Ultrasound Elastography for Lung Disease Assessment

Abstract

Ultrasound elastography (US-E) is a noninvasive, safe, cost-effective and reliable technique to assess the mechanical properties of soft tissue and provide imaging biomarkers for pathological processes. Many lung diseases such as acute respiratory distress syndrome, chronic obstructive pulmonary disease, and interstitial lung disease are associated with dramatic changes in mechanical properties of lung tissues. Nevertheless, US-E is rarely used to image the lung because it is filled with air. The large difference in acoustic impedance between air and lung tissue results in the reflection of the ultrasound wave at the lung surface and, consequently, the loss of most ultrasound energy. In recent years, there has been an increasing interest in US-E applications in evaluating lung diseases. This article provides a comprehensive review of the technological advances of US-E research on lung disease diagnosis. We introduce the basic principles and major techniques of US-E and provide information on various applications in lung disease assessment. Finally, the potential applications of US-E to the diagnosis of COVID-19 pneumonia is discussed.

I. Introduction

Lung disease is one of the leading causes of death worldwide. Interstitial lung diseases (ILDs) include more than 200 diseases, such as pulmonary fibrosis, pulmonary edema, pneumonia, and so on. Idiopathic pulmonary fibrosis (IPF) is increasingly recognized as a public health concern, with more than 50 000 cases annually diagnosed [1], and lung cancer is by far the leading cause of cancer death among both men and women, making up almost 25% of all cancer deaths. Meanwhile, in 2020, the world is experiencing the Covid-19 pandemic, which has resulted in many cases of pneumonia. More than ever, accurate diagnosis and monitoring of lung diseases are concerns of global scale.

Chest X-ray and computed tomography (CT) are the conventional imaging techniques that play a predominant role in the diagnosis and management of lung diseases. Chest X-ray can detect cancer, infection, or air collecting in the space around a lung, which can cause the lung to collapse. It can also show chronic lung conditions, such as emphysema or cystic fibrosis. Although chest X-ray has widely been used for diagnosing pneumonia, it has limited sensitivity and specificity as well as high interobserver variability. CT is used to check the size and structure of an organ and determine if it is infected, solid or filled with fluid. This technique is used to diagnose tumors, cancer, and various other abnormalities within the body [2]. Although a chest CT-scan is more sensitive than a radiograph in detecting the presence of lung infiltrates and regarded as the “gold standard” for diagnosing pneumonia, chest CT is not used for routine examination for patients with suspected pneumonia for reasons including unavailability, concerns about radiation, and cost.

Lung ultrasound (LUS) is an emerging imaging technology that is used in the diagnosis and monitoring of various lung diseases, such as pneumonia [3], [4], fibrosis [5], edema [6], and pneumothorax [7]. Given the COVID-19 pandemic, LUS could be beneficial for early screening of this disease. For example, Sperandeo et al. [8] reported the use of thoracic ultrasound in the diagnosis and follow-up of community-acquired pneumonia, Their study suggested that thoracic ultrasound was a reproducible and useful complementary imaging tool in this context [8]. Given the promptness of the diagnosis, thoracic ultrasound is especially useful for identifying pneumonia in patients in a semi-recumbent or supine position in an emergency department or intensive care unit [9], [10].

Ultrasound elastography (US-E) provides a quantitative assessment of tissue elasticity for diagnostic applications. In a qualitative, semi-quantitative, or quantitative manner, US-E has been used for disease diagnoses in various organs, such as liver, breast, thyroid, prostate, kidney, and lymph nodes. However, given the air-filled alveoli of the lung and total reflection of the ultrasound beam, this technique is not commonly used for lung disease diagnosis. Because of advances in this technique in recent years, there has been an increasing interest in the use of US-E for evaluating lung disease.

