Page 39 - Handbook of Deep Learning in Biomedical Engineering Techniques and Applications

Chapter 2 Deep convolutional neural network in medical image processing 27

performance measures such as precision, specificity, sensitivity,
F-measure, and accuracy. As the medical data are growing day
by day, an appropriate method is required to analyze them. In
this condition, DL techniques are shown to be a supreme candi-
date for learning medical data effectively. DL imitates the tech-
niques of the human brain [5]. DL has a deep design that
comprises several layers of the transformation of the data, which
is quite similar to the working of the human brain [6].
Extraction of the most appropriate features requires deep in-
formation of the basic features in data gathering. The process
of feature extraction is the most tedious and challenging task
when the amount of data to be handled is quite large. The
most important advantage of DL is its capability to learn complex
features directly from the raw data. This gives the freedom of con-
structing a system that does not depend on manually extracted
features such as in ML techniques [7]. This asset of DL has
attracted many researchers to work on the domain of medical im-
aging. Fig. 2.2 shows the basic architecture of DL, which can learn
the features automatically. The future of medicine can highly
benefit from the recent trends of DL implemented in medical im-
aging. But the major challenge is the limited medical knowledge of
DL experts in application viewpoint and limited DL knowledge of
the medical professionals [7]. A current tutorial makes an effort to
reduce the gap between medical and engineering professionals by
giving a detailed procedure of implementing DL to digital patho-
logical images [8]. A detailed knowledge of medical image seg-
mentation using the concept of DL is provided in Ref. [9] for
professionals to get a better idea.
This chapter gives a brief idea about the recent procedure of
medical imaging that uses the concept of deep convolutional
neural network (CNN). In Section 2, the process of medical image
analysis is discussed. In Section 3, a brief idea on CNN is given fol-
lowed by various architectures of CNN. In Section 4, the applica-
tion of CNN to different organs of human body imaging is

Figure 2.2 The basic architecture of deep learning.