Artificial Intelligence Techniques in Breast Cancer Diagnosis and Prognosis

Reviewed by:


A Jain et al (Eds)
Published by World Scientific (2000)
ISBN: 981-02-4374-X. 326 pages plus index.
RRP: A$137.48

Knowledge of clinical outcome, or prognosis, is integral to medical decision-making and resource prioritisation. The automated analysis of an individual patient’s records and genomic data to extract prognostic factors may be enhanced by the development of new statistical techniques and artificial intelligence methods. But can the application of artificial intelligence techniques improve diagnosis and prognostic estimates in the domain of breast cancer research? Has there been enough progress in the past decade to justify a first book on this subject? The editors of the series in Machine Perception and Artificial Intelligence obviously believe so, and they have done an excellent job of putting together a book that highlights the advances and controversies that surround the subject. Contributors from Australia, Germany, Italy and the USA combine the pathological, intelligent and statistical approaches to enable more precise definition of disease extent and prediction of tumour behaviour and response to treatment.

The first three chapters are keys to these approaches. Two chapters provide a brief introduction to breast cancer diagnosis, risk assessment and image feature extraction. The third chapter describes recent advances in prognostic and predictive techniques in breast cancer research. It demonstrates the ability of neural networks to effectively recognise and represent the complicated dependence of the disease on the range of demographic and clinical factors and to distinguish this dependence from the noise. These chapters are clear, concise, and potentially useful for teaching of both medical informatics and computer science.

A whole chapter is devoted to the MammoNet, a Bayesian networks-based tool for breast cancer. Authors spend a good deal of time carefully explaining its topology, system architecture and the use of its decision support features. The next chapter deals mainly with the use of continuous and categorical biological variables as prognostic biomarkers for breast cancer patients. The last three chapters review the application of artificial intelligence paradigms to computer-assisted interpretation of mammograms and cytodiagnosis of breast cancer, including computer vision techniques and the current state of the ongoing research program in the USA on Xcyt, a new system for remote cytological diagnosis of breast cancer. Perhaps one of the most interesting issues arising from this collection is the examination of different decision support technologies for their application in laboratory medicine.

A degree of knowledge in algebra, cancer biology and probability theory is assumed, but there is nothing much in here to frighten the non-computer scientist and the writing is easy to follow. The book will be of particular interest to clinical decision support systems designers and academic oncologists.


Be the first to know when a new issue is online. Subscribe today.