This article provides a systematic review of US-E research in lung diseases. We conducted a comprehensive literature search using the search terms “ultrasound elastography” and “lung disease” in PubMed, Web of Science, and Google Scholar from 2000 to 2020. Search results were screened to exclude articles that were irrelevant. A total of 42 articles were included and all studies had IRB or IACUC approvals. Table I shows a list of references that use US-E in lung diseases. We organized the studies into three disease categories: interstitial lung disease, lung lesions, and pleural effusion and pulmonary edema. In Section II, we introduce the basic principles and various types of US-E. In Section III, we present the various US-E applications in lung disease evaluation. In Section IV, we discuss the potential US-E applications in COVID-19 pneumonia diagnostics. Section V summarizes our conclusions.

TABLE I. Summary of Studies Using Ultrasound Elastography for Lung Disease Evaluation.

References	Methods	Disease	Frequency	Outcome
[21] – [36]	SWE	Interstitial lung disease (ILD)	6 MHz, linear	Significant difference in the wave speed between control and ILD groups
[39] – [45], [48]	Transthoracic ARFI	Lung lesions	7.5 MHz, convex	Differentiation of different lesion types
[46]	Transthoracic SWE	Lung lesions	4–9 MHz linear 1–4 MHz convex	Differentiate malignant from benign subpleural solid lesions
[47]	Transthoracic SE, ARFI, p-SWE	Peripheral lung lesions (PLLs)	1.5–4 MHz, convex	ARFI and p-SWE can differentiate malignant from benign PLLs.
[51], [52]	SWE	Malignant pleural effusion	4–15 MHz, linear	Differentiate malignant from benign pleural effusion
[53], [54]	SWE	Edema	6 MHz, linear	Magnitude of wave speed correlate with markers of extravascular lung water

II. Ultrasound Elastography

A. Principles

Elastography assesses tissue elasticity, which can be described by Hooke’s Law

where stress ( ) is the force per unit area, strain is the deformation due to stress, and the elastic modulus ( ) relates stress to strain [11]. Given the different methods of static deformation, three kinds of elastic moduli ( ) are defined: Young’s modulus ( ), Shear modulus ( ), and Bulk modulus ( ).

Young’s modulus ( ) is defined as the relationship between normal stress ( ) and normal strain ( )

Shear modulus ( ) is defined as the ratio of shear stress ( ) to shear strain ( )

Bulk modulus ( ) is defined as the ratio of bulk stress ( ) to bulk strain ( )

In addition to governing the static deformation, elastic moduli also characterize wave propagation in the material

where is the material density and is the wave speed.

There are two kinds of wave propagation in ultrasound: longitudinal and shear. Longitudinal waves are waves in which the displacement of the medium is parallel to the wave propagation. For shear waves, the displacement of the medium is perpendicular to the direction of wave propagation.

The longitudinal wave speed can be calculated as

The shear wave speed can be calculated as

The longitudinal wave speed is approximately 1540 m/s, while shear wave speed is roughly 1–10 m/s in soft tissues.

B. Techniques

Strain elastography (SE) assesses tissue strain using mostly manual compression [12]. This technique works well for superficial organs such as breast [13] and thyroid [14], yet is challenging for deep organs such as liver [15]. Correlation of radiofrequency echo signals between the states before and after compression is used to measure the tissue displacement. The strain ratio is defined as the ratio of strain measured in the adjacent (normal) tissue to the strain measured in the target region of interest (ROI). A strain ratio >1 indicates lower strain and greater tissue stiffness. Young’s modulus ( ), shear modulus ( ), and bulk modulus ( ) are usually used in the SE. Although this technique is widely used in the clinic, numerous studies showed that manual compression is difficult to maintain and operator-dependent. Therefore, SE is often associated with low reproducibility and high inter-observer variability.

Acoustic radiation force impulse (ARFI) elastography is an alternative technique to measure tissue strain. ARFI displaces tissue perpendicular to the tissue surface via a short-duration, high-intensity acoustic pulse and then measures the displacement within a specified ROI [16].

