breast Cancer
		3.1.2 Leukemia
		3.1.3 Cervical Cancer
	3.2 Literature Survey
		3.2.1 Deep Learning Methods for Leukemia Prediction
		3.2.2 Machine Learning Methods for Cervical Cancer Prediction
		3.2.3 Deep Learning Methods for Breast Cancer Prediction
	3.3 Conclusion
	References
Chapter 4: An Optimized Deep Learning Technique for Detecting Lung Cancer from CT Images
	4.1 Introduction
	4.2 Literature Review
	4.3 Design Approach and Details
		4.3.1 Basic CNNs
		4.3.2 Convolutional Layer and Sub-Sampling Method
	4.4 Proposed CNN Architecture
		4.4.1 Data Augmentation
	4.5 Experimental Analysis
		4.5.1 Parameter Setting
	4.6 Conclusion
	References
Chapter 5: Brain Tumor Segmentation Utilizing MRI Multimodal Images with Deep Learning
	5.1 Introduction
	5.2 Material and Methods
		5.2.1 Pre-Processing
		5.2.2 Tumor Representation in Each Slice
		5.2.3 Finding the Expected Area of the Tumor
		5.2.4 Deep Learning Architecture
		5.2.5 Proposed Structure
		5.2.6 Distance-Wise Attention Module
		5.2.7 Cascade CNN Model
	5.3 Experiments
		5.3.1 Data and Implementation Details
		5.3.2 Evaluation Measure
		5.3.3 Experimental Results
	5.4 Conclusion
	References
Chapter 6: Detection and Classification of Brain Tumors Using Light-Weight Convolutional Neural Network
	6.1 Introduction
	6.2 Related Works
	6.3 Dataset Detail
	6.4 Methodology
		6.4.1 Detection of Brain Tumor
		6.4.2 Classification of Brain Tumor
	6.5 Results and Discussions
		6.5.1 Comparison of the Proposed Approach with Other Light-Weight Models
	6.6 Conclusion
	References
Chapter 7: Parallel Dense Skip-Connected CNN Approach for Brain Tumor Classification
	7.1 Introduction
	7.2 Parallel Dense Skip-Connected CNN (PDSCNN)
	7.3 Results and Discussion
		7.3.1 Network Training Parameters
		7.3.2 Brain Tumor MRI Dataset
		7.3.3 Tumor Identification Accuracies
		7.3.4 Confusion Matrices
	7.4 Conclusion
	References
Chapter 8: Liver Tumor Segmentation Using Deep Learning Neural Networks
	8.1 Introduction
	8.2 Prior Work
	8.3 Proposed Solution and Architecture
		8.3.1 Data Set Used
		8.3.2 FastAI Library
		8.3.3 U-Net Architecture
		8.3.4 Employed Dynamic U-Net with ResNet34 Encoder
		8.3.5 Data Preprocessing
		8.3.6 Proposed Architecture and Methodology
		8.3.7 Model Training Metrics
	8.4 Model Analysis
	8.5 Model Specifications and Runtime Analysis
	8.6 Performance Evaluation
		8.6.1 Dice Similarity Coefficient
		8.6.2 Comparative Analysis
	8.7 Conclusion
	References
Chapter 9: Deep Learning Algorithms for Classification and Prediction of Acute Lymphoblastic Leukemia
	9.1 Introduction
	9.2 Related Works
	9.3 Dataset Description
	9.4 Methodology
	9.5 Results and Discussion
	9.6 Conclusion
	Details of Authors
	References
	Note
Chapter 10: Cervical Pap Smear Screening and Cancer Detection UsingDeep Neural Network
	10.1 Introduction
	10.2 Related Work
	10.3 Methodology
	10.4 Dataset
		10.4.1 Normal
		10.4.2 Abnormal
		10.4.3 Benign
	10.5 Experimental Results
	10.6 Conclusion
	References
Chapter 11: Cancer Detection Using Deep Neural Network Differentiation of Squamous Carcinoma Cells in Oral Pathology
	11.1 Histopathology – A Review
	11.2 Computer Vision in Feature Extraction
	11.3 Deep Neural Nets for Cancer Diagnosis
	11.4 Differential Diagnosis in Oral Pathology
		11.4.1 Convolutional Neural Network
			11.4.1.1 Convolution Operation
			11.4.1.2 Pooling Operation
			11.4.1.3 Decision Function
			11.4.1.4 Normalization
			11.4.1.5 Dropout
			11.4.1.6 Fully Connected Layer
	11.5 Automated Detection and Grading of Squamous Cell Carcinoma forDiagnosis of Oral Cancer
		11.5.1 Problem Statement
		11.5.2 Objectives
		11.5.3 Methodology/Experimental Design and Sampling Strategy
		11.5.4 Methodology/Experimental Design and Sampling Strategy
		11.5.5 Methodology/Experimental Design
		11.5.6 Methodology/Experimental Design
		11.5.7 Performance Measures and Metrics
			11.5.7.1 Recall
			11.5.7.2 Dice Similarity Coefficient (DSC)
			11.5.7.3 Intersection over Union (IOU)
			11.5.7.4 Confusion Matrix (CM)
			11.5.7.5 Accuracy
			11.5.7.6 F1 Score
	11.6 Research Challenges in Digital Pathology
	11.7 Conclusion
	References
Chapter 12: Challenges and Future Scopes in Current Applications of Deep Learning in Human Cancer Diagnostics
	12.1 Introduction
		12.1.1 Challenges in Deep Learning
		12.1.2 Advantages of Deep Learning
		12.1.3 Current Application of Deep Learning in Cancer Prognosis
	12.2 Neural Networks and Their Types
		12.2.1 Non-Feature-Extracted NN Models
		12.2.2 Creation of Fully Connected NNs by Extracting Features from Gene Expression Data
		12.2.3 CNN-Based Models
		12.2.4 Cancer Imaging with Convolutional Neural Networks
		12.2.5 Digital Pathology
		12.2.6 Electronic Medical Records (EMRs)
		12.2.7 Deep Learning and Artificial Neural Networks (DL) in Healthcare
	12.3 Challenges and Opportunities of Deep Learning in Cancer Diagnostics
		12.3.1 Enhancement of Features
			12.3.1.1 Federated Inference
			12.3.1.2 Model Privacy
			12.3.1.3 Incorporating Expert Knowledge
			12.3.1.4 Temporal Modeling
			12.3.1.5 Interpretable Modeling
	12.4 Conclusion
	12.5 Acknowledgment
	12.6 Conflict of Interest
	12.7 Funding Statement
	References
Index