Deep Learning for Brain Tumor Segmentation: A Survey of State-of-the-Art

Abstract

Quantitative analysis of the brain tumors provides valuable information for understanding the tumor characteristics and treatment planning better. The accurate segmentation of lesions requires more than one image modalities with varying contrasts. As a result, manual segmentation, which is arguably the most accurate segmentation method, would be impractical for more extensive studies. Deep learning has recently emerged as a solution for quantitative analysis due to its record-shattering performance. However, medical image analysis has its unique challenges. This paper presents a review of state-of-the-art deep learning methods for brain tumor segmentation, clearly highlighting their building blocks and various strategies. We end with a critical discussion of open challenges in medical image analysis.

Kidney Tumor Semantic Segmentation Using Deep Learning: A Survey of State-of-the-Art

“…Neural networks are a sort of learning algorithm that serves as the foundation for the majority of DL techniques [ 21 , 27 ]. As seen in Figure 4 , neural networks are nothing more than a set of arithmetic operations that convert an input (x) into an output (y).…”

“…CNN are capable of automatically extracting useful feature representations with fully connected layers from raw images and optimizing them to represent specific target classes by combining several convolutional, activation, batch normalization, and pooling layers [ 76 ]. In comparison to alternative models, CNN has recently become the de facto model for medical image segmentation due to its record-breaking performance in conventional computer vision tasks as well as medical image analysis [ 21 ]. CNN models may learn spatial hierarchies of features within data, e.g., the first layer will learn tiny local patterns, such as edges, while the second convolutional layer may learn bigger patterns constructed from the first layer characteristics, and so forth.…”

“…They are more suited to image analysis jobs because of this capacity. Furthermore, convolutional layers’ units share weights, decreasing the number of parameters to train and improve the network’s efficiency [ 21 ]. CNN has recently become the de facto paradigm for kidney tumor segmentation because of its superior performance in traditional computer vision and medical image analysis when compared to alternative models.…”

“…Manual segmentation is time-consuming. It also has a lot of intra- and inter-rater variability [ 21 ]. CT technologies can significantly enhance our ability to detect and monitor diseases and patients, and this may improve patient care and facilitate the evaluation of treatment strategies.…”

Advanced Computational Methods for Oncological Image Analysis

“…The effort of the research community to propose automatic brain tumor segmentation and classification methods has been tremendous. As a result, ample literature exists on segmentation using region growing, traditional machine learning and deep learning methods [22,23]. Similarly, a number of tasks have been successfully conducted in the area of brain tumor classification into their respective histological type.…”