Shear wave elastography (SWE) utilizes a dynamic stress to generate a shear wave in parallel or perpendicular, measures the shear wave speed of the tissue, and quantitatively assesses the tissue stiffness in terms of shear modulus ( ) or Young’s modulus ( ). This technique consists of 1-D transient elastography, point SWE (p-SWE), and 2-D SWE (2-D-SWE). The 1-D transient elastography is mostly used for characterizing liver fibrosis. It uses a single device integrating a mechanical vibrator to generate a shear wave and an ultrasound probe to measure the shear wave speed [17]. For p-SWE, ARFI is used to induce tissue motion in the normal direction within a focal point, and some of the longitudinal waves produced by ARFI are converted to shear waves [16]. In contrast to ARFI and p-SWE, 2-D-SWE uses acoustic radiation force to generate shear waves at multiple focal zones in rapid succession, creating a nearly cylindrical shear wave cone and allowing real-time visualization of 2-D shear waves. The shear wave speed can be measured and a quantitative elastogram superimposed on a B-mode image can be generated, providing both anatomical and elastographical information [18], [19].

Safety issues should be considered when using ARFI and SWE for lung imaging. Miller et al. [20] conducted an experimental study on rats to investigate the relative incidence of pulmonary capillary hemorrhage when using different imaging modes of diagnostic ultrasound. Their results showed that only ARFI pushes led to hemorrhage and suggested reducing the examination duration might not be very effective in reducing the risk of hemorrhage, while reducing the power (or on screen mechanical index) may be a better strategy for potentially vulnerable patients.

III. Ultrasound Elastography in Lung Disease Diagnoses

A. Ultrasound Surface Wave Elastography for Interstitial Lung Disease

ILD includes a number of lung disorders in which the lung tissue is stiffer due to the fibrosis of the lung tissue [21]. Sedie et al. [22] reviewed the role of ultrasound in the diagnosis and clinical evaluation of lung involvement in systematic sclerosis. They reported an update of the available data regarding the use of ultrasound in lung conditions that might cause disability and mortality in patients with systemic sclerosis.

Zhang et al. [21], [23] developed a LUS surface wave elastography (LUSWE) technique for assessing superficial lung stiffness and tested it in an ex vivo porcine muscle–lung model. In this model, the pig muscle was laid on top of a pig lung. A vibration excitation at 100 Hz was generated for 0.1 s on the muscle surface and the wave propagation of the lung surface was detected using a linear array ultrasound probe with a central frequency of 5 MHz. Zhou et al. [24] further validated this technique in pulmonary fibrosis at different levels (control, mild, and severe) in the ex vivo bleomycin mouse model. A gelatin mixture was used for supporting the lung sample and reducing wave reflection from the boundary of the testing container (Fig. 1). A mechanical shaker was used to generate a vibration on the gelatin and a high-frequency linear array ultrasound probe with central frequency of 18 MHz was used for capturing the wave propagation of the lung. The resultant lung motion was estimated via tracking displacement with an autocorrelation technique. Eight locations over a length of approximately 8 mm on the mouse lung surface were selected and used to measure the wave speed based on the phase-delay of selected points using a phase gradient method

where is the wave speed, is the excitation frequency in Hz, and is the slope of the linear regression between phase delay and lateral distance of selected points (Fig. 2). The results showed statistically significant differences in the magnitudes of lung surface wave speed, pulse oximetry, and compliance between control mice and mice with severe pulmonary fibrosis.

Fig. 1.

(a) Three groups (control, mild, and severe) of mouse lungs embedded in gelatin in plastic containers for measuring the lung surface wave speed. (b) Schematic of the experimental setup (adapted from [24]).

Fig. 2.

(a) Eight locations over 8 mm on the mouse lung surface were used to measure the wave speed using US tracking beams. (b) Phase delay of the remaining locations, relative to the first location, is used to measure the surface wave speed. (adapted from [24]).

