Use of Artificial Neural Network in Segmenting Clinical Images

Inaccurate detection of tumors, fractures, and breast cancer in clinical images has become one of the major issues in the medical field. Variations or errors in medical reports caused by operators, machines, or the environment become a common cause of delay or incorrect diagnosis. Therefore, correct segmentation of areas of interest in clinical images like X-rays and MRIs is highly required. To solve this problem, many researchers have provided various state-of-the-art automatic or semiautomatic methods of segmentation. Artificial neural networks play a significant role in increasing the accuracy of clinical image segmentation. In this chapter, the workings of ANN and the difference between gradient and stochastic gradient descent are discussed. Also, the application of ANN in tumor, fracture, and breast cancer segmentation is discussed using authentic and publically available datasets. This chapter mentions the results and confusion matrix of some state-of-the-art methods. This chapter will help readers know about ANN, the use of gradient and stochastic gradient descent, the application of ANN in segmenting clinical images, and the confusion matrix.

Keywords: ANN, Clinical images, Deep learning, Segmentation, Tumor segmentation.