High-level prior-based loss functions for medical image segmentation: A survey

• Published in: Computer vision and image understanding Vol. 210; p. 103248
• Main Authors: El Jurdi, Rosana, Petitjean, Caroline, Honeine, Paul, Cheplygina, Veronika, Abdallah, Fahed
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01.09.2021; Elsevier
• Subjects: Anatomical constraint losses; Artificial Intelligence; Computer Science; Computer Vision and Pattern Recognition; Computers and Society; Convolutional neural networks; Deep learning; Engineering Sciences; Machine Learning; Mathematics; Medical image segmentation; Neural and Evolutionary Computing; Prior-based loss functions; Signal and Image processing; Statistics; Deep learning; Anatomical constraint losses; Convolutional neural networks; Medical image segmentation; Prior-based loss functions
• ISSN: 1077-3142; 1090-235X
• DOI: 10.1016/j.cviu.2021.103248

Today, deep convolutional neural networks (CNNs) have demonstrated state of the art performance for supervised medical image segmentation, across various imaging modalities and tasks. Despite early success, segmentation networks may still generate anatomically aberrant segmentations, with holes or inaccuracies near the object boundaries. To mitigate this effect, recent research works have focused on incorporating spatial information or prior knowledge to enforce anatomically plausible segmentation. If the integration of prior knowledge in image segmentation is not a new topic in classical optimization approaches, it is today an increasing trend in CNN based image segmentation, as shown by the growing literature on the topic. In this survey, we focus on high level prior, embedded at the loss function level. We categorize the articles according to the nature of the prior: the object shape, size, topology, and the inter-regions constraints. We highlight strengths and limitations of current approaches, discuss the challenge related to the design and the integration of prior-based losses, and the optimization strategies, and draw future research directions. •Review of methods that incorporate prior knowledge in deep learning loss function for medical image segmentation•Understanding the mechanisms behind the design and implementation of prior based losses.•Categorization of prior-based losses according to the nature of the prior constraints.•Overview on:•The types of priors existing in the literature and how they are modeled.•The major challenges linked to the design of such prior-based losses.•Their common training and optimization strategies.