Zhang et al. [25], [28] Kalra et al. [26], and Osborn et al. [27] then conducted a pilot clinical study of using LUSWE in ten patients with ILD and ten healthy controls. LUSWE was performed on regions of both lungs through six intercostal spaces in the upper anterior, lower lateral, and lower posterior lungs (Fig. 3). The lung was tested at total lung capacity (TLC) when the subject was asked to take a deep breath and hold it for a few seconds. A mechanical shaker was placed adjacent to the ultrasound transducer with a distance of roughly 5 mm and used to generate a wave, while the ultrasound transducer with a central frequency of 6.4 MHz was used for capturing the wave propagation of the tissue. The normal component of the lung surface motion was quantified by cross-correlation analysis of the ultrasound tracking beams [29]. Eight locations covering a length of approximately 6 mm on the lung surface were used to measure the normal component of the lung surface motion. The results demonstrated that the lung surface wave speed in patients with ILD (3.3 ± 0.37 m/s at 100 Hz, 4.38 ± 0.33 m/s at 150 Hz, and 5.24 ± 0.44 m/s at 200 Hz) was statistically significantly higher than that of healthy control (1.88 ± 0.11 m/s at 100 Hz, 2.74 ± 0.26 m/s at 150 Hz, and 3.62 ± 0.13 m/s at 200 Hz), indicating lung fibrosis. This is in agreement with the results from Marinelli et al. [30] who used magnetic resonance elastography (MRE) to quantitatively assess the shear stiffness of lung parenchyma and diagnose ILD. Fifteen patients diagnosed with ILD and 11 healthy controls were recruited in this study. The data were collected at residual volume (RV) and TLC. The results from this study showed that the magnitudes of the parenchymal shear stiffness of patients with ILD at RV (1.32 ± 0.3 kPa) and TLC (2.74 ± 0.896 kPa) were statistically significantly higher than those of healthy controls at RV (0.849 ± 0.25 kPa) and TLC (1.33 ± 0.195 kPa). Zhang et al. [21] used the wave speed of lung surface to characterize the tissue stiffness. Given the density of lung surface was unknown, they did not calculate the stiffness. Assuming the lung mass density is 0.3 kg/m³, the shear stiffness of lung surface can be calculated as , is the wave speed, is the shear stiffness, and is the tissue density. After calculating the shear stiffness from wave speed, it showed that the stiffness measured by MRE and US-E were of the same order of magnitude. Moreover, the differences between ILD patients and controls were of the same order.

Fig. 3.

Experimental setup and result of LUSWE. (a) Position of the handheld shaker. (b) Magnified view of the US probe in the intercostal space. (c) US B-mode image; the blue dots are the selected points for wave speed measurements. (d) Wave phase delay relative to the first location (leftmost blue dot) measures the lung surface wave speed (adapted from [31]).

Based on the results from both US-E and MRE, it showed that the differences in the lung stiffness between ILD patients and healthy controls were of the same order.

Clay et al. [32] explored the correlation between LUSWE and a recently developed CT imaging technique called Computer-Aided Lung Informatics for Pathology Evaluation and Rating (CALIPER). This study was conducted on 77 patients with ILD and 19 healthy controls. The results showed that the lung surface wave speed positively correlated with grading from a radiologist and CALIPER-detected interstitial abnormalities (Fig. 4).

Fig. 4.

Examples of radiologic appearance, quantitative analysis, and clinical severity, along with LUSWE velocity. Case A is a healthy control with a normal chest CT and normal lungs quantitatively color-coded as dark and light green by CALIPER. LUSWE velocity is 3.94 m/s, expectedly slower than the ILD cases. Case B is a clinically and radiologically identified mild ILD case with ILAs involving 35% of the total lung volume (shown in foci of yellow and orange) and significantly higher LUSWE velocity. Case C is a case with both high clinical and visually assessed severity with diffuse disease involving 83% of the lung and high LUSWE velocity (adapted from [32]).

