Artificial neural networks in cancer diagnosis, prognosis, and patient management / [edited by] Raouf N.G. Naguib and Gajanan V. Sherbet.

Format:

Book

Contributor:

Naguib, Raouf N. G.

Sherbet, G. V. (Gajanan V.)

Series:

Biomedical engineering series (Boca Raton, Fla.)

Biomedical engineering series

Language:

English

Subjects (All):

Cancer--Computer simulation.

Cancer.

Neural networks (Computer science).

Artificial intelligence--Medical applications.

Artificial intelligence.

Neoplasms--diagnosis.

Diagnosis, Computer-Assisted.

Neoplasms--therapy.

Neural Networks, Computer.

Prognosis.

Computer simulation.

Medical Subjects:

Neoplasms--diagnosis.

Diagnosis, Computer-Assisted.

Neoplasms--therapy.

Neural Networks, Computer.

Prognosis.

Physical Description:

187 pages : illustrations ; 26 cm.

Place of Publication:

Boca Raton : CRC Press, 2001.

Summary:

The use of artificial neural networks (ANNs) has proliferated over the past few years, and the medical applications of ANNs have touched on a variety of areas. However, their application to cancer research is just beginning to be realized. Research and case studies are mostly scattered throughout various journals. This volume brings together the work of top researchers-primarily clinicians-who present the results of their state-of-the-art work with ANNs as applied to nearly all major areas of cancer for diagnosis, prognosis, and management of the disease

Contents:

II. Artificial Neural Networks 2

Chapter 2 Analysis of Molecular Prognostic Factors in Breast Cancer by Artificial Neural Networks / B. Angus, T.W.J. Lennard, R.N.G. Naguib, G.V. Sherbet 9

II. Prognostic Factors in Breast Cancer 10

A. Established Prognostic Factors in Clinical Use: Stage, Grade, and Size 10

B. Hormone Receptors and Oestrogen Regulated Proteins 11

C. Markers of Cellular Proliferation and Cell Cycle Regulators 12

D. p53 and Related Proteins 12

E. Type 1 Growth Factor Receptors 12

III. Prediction of Nodal Metastasis by Neural Analysis 13

IV. Proteins Associated with Metastatic Potential 14

A. h-mts1 and not nm23 Expression Correlates with Nodal Sread of Cancer 15

B. h-mts1 Expression and ER/PgR Status 16

V. Anaylsis of h-mts1 and nm23 Expression by Artificial Neural Networks 16

Chapter 3 Artificial Neural Approach to Analysing the Prognostic Significance of DNA Ploidy and Cell Cycle Distribution of Breast Cancer Aspirate Cells / R.N.G. Naguib, G.V. Sherbet 23

II. Breast Cancer Fine-Needle Aspirates 23

III. Analysis of Image Cytometric Data Using Artificial Neural Networks 24

Chapter 4 Neural Networks for the Estimation of Prognosis in Lung Cancer / H. Esteva, M. Bellotti, A.M. Marchevsky 29

II. Carcinoma of the Lung 29

III. Medical Applications of Artificial Neural Networks 30

IV. Artificial Neural Networks 31

A. Architecture 31

B. Training Artificial Neural Network Systems 31

C. Estimating the Reliability of Artificial Neural Networks 32

V. Practical Application of Neural Networks in the Prognosis of Lung Cancer Patients 32

A. Tumour Type 33

B. Ploidy Levels 33

C. Immunohistochemical Proliferation Markers 33

D. Neural Processing 34

Chapter 5 The Use of a Genetic Algorithm Neural Network (GANN) for Prognosis in Surgically Treated Nonsmall Cell Lung Cancer (NSCLC) / M.F. Jefferson, N. Pendleton, S.B. Lucas, M.A. Horan 39

A. The NSCLC Data 40

III. Mathematical Method: The Genetic Algorithm Neural Network 40

A. Selection of Predictors by Genetic Algorithm 40

B. Classification 41

C. Choosing the Optimal Genome for Use with a Neural Network 41

D. Updating the Optimal Information Genome 41

E. Defining the Neural Network Architecture 41

F. Training the Neural Network 41

G. Stopping Rule for the Training Process 41

H. Classification Solution 42

I. Application of Bayes' Theorem 42

J. Computational Method 42

IV. Statistical Methods 42

A. Predictive Statistics 42

B. Comparison of Predictive Statistics 42

V. Survival 43

A. Survival by Logistic Regression 44

B. Survival Classification by GANN 44

C. Comparison of Logistic Regression and GANN for Classification of Survival 45

A. Interpretation of Clinical Findings 48

B. Methodological Considerations 49

1. Classification of Survival Outcome 49

2. Losses and Gains in Information 50

3. Adaptability and Retraining 51

4. Variable Selection by the GA 52

5. Performance of Genetic Algorithm Neural Network Compared with Logistic Regression 52

6. Adaptations in the Genetic Algorithm Neural Network Method: Simulating Preprocessing in Basic Neural Systems 52

7. Components of Methodological Performance 52

Chapter 6 The Use of Machine Learning in Screening for Oral Cancer / P.M. Speight, P. Hammond 55

