Clinical Feasibility of Quantitative Ultrasound Texture Analysis: A Robustness Study Using Fetal Lung Ultrasound Images

Abstract

Objectives To compare the robustness of several methods based on quantitative ultrasound (US) texture analysis to evaluate its feasibility for extracting features from US images to use as a clinical diagnostic tool. Methods We compared, ranked, and validated the robustness of 5 texture‐based methods for extracting textural features from US images acquired under different conditions. For comparison and ranking purposes, we used 13,171 non‐US images from widely known available databases (OUTEX [University of Oulu, Oulu, Finland] and PHOTEX [Texture Lab, Heriot‐Watt University, Edinburgh, Scotland]), which were specifically acquired under different controlled parameters (illumination, resolution, and rotation) from 103 textures. The robustness of those methods with better results from the non‐US images was validated by using 666 fetal lung US images acquired from singleton pregnancies. In this study, 2 similarity measurements (correlation and Chebyshev distances) were used to evaluate the repeatability of the features extracted from the same tissue images. Results Three of the 5 methods (gray‐level co‐occurrence matrix, local binary patterns, and rotation‐invariant local phase quantization) had favorably robust performance when using the non‐US database. In fact, these methods showed similarity values close to 0 for the acquisition variations and delineations. Results from the US database confirmed robustness for all of the evaluated methods (gray‐level co‐occurrence matrix, local binary patterns, and rotation‐invariant local phase quantization) when comparing the same texture obtained from different regions of the image (proximal/distal lungs and US machine brand stratification). Conclusions Our results confirmed that texture analysis can be robust (high similarity for different condition acquisitions) with potential to be included as a clinical tool.

Fetal MRI radiomics: non-invasive and reproducible quantification of human lung maturity

“…Despite the critical importance of radiomics feature reproducibility analyses, there is a paucity of evidence in fetal imaging. One study by Perez-Moreno et al found that gray-level co-occurrence matrix, local binary patterns, and rotation-invariant local phase quantization delivered reproducible texture features from different lung regions in ultrasound images [ 35 ]. However, fetal ultrasound-based lung radiomics analysis has thus far been performed based on two-dimensional image data at the level of the four-chamber view, in lung tissue that is representative of the whole lung according to the examiner’s subjective impression.…”

Artificial Intelligence in Obstetric Ultrasound: An Update and Future Applications

“…Also, the feasibility of these software was validated in different clinical situation including proximal/distal lungs and US machines of different brands, etc. (35). The common thread was that the analysis was not affected by the adjustment of the gray value of the instrument.…”

Deep learning in the ultrasound evaluation of neonatal respiratory status

“…Among the most recently proposed approaches, several studies have focused their attention on fetal lung US image analysis. These methods, mostly based on very well-known textural descriptors, aimed to assess the respiratory status [6], [7] or predict the newborn diseases [8]. Texture analysis has been also employed to determine the respiratory morbidity in newborns [18], while recent fully automatic methods focused on predicting the neonatal maturation degree [19] as well as the respiratory morbidity [20].…”

“…Recently, the activity in this research area is boosted by the availability of powerful image processing tools, in particular those based on deep learning, able at solving complex tasks about image analysis. Recent studies have focused their attention on fetal lung US images and proposed new approaches to assess the respiratory status [6], [7] or to predict the newborn diseases [8]. In this work, we collected a large and challenging dataset and tested different deep convolutional neural networks.…”