Zhou et al. [31] later used LUSWE for ILD staging in a prospective study. Of 91 patients with ILD and 30 healthy controls, severity of ILD was graded as healthy (F0), mild (F1), moderate (F2), or severe (F3) based on pulmonary function testing (PFT), high-resolution CT (HRCT), and clinical assessments. For PFT, the lowest values for predicted TLC, diffusion capacity, and forced vital capacity were used to grade ILD. Healthy controls had no clinical symptoms of lung disease and were grade F0. For HRCT, the extent of involvement was ranked. The obtained results showed that LUSWE had a sensitivity of 92% and a specificity of 89% for differentiating between and – , a sensitivity of 50% and a specificity of 81% for differentiating – from , and a sensitivity of 88% and a specificity of 97% for differentiating between and . The highest area under the receiver operating characteristics (ROC) curve was observed between 0.83 and 0.94 to distinguish between and – , suggesting that LUSWE could be an adjunct to HRCT for noninvasive ILD evaluation. Verschakelen commented on this study and stated that LUSWE could be a promising technique to detect early ILD in systemic sclerosis [33]. LUSWE could be an easy-to-use alternative for HRCT and does not expose the patient to ionizing radiation. A noninvasive, easy-to-repeat imaging technique such as LUSWE would be very welcome to follow patients diagnosed with IPF and treated with antifibrotic therapy. Zhang et al. [34] then did a follow-up study using LUSWE for assessing the progression of 52 patients with ILD (interval was 9.2 ± 3.5 months). The disease progression between baseline and follow-up was clinically assessed based on changes in PFT, HRCT, and echocardiography. They showed that the correlations of changes in surface wave speed obtained from LUSWE in the lower lateral and posterior portions of the lung with clinical evaluations were good ( ).

Studies have shown that the lung mass density increases with the degree of fibrosis. Zhou and Zhang [35] developed a deep neural network (DNN) technique for predicting lung mass density based on the lung surface wave speed obtained from LUSWE and lung elasticity from PFT measurements. The proposed DNN is composed of five fully connected layers, four neurons in the input layer, 1024 neurons in each of three hidden layers, and one neuron in the output layer. They validated this technique on the synthetic data (validation accuracy of 0.992) and then on a sponge phantom. The synthetic data is composed of surface wave speeds at 100, 150, and 200 Hz, lung mass density (kg/m³) as well as shear elasticity and viscosity. Based on Voigt’s model, the surface wave speeds at three frequencies can be calculated by specifying lung mass density, shear elasticity, and viscosity. They further validated this technique in patients with ILD and healthy subjects (77 in total) [36]. The predefined lung mass density was calculated based on the Hounsfield unit obtained from HRCT. The performances of the DNN with two types of activation functions [rectified linear activation unit (ReLU) and exponential linear unit (ELU)] and various machine learning models (support vector regression, random forest, and Adaboost) were compared, showing that the predictions via the DNN with ELU obtained a better performance in the testing dataset (accuracy = 0.89) than those of the DNN with ReLU and machine learning models, suggesting that this technique may be able to noninvasively quantify the lung mass density.

B. Ultrasound Elastography for Assessing Lung Lesions

Lung cancer is one of the most common malignant tumors worldwide. Cancer cells exhibit quantitatively different biophysical properties, such as cell stiffness and elasticity, compared with normal cells [37]. Tumor progression alters the composition and physical properties of the extracellular matrix. In particular, increased matrix stiffness has profound effects on tumor growth and metastasis [38]. Thus, studies have indicated that malignant tumors are usually stiffer than the benign lesions. Discrimination between benign and malignant tumors plays a vital role in the diagnosis and treatment of patients with lung cancers. US-E could be used for assessing the tumor stiffness at the tissue level.

