UNet Deep Learning Architecture for Segmentation of Vascular and Non-Vascular Images: A Microscopic Look at UNet Components Buffered With Pruning, Explainable Artificial Intelligence, and Bias

Abstract

Background & Motivation: Biomedical image segmentation (BIS) task is challenging due to the variations in organ types, position, shape, size, scale, orientation, and image contrast. Conventional methods lack accurate and automated designs. Artificial intelligence (AI)-based UNet has recently dominated BIS. This is the first review of its kind that microscopically addressed UNet types by complexity, stratification of UNet by its components, addressing UNet in vascular vs. non-vascular framework, the key to segmentation challenge vs. UNet-based architecture, and finally interfacing the three facets of AI, the pruning, the explainable AI (XAI), and the AI-bias. Methods: PRISMA was used to select 267 UNet-based studies. Five classes were identified and labeled as conventional UNet, superior UNet, attention-channel UNet, hybrid UNet, and ensemble UNet. We discovered 81 variations of UNet by considering six kinds of components, namely encoder, decoder, skip connection, bridge network, loss function, and their combination. Vascular vs. non-vascular UNet architecture was compared. AP(ai)Bias 2.0-UNet was identified in these UNet classes based on (i) attributes of UNet architecture and its performance, (ii) explainable AI (XAI), and, (iii) pruning (compression). Five bias methods such as (i) ranking, (ii) radial, (iii) regional area, (iv) PROBAST, and (v) ROBINS-I were applied and compared using a Venn diagram. Findings and Conclusions: Vascular and non-vascular UNet systems dominated with sUNet classes with attention. The majority of the studies suffered from low interest in XAI and pruning strategies. None of the UNet models qualified to be bias-free. There is a need to move from paper-to-practice paradigms for clinical evaluation and settings. INDEX TERMS Image segmentation; vascular; non-vascular; UNet classes; UNet variations; UNetcomponents; explainable AI; pruning; bias.

Deep Learning Paradigm and Its Bias for Coronary Artery Wall Segmentation in Intravascular Ultrasound Scans: A Closer Look

“…Still, the conventional model had dominated for a long time due to UNet's strong abilities, such as automatic feature extraction, the ability to add a transformer, and its attention-enabled solutions [39,117].…”

“…The most powerful paradigm was the addition of addition of intermediate layers between the input and output layers [145]. We could not only add a layer between these input and output layers, but we could add ones large in number and shape to these networks for superior feature extraction followed by classification or risk stratification [39]. These deep layers are a special case of machine learning, where the features extracted were limited and ad hoc and not like those of the deep layers, where the features extracted were stronger compared to machine learning model features [146].…”

“…The ability of UNet-based deep learning models as shown before is very powerful in the imaging domain and can handle image noise, structure, scale, size, resolution, and further, the variability in the shapes [39]. Thus, we applied that as an assumption to hypothesize that UNet-based solutions are more powerful than conventional models for wall segmentation in IVUS scans.…”

Extending Convolution to Semantic Segmentation with Graph Neural Networks

“…The segment classification network is based on the UNet architecture [104]. An overview of the developed segment classification network is shown in fig.…”

Automatic 3D Segmentation and Identification of Anomalous Aortic Origin of the Coronary Arteries Combining Multi-view 2D Convolutional Neural Networks

“…Different variations of CNNs have been explored, including 3D CNNs, multi-channel U-Nets, and graph CNNs, with reported Dice score coefficients (DSCs) ranging from 0.6 to 0.9 [ 18 ]. Recent review works have highlighted the growing popularity of CNN-based methods for coronary artery segmentation and emphasized the need for further development to translate research into clinical practice [ 17 , 19 ].…”