Page 267 - Handbook of Deep Learning in Biomedical Engineering Techniques and Applications

258 Chapter 9 Applications of deep learning in biomedical engineering

3. benign from malignant cell clusters, and
4. benign and malignant breast lesions [19].
Histopathology or histology defines the process of examining
tissues using microscope. It is the study of tissues concerning
diseased one. Employment of DL in histopathological images
assists human pathologists to overcome manual examination of
suspected tissue. The malignant markers that are used by neural
network to identify pathological images such as
1. increased nucleus to cytoplasm ratios,
2. increased numbers of mitotic figures indicating increased cell
replication,
3. a typical cellular architecture,
4. signs of cellular necrosis, and
5. high cell proliferation index from molecular markers like
Ki-67 [16].

23. Brain, body, and machine interface

23.1 Brainemachine interface
The activity of manipulating an artificial device by extracting
neural impulses of the brain is referred as brainemachine
interface (BMI). The interface that helps to decode and process
these signals for controlling an external device is known as neuro-
prosthetic device. The neuroprosthetic devices (e.g., cochlear
implant) can be implemented in real-time applications by
connecting with the central nervous system, for example, restora-
tion of audition for patients having hearing disorders. The neuro-
pathetic devices may be invasive or noninvasive [20].
Biofeedback: This is the process in which the brain generates
control signals according to the external stimuli and adapts the
body response to it [20].
The BMI uses biofeedback in applications regarding visual,
auditive, and tactile or mixture of them.
The scheme of a typical BMI is shown in Fig. 9.9.

24. Classification of the brainemachine
interfaces

The BMIs are classified into invasive and noninvasive
techniques.