Uramoto et al. [39] hypothesized that the ARFI may be used to visualize subpleural tumors in ex vivo pig lungs yet intraoperative US-E may not be able to detect deep tumors. Thoracic ultrasound has been used for diagnosis and intervention in pulmonary disease [40]. Sperandeo et al. [41], [42] used transthoracic ARFI for identifying subpleural cancer. Ninety-five consecutive patients with lesions detected on chest X-ray or CT were included. They showed that a significantly lower elasticity of squamous cell carcinoma (4.67 ± 0.492) was observed versus other types of lung cancer ( ), and versus pneumonia (2.35 ± 0.48). Lim et al. [43] used transthoracic ARFI to differentiate pulmonary lesions, showing that the strain ratios were significantly different among lesions (1.03 ± 0.71 [necrosis, ] versus 2.51 ± 1.14 [atelectasis, ] versus 19.98 ± 15.59 [consolidation, ] versus 36.19 ± 20.18 [tumor, ]; ) [43]. Moreover, the strain ratio of primary lung cancer was significantly different from pneumonia ( ) and metastatic lung cancer ( ). In contrast to the invisibility of pulmonary lesions in B-mode ultrasound, they hypothesized that the interaction between acoustic impulse and the interstitium would enable ARFI to visualize the pulmonary lesions, providing valuable diagnostic information on peripheral pulmonary lesions and help for interventional procedures.

Adamietz et al. [44] used both B-mode ultrasound and ARFI to visualize and identify the pulmonary lesions. Seventeen lesions (eight female patients suffering from breast cancer, ovarian cancer, and uterine leiomyosarcoma) were solid in CT and appeared as inelastic in ARFI. The time interval between CT and ultrasound examinations was less than two weeks for all patients. All lesions were located in the peripheral rim of the lower lobes and were not smaller than 5 mm. The transducer was placed perpendicular to the skin surface between the ribs with no pressure. Interestingly, the results showed that ARFI was able to visualize and detect all pulmonary lesions in contrast to B-mode ultrasound (Fig. 5). However, Mostbeck questioned whether the solid nodules surrounded by the aerated lung, which were not visible by real-time ultrasound, could be visualized by ARFI [45]. There is a need to further validate this technique on the detection of the pulmonary lesions and understand the physical mechanism of this phenomenon.

Fig. 5.

B-mode and real-time US-E images of a lung lesion. The pulmonary lesion was invisible on the B-mode imaging. The lesion appeared inelastic (red on color mapping) on the US-E image. A fibrotic string connecting the lesion with the pleura was reproducible by real-time US-E imaging (adapted from [44]).

Ozgokce et al. [46] used transthoracic ultrasound SWE to differentiate malignant from benign subpleural solid lesions. A total of 33 cases with subpleural solid lesions were included. ARFI elastography was performed using linear transducer (4–9 MHz) and convex transducer (1–4 MHz) with optimal intercostal distance and breath holding of the patient when necessary. Shear wave velocity (SWV) quantification was performed in the axial plane originating from the center of the lesion. The mean value of SWV of the benign lesions (2.18 ± 0.49 m/s, ) was statistically significantly lower than that of the malignant lesions (3.50 ± 0.69 m/s, ). Sensitivity and specificity were 97.7% when the SWV cutoff value was set to 2.47 m/s.

Wei et al. [47] used US-E to differentiate between benign and malignant peripheral lung lesions (PPLs). The PLL makes the acoustic window of the lung available, enabling the passage of ultrasound waves and therefore the use of transthoracic US-E. SE, ARFI, and p-SWE were performed on 91 patients with PLLs (36 benign and 55 malignant confirmed by US-guided biopsies or surgeries). SE was performed with mild manual compression. The ROI was the whole lesion and some peripheral normal tissues. ARFI was then performed on the same ROI. The p-SWE mode was then initiated with a mm ROI box placed over the solid portion without air covering the lesion. The results showed that SE was not able to differentiate malignant from benign PPLs ( ). An elasticity score of 3 or greater from ARFI was predictive of malignancy with a sensitivity of 83.6% and a specificity of 52.8%. With regard to p-SWE, a sensitivity of 70.9% and a specificity of 69.4% were obtained when using 1.951 m/s as the cutoff value to differentiate malignant from benign PPLs.

