Image‐ versus histogram‐based considerations in semantic segmentation of pulmonary hyperpolarized gas images

• Published in: Magnetic resonance in medicine Vol. 86; no. 5; pp. 2822 - 2836
• Main Authors: Tustison, Nicholas J., Altes, Talissa A., Qing, Kun, He, Mu, Miller, G. Wilson, Avants, Brian B., Shim, Yun M., Gee, James C., Mugler, John P., Mata, Jaime F.
• Format: Journal Article
• Language: English
• Published: United States Wiley Subscription Services, Inc 01.11.2021
• Subjects: Advanced Normalization Tools; Algorithms; Artificial neural networks; Computer applications; convolutional neural network; Deep learning; Domains; Ecosystem; Fields (mathematics); functional lung imaging; Histograms; Humans; Image processing; Image Processing, Computer-Assisted; Image segmentation; Lung - diagnostic imaging; Lungs; Machine learning; Magnetic Resonance Imaging; Neural networks; Optimization; Probabilistic models; Retrospective Studies; segmentation; Semantic segmentation; Semantics; Spatial data; Xenon 129; Xenon Isotopes; deep learning; Advanced Normalization Tools; functional lung imaging; segmentation; convolutional neural network
• ISSN: 0740-3194; 1522-2594; 1522-2594
• DOI: 10.1002/mrm.28908

Purpose To characterize the differences between histogram‐based and image‐based algorithms for segmentation of hyperpolarized gas lung images. Methods Four previously published histogram‐based segmentation algorithms (ie, linear binning, hierarchical k‐means, fuzzy spatial c‐means, and a Gaussian mixture model with a Markov random field prior) and an image‐based convolutional neural network were used to segment 2 simulated data sets derived from a public (n = 29 subjects) and a retrospective collection (n = 51 subjects) of hyperpolarized 129Xe gas lung images transformed by common MRI artifacts (noise and nonlinear intensity distortion). The resulting ventilation‐based segmentations were used to assess algorithmic performance and characterize optimization domain differences in terms of measurement bias and precision. Results Although facilitating computational processing and providing discriminating clinically relevant measures of interest, histogram‐based segmentation methods discard important contextual spatial information and are consequently less robust in terms of measurement precision in the presence of common MRI artifacts relative to the image‐based convolutional neural network. Conclusions Direct optimization within the image domain using convolutional neural networks leverages spatial information, which mitigates problematic issues associated with histogram‐based approaches and suggests a preferred future research direction. Further, the entire processing and evaluation framework, including the newly reported deep learning functionality, is available as open source through the well‐known Advanced Normalization Tools ecosystem.