I. Epidemiology 55

II. Clinical Features and Diagnosis 55

III. Screening for Oral Cancer 56

IV. Selection of High-Risk Groups Using a Neural Network 57

V. Training and Testing the Neural Network 57

VI. Comparison of Neural Networks and Other Machine Learning Techniques 60

VII. Data Visualisation in Clementine 60

VIII. Inducing Models in Clementine 64

IX. Evaluating the Potential Performance of Machine Learning for Detection of High-Risk Individuals 68

Chapter 7 Outcome Prediction of Oesophago-Gastric Cancer Using Neural Analysis of Pre- and Postoperative Parameters / J. Wayman, S.M. Griffin 73

II. Artificial Neural Networks for the Prediction of Outcome of Oesophago-Gastric Cancer 74

Chapter 8 Artifical Neural Networks in Urologic Oncology / T.H. Douglas, J.W. Moul 93

II. Kidney Cancer 94

III. Bladder Cancer 95

IV. Prostate Cancer 95

V. Testicular Cancer 98

Chapter 9 Neural Networks in Urologic Oncology / C. Niederberger, D. Ridout 103

II. Renal Cell Carcinoma 103

III. Prostate Cancer 107

IV. Bladder Cancer 110

V. Testicular Carcinoma 111

Chapter 10 Comparison of a Neural Network with High Sensitivity and Specificity of Free/Total Serum PSA for Diagnosing Prostate Cancer in Men with a PSA < 4.0 ng/mL / T.A. Stamey, S.D. Barnhill, Z. Zhang, C.M. Yemoto, H. Zhang, K.R. Madyastha 115

II. Material and Methods 116

A. Patient Selection 116

B. Laboratory Analysis 116

C. Neural Network Inputs and Derivation of the PI 116

III. Results 116

Chapter 11 Artificial Neural Networks and Prognosis in Prostate Cancer / F.C. Hamdy 125

II. Prostate Cancer 126

III. Patients and Methods 126

A. Patients 126

B. Methods 126

1. Immunohistochemistry 126

2. Artificial Neural Networks 127

3. Statistical Analysis 127

IV. Results 127

A. Immunohistochemistry 127

B. Neural Network Analysis 127

C. Comparison of ANNs with Statistical Analysis 129

Chapter 12 Comparison between Urologists and Artificial Neural Networks in Bladder Cancer Outcome Prediction / K.N. Qureshi, J.K. Mellon 133

II. Materials and Methods 135

III. Results 136

A. Clinical Outcome: Recurrence, Progression, and Survival 136

B. Predictions by the ANN and Clinicians 136

Chapter 13 A Probabilistic Neural Network Framework for the Detection of Malignant Melanoma / M. Hintz-Madsen, L.K. Hansen, J. Larsen, K.T. Drzewiecki 141

A. Malignant Melanoma 141

B. Evolution of Malignant Melanoma 142

C. Image Acquistion Techniques 142

1. Traditional Imaging 142

2. Dermatoscopic Imaging 142

D. Dermatoscopic Features 143

II. Feature Extraction in Dermatoscopic Images 145

A. Image Acquisition 145

B. Image Preprocessing 146

1. Median Filtering 147

2. Karhunen-Loeve Transform 148

C. Image Segmentation 149

1. Optimal Thresholding 149

D. Dermatoscopic Feature Description 152

1. Asymmetry 152

2. Edge Abruptness 154

3. Colour 156

III. A Probabalistic Framework for Classification 160

A. Bayes' Decision Theory 160

B. Measuring Model Preformance 161

1. Cross-Entropy Error Function for Multiple Classes 163

C. Measuring Generalisation Performance 163

1. Empirical Estimates 164

2. Algebraic Estimates 165

D. Controlling Model Complexity 165

1. Weight Decay Regularisation 166

2. Optimal Brain Damage Pruning 166

IV. Neural Classifier Modelling 167

A. Multilayer Perception Architecture 168

1. Softmax Normalisation 168

2. Modified Softmax Normalisation 169

B. Estimating Model Parameters 170

1. Gradient Descent Optimisation 172

2. Newton Optimisation 172

C. Overview of Design Algorithm 173

V. Experiments 173

A. Experimental Setup 173

B. Results 174

1. Classifier Results 174

2. Dermatoscopic Feature Importance 178

A. Dermatoscopic Feature Extraction 180

B. Probabilistic Framework for Classification 180

C. Neural Classifier Modelling 180

D. The Malignant Melanoma Classification Problem 181.

Notes:

Includes bibliographical references and index.

ISBN:

0849396921

OCLC:

46402503