The diagnostic accuracy of ultrasound is greatly improved by using ultrasound bronchoscopy, in which an ultrasound transducer is coupled with the bronchoscope so as to place the probe through the airway to reach the lesion. He et al. [48] performed endobronchial ultrasound (EBUS) elastography on 57 patients with central lung lesions and the tissue stiffness were quantified. A convex probe was inserted through an oral connecting tube. Scanning was performed at the central frequency of 7.5 MHz. The scan range included the entire lung lesion and surrounding normal tissue. Elastographic images were acquired based on the compression generated by the pulsation of vessels in the thoracic cavity and respiratory movement. The elasticity of tissue within the scanned region was reconstructed by comparing it with the surrounding tissue. The results showed that the elastography grading score was more sensitive and specific than the standard EBUS criteria in differentiating malignant from benign lung lesions with a sensitivity of 72.2%, a specificity of 76.2%, and accuracy of 73.3%.

C. Ultrasound Elastography for Assessing Pleural Effusion and Pulmonary Edema

Pleural effusion manifests as compression of pleural fluid on the lung parenchyma contributing to hypoxemia. LUSWE can be used to assess the mechanical properties of lung parenchyma for patients with pleural effusion. In order to investigate the effect of pleural effusion on the LUSWE measurement, Zhou and Zhang [49], [50] developed a sponge phantom model for simulating pleural effusion. Different thicknesses of ultrasound transmission gel were used to simulate pleural fluid by insertion into a condom and placement between the ultrasound transducer and a sponge phantom. Mechanical waves were generated by the shaker at different frequencies ranging from 100 to 300 Hz, and the ultrasound transducer captured the surface wave speed of the sponge. The results showed that the thickness of the ultrasound transmission gel had a negligible effect on the surface wave speed of the sponge at 100 and 150 Hz.

A malignant pleural effusion (MPE) is the buildup of fluid and cancer cells that collects between the chest wall and the lung. It is a fairly common complication in a number of different cancers. This complication can be relieved via a variety of chemical or mechanical means, after complete drainage of the effusion that allows parietal-to-visceral pleural apposition. Instillation of the sclerosing agent is thereafter followed by a profound inflammatory response between the layers, which, in turn, result in fibrin accumulation and pleural fibrosis. Jiang et al. [51] and Hou et al. [52] used US-E for diagnosing 244 patients with MPE. These patients were divided into a development set (114 in total, 53 malignant and 61 benign) and a validation set (130 in total, 55 malignant and 75 benign). An ultrasound system with a bandwidth between 4 and 15 MHz and a 55-mm-long linear array transducer was used in this study. Patients in a seated position were asked to raise their arms above their head so as to widen the intercostal space as an acoustic window. The intercostal spaces were sequentially scanned from top to bottom. The results showed that the pleural US-E had a sensitivity of 83.64% and a specificity of 90.67% for diagnosing MPE in the validation set, indicating that US-E was better in differentiating MPE from benign pleural effusion than transthoracic US.

Zhang et al. [53], [54] performed LUSWE on an ex vivo porcine lung model to assess the effects of pulmonary edema on the mechanical properties of lung tissue. Various amounts (0, 120, and 240 mL) of water were poured into the lung through the trachea with ultrasound imaging used to guide the water filling until the water overflowed the lung surface (Fig. 6). The mechanical wave was generated from 100 to 400 Hz, and the surface wave speed at each frequency was measured. It was shown that lung surface wave speed increased with frequency and decreased with injected water volume.

Fig. 6.

Experimental design for testing the surface wave speed. An ex vivo fresh swing lung was tested and a thick rubber pad was placed under the lung to reduce wave reflection. Water was injected into the lung through the trachea. The surface wave propagation was generated using a small vibrator and measured using an ultrasound probe (adapted from [53]).

