Deep Learning and its Impact on Medical Image Analysis

By: , Posted on: November 30, 2016


Deep Learning for medical image analysis has a growing impact on medical imaging, we talk to the Editors of Deep Learning for Medical Image Analysis to find out more about their latest book.

Q: Deep Learning is having a major impact in computer vision and now in medical imaging. Can you explain why it is having such an impact?

A: Deep learning is one of the most effective machine learning methods in a supervised learning setting. The term deep learning implies the use of a deep neural network model. The basic computational unit in a neural network is the neuron, a concept inspired by the study of the human brain, which takes multiple signals as inputs, combines them linearly using weights, and then passes the combined signals through nonlinear operations to generate output signals. A deep neural network hierarchically stacks multiple layers of neurons, forming a hierarchical feature representation. The number of layers now extends to over 1,000! With such a gigantic modeling capacity, a deep network can essentially “memorize” all possible mappings after successful training with a sufficiently large knowledge database and make “intelligent” predictions e.g. interpolations and/or extrapolations for unseen cases. Thus, deep learning is generating a major impact in computer vision and medical imaging. In fact, similar impact is happening in domains like text, voice, etc.

 Q: How do you see the use of Deep Learning continuing in medical imaging?

A: We observe that the momentum is definitely picking up. For example, the number of deep learning related MICCAI papers almost doubled (52 papers in MICCAI 2016 vs 23 papers in MICCAI 2015). The typical applications that involve deep learning include medical image detection, segmentation, registration, computer aided diagnosis, etc. In terms of imaging modality, the use of deep learning spans to all different modalities. We believe that this trend will stay strong and most likely become even stronger and extend, beyond the abovementioned applications, into the fields like reconstruction, intervention, etc.

Editor Kevin Zhou pictured with Nicholas Ayache, Editor in Chief, Medical Image Analysis Journal who has written the forward for the book.
Editor Kevin Zhou pictured with Nicholas Ayache, Editor in Chief, Medical Image Analysis Journal who has written the foreword for the book.

 Q: What was your objective in developing this book?

A: The objective was to provide interested readers the access to foundational knowledge about (deep) neural networks as well as the state-of-the-art approaches that integrate deep learning into real solutions in the medical imaging domain. The hope is that after reading the book, the readers are adequately equipped to start their own research and development of deep learning based solutions to their own problems.

 Q: What approach does the book take? What application areas does it cover?

A: The book begins with introductory chapters on (deep) neural networks, followed by chapters that utilize deep learning methods to provide real solutions to various applications. Examples of such applications include medical image detection, recognition, segmentation, and registration, computer-aided diagnosis and disease quantification, and several other miscellaneous applications.

Q: How will readers find it useful to their research or study?

A: Our hope is that after reading the book, the readers are adequately equipped and ready to embark on their own research and development of deep learning based solutions to their specific problem domains

Q: What do you think are the future challenges of using Deep Learning methods for medical imaging?

A: There are still many challenges ahead in using Deep Learning methods for medical imaging. For example, the current methods are supervised in nature, relying on the existence of a large number of annotated datasets. In case of rare disease, it is difficult to obtain a sufficient number of cases. Also it is expensive to have all images manually annotated. One challenge is therefore how to perform effective deep learning using a small number of training examples and/or without supervision. A neural network in the current form is mostly a black box, lacking the desired interpretation for medical image diagnosis. Last but not the least, deep learning models should not replace and should be integrated into existing knowledge (logic, graphical models, ontology, etc.). How to do this is an important research topic.

Editors Kevin Zhou, Hayit Greenspan and Dinggang Shen

deep learning for medical image analysis

Deep Learning for Medical Imaging

  • Covers common research problems in medical image analysis and their challenges
  • Describes deep learning methods and the theories behind approaches for medical image analysis
  • Teaches how algorithms are applied to a broad range of application areas, including Chest X-ray, breast CAD, lung and chest, microscopy and pathology, etc.

Save up to 30% on your own copy when you order via the Elsevier Store. Enter STC215 at the checkout.

Electronics & Electrical Engineering

Electronics and electrical engineering have practically limitless applications. From power engineering, telecommunications, and consumer electronics to circuit design, computer engineering, and embedded systems, these disciplines form the backbone of our increasingly tech-dependent world. Elsevier’s collection of electronics and electrical engineering content — particularly our Newnes and Academic Press Imprints — encompasses these areas and more. Our books and journals provide fundamental knowledge and practical, up-to-date toolkits for professional engineers and technicians, undergraduate and postgraduate students, and electronics enthusiasts.