Page 265 - Handbook of Deep Learning in Biomedical Engineering Techniques and Applications

256 Chapter 9 Applications of deep learning in biomedical engineering

18. Biomedical image analysis

The applications of DL in biomedical image analysis are as
follows.

19. Image detection and recognition

Image detection deals with the problem of identifying or
detecting a certain portion/biomarker in a medical image, for
example, organ, region, and landmark localization [4]. A myriad
of DL applications in image detection are as follows:
1. Brain lesion detection in MRI images
2. Automatic lesion detection in fundus images
3. Gastric cancer detection
4. Detection of tumors
5. Detection of dendritic cells
6. Detecting malignant skin cells [16]
Image recognition (or image classification) is the process of
recognizing input images and classifying them in one of the pre-
defined distinct classes. The classifiers may be binary/multiclass.
DL algorithms have the potential to classify the normal and
abnormal images by learning 2D and 3D structure of a body
part. CNN is the prominent neural network architecture, and it
is compatible to perform classification. One instant application
of DL is classifying the images as benign, malignant, or nonneo-
plastic lesions. The main difficulty of deep learning in medical
image classification is the shortage of hand-featured samples [16].

20. Image acquisition and image
interpretation

In bio and medical imaging, the accurate of the diagnosis and/
or assessment of a disease depends on both image acquisition
and image interpretation. Large medical high-resolution imaging
acquisition systems are available, such as parallel MRI, multislice
CT, US transducer technology, digital PET, or 2D/3D X-ray [17].
These medical images contain a multitude of data, which are
conflict in between interpreters. The main part of the DL applica-
tions in bio and medical imaging concerns the computer-aided
interpretations and analyses, for example, analyzing histopathol-
ogy images for breast cancer diagnosis or analyzing the patholog-
ical images with a focus on research and biomarker discovery [18].