Given that presence of extravascular lung water induces the reduction of lung compliance and increases lung stiffness, Wiley et al. [55] evaluated the feasibility of LUSWE for quantitative assessment of symptomatic pulmonary edema in 14 patients hospitalized with acute congestive heart failure. Diuretic therapy was administered to remove extravascular lung water from patients with pulmonary edema, and the net fluid balance was recorded at the baseline and follow-up examinations. Each patient was imaged using 2-D LUS in the supine position with B-lines counting. Meanwhile, LUSWE was performed in the second, third, and fourth intercostal spaces for each patient, and the corresponding lung surface wave speed was measured. The results showed that, from baseline to follow-up examinations, a significant reduction in the magnitude of lung surface wave speed correlated with the markers of decreased extravascular lung water in terms of a decline in the number of B-lines and fluid removal.

IV. Potential Applications of Ultrasound Elastography in COVID-19 Pneumonia Diagnoses

COVID-19, the illness caused by the new coronavirus, has sickened over ten million people and taken over 500 000 lives worldwide as of June 28, 2020 [56]. While the majority of cases experience mild illness and are able to recover, the severe cases can develop fetal complications such as organ failure, septic shock, and severe pneumonia [57]. One of the most important tasks in the management of COVID-19 pneumonia is early screening of high risk and critically ill patients [2], [58], [59]. CT examination plays a vital role in early detection of COVID-19 pneumonia and in managing the current COVID-19 outbreak, especially in the highly suspicious, asymptomatic cases with negative nucleic acid testing [60]. Yet, in the context of screening and frequent monitoring in the follow-up examinations, chest CT may increase patient’s burden given its involvement of high radiation dose. Moreover, the rooms and systems for CT examination need to be rigorously cleaned to prevent contamination for patients with a high COVID-19 suspicion.

Bedside ultrasound is a portable, cost-effective and safe imaging modality for pulmonary disease diagnostics [61]. This technique could be useful for frequent and repeated examinations for patients with COVID-19 pneumonia. Under ultrasound imaging, pleural line thickening with uneven B-lines can be observed in patients with COVID-19 pneumonia [62], [63]. LUS findings have been shown to correlate well with CT [59], [62], [64]. Huang et al. [65] found that COVID-19 foci were mostly observed in the posterior lower portions of both lungs. An acquisition protocol for LUS in COVID-19 patients has been suggested [66], [67]. The recommendations include using convex or linear probes and setting the focal point on the pleural line. The mechanical index should be kept as low as possible; avoid the use of cosmetic filters and special imaging modalities, such as harmonic, Doppler, contrast, and compounding, should be avoided; and it is ideal to achieve the highest frame rate possible. Although 2-D LUS is frequently used in the diagnosis of COVID-19, its clinical role is still unclear, partially due to its qualitative nature. Clinical interpretation of 2-D lung images is subjective, and at times, inaccurate, and irreproducible. Roy et al. [68] developed a deep learning framework for classification and localization of COVID-19 markers in LUS. This framework derived from a spatial transformer network was able to predict the disease severity score associated with the input frame and provide localization of pathological artifacts in a weakly supervised manner. Moreover, they further predicted the disease severity score at a video-level based on uninorms for effective frame score aggregation. The quantities such as wave speed or local stiffness of superficial lung tissue obtained from US-E could be used as features for this framework to further improve the diagnostic accuracy of COVID-19 pneumonia.

V. Conclusion

Lung disease is a major health concern worldwide, making accurate assessment and safe monitoring of lung pathologies an area of increasing importance. This is especially true in the time of COVID-19 pandemic. The main advantages of LUS are portability, cost-effectiveness, and safety that will enable bedside evaluation and repeated examinations during follow-up. Performing LUS at the bedside could minimize the need for transferring the patient, reduce exposure to radiation from CT, and reduce the risk of spreading the COVID-19 infection among the healthcare personnel. The combination of LUS with US-E could provide a more robust, quantitative method for evaluating superficial lung stiffness and potentially generate an automated “COVID-19 lung score.” Such quantitative assessment could allow precise longitudinal tracking (hour-to-hour or day-to-day) of an individual patient’s lung condition to evaluate the efficacy of the management of the disease.

Funding Statement

This work was supported by the Emory Winship Cancer Institute.