Over the past few years, deep learning has been a drastic change in how images are being analyzed with ever increasing accuracy particularly when it comes to tasks like image classification, image segmentation etc. This part elucidates the major principles and basic architectures of deep learning and focuses on its purpose to understand images. The first part of this chapter defines and explains neural networks and is particularly concerned with the specifics of deep learning as a trend of machine learning while fully or partially eliminating the need on any prior feature engineering. This chapter lays out the intricacies of training of Convolutional Neural Networks (CNNs), the most popular architecture for tasks associated with images. CNNs learn the spatial hierarchies and patterns of images making such networks critical for performing tasks like object detection, image recognition and segmentation. Important also is understanding what comprises CNNs described one of the chapter's sections including convolutional layers, pooling, fully connected layers and how purposes of those components assist in image processing effectively. Next along the lines of following developments in medical imaging technologies, what is covered in the chapter is also dedicated to other modern methods for biomedical images processing, such as U-Net and Fully Convolutional Networks (FCN), which are used for pixel-wise images segmentation. Certain special focus is directed on the use of these models for classification and segmentation of white blood cells, which is an important and difficult area of clinical diagnostics. The chapter provides key challenges that come with the application of a deep learning approach towards the image analysis, including but not limited to overfitting, data imbalance, and an absence of interpretability. Possible solutions for these challenges, especially in contexts with little labelled data will also be reviewed, including the use of dropout, data augmentation and transfer learning. In conclusion, the chapter delineates the probable future developments in the field, which include attention mechanisms and self-supervised learning technologies which will take deep learning models for image analysis a step further than their current state. As a result of the defining of the key features of the deep learning application as well new approaches, this chapter seeks to create an understanding of the impact that deep learning is having on image analysis and imaging in general especially in areas such as biosciences and other medical applications.