Radiation dose reduction considerations and imaging patterns of ground glass opacities in coronavirus: risk of over exposure in computed tomography.

This article aims to summarize the available data on the severe acute respiratory syndrome coronavirus 2 (SAR-CoV-2) imaging patterns as well as reducing radiation dose exposure in chest computed tomography (CT) protocols. First, the general aspects of radiation dose in CT and radiation risk are discussed, followed by the effect of changing parameters on image quality. This article attempts to highlight some of the common chest CT signs that radiologists and emergency physicians are likely to encounter. With the increasing trend of using chest CT scans as an imaging tool to diagnose and monitor SAR-CoV-2, we emphasize that pattern recognition is the key, and this pictorial essay should serve as a guide to help establish correct diagnosis coupled with correct scanner parameters to reduce radiation dose without affecting imaging quality in this tragic pandemic the world is facing.

Introduction

The rapid rise of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2; previously known as 2019 novel coronavirus or 2019-nCoV) disease (COVID-19) has become a global emergency that has sent shockwaves in medicine to each corner of the globe [1]. The coronavirus infection took place in Wuhan, Hubei Province, China. In humans, coronaviruses are a series of viruses that cause the symptoms of upper respiratory infections which can become severe. The current status of mortality rates between China and Italy are similar with fatalities in mostly the elderly with known comorbidities [2]. Severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERS) have mortality rates of 10% and 37%, in each of their respective continents [1, 3, 4]. This global emergency has resulted in increased utilization of chest imaging in the confirmation and progress of disease throughout their treatment.

Over the last century, chest radiography was the first line of imaging in respiratory related pandemics that occurred such as Malaria, Influenza, Tuberculosis, SARS and HIV/AIDS [5]. However, the explosion of computed tomography (CT) technology and the effect of lower doses in chest CT around the world have been heavily employed to monitor SARS-CoV-2 [6-10]. Standard chest CT doses can range from 3 to 4.8 mSv and low/ultra-low dose chest CT 0.13–1.5 mSv [11, 12]. A recent study demonstrated that no effect of low-dose CT on human DNA was detected with chest CT scans below 1.5 mSv, and increasing beyond this dose increases double-strand breaks in DNA and chromosome aberrations [13]. As such, depending on the severity of disease in SARS-Covid-2, the mean time between the first CT and the subsequent CT range from mild, common, severe and critical is from 5.8  ±  1.2, 4.9  ±  1.4, 4.5  ±  1.0 and 3.7  ±  1.1 days, respectively [14]. Therefore, in the current global pandemic there have been patients undergoing routine chest CT examinations up to 14 CT scans in the monitoring of disease [15], which can potentially place the patient at risk with a combined radiation dose delivered within 20 days from 21 to 67 mSv which the latter has up to 3.5 times the allowable dose as per the recommendations set by the International Commission on Radiological Protection (ICRP) Report 103 [16].

As of April 27, 2020, a total of 3,017,806 cases and 209,661 deaths had been reported in more than 100 countries [17]. With this ever-expanding number of SARS-CoV-2 cases, the radiological society bodies [18, 19] were first to release statements that the first line of imaging should be a chest radiograph and not a chest CT. Prior to conducting this retrospective review, the institutional review board approval was waived. The aim of this review is to highlight the scanning techniques that reduce radiation dose without affecting imaging quality in low-dose lung CT in the diagnosis of ground glass opacities (GGO).

Anatomical considerations

The anatomy of the secondary pulmonary lobule is key to understanding of GGO. The identification of the patterns of infiltration and distribution is a key to the establishment of a correct list of differential diagnoses, and sometimes to the diagnosis itself [20]. CT provides submillimeter spatial and contrast resolution into lung parenchymal anatomy and architecture. The following anatomic structures and architectural components need to be considered:

Secondary pulmonary lobule

The secondary pulmonary lobule (SPL) is the smallest anatomical unit of the lung parenchyma that can be visualized on CT examinations (Fig. 1). SPL is a fundamental unit of lung structure, and an understanding of lobular anatomy is essential to the interpretation of thin-section CT of the lung [21]. Whereas in normal lung parenchyma, these polyhedral structures (size range: 1–2.5 cm diameter) [21] are only visualized in the anterior and lateral aspects of lung parenchyma, these become clearer when interstitial lung disease is present [20]. The SPL is surrounded by connective tissue septa. A central bronchovascular bundle, consisting of the pulmonary bronchiole and artery, perforates the center of the SPL, where the bronchiole divides into three to five terminal bronchioles [20]. At the origin of the terminal bronchioles is termed the “centrilobular” region.

Fig. 1

Secondary pulmonary lobule (Pulmonary arteries and bronchioles with a diameter of approximately 1 mm; interlobular septa with a thickness of approximately 0.1 mm; pulmonary vein and lymphatic branch with diameters of 0.5 mm each; and acinus—never visible on CT scans)

The acinus is located within the SPL. It is a portion of lung distal to a terminal bronchiole (the last purely conducting airway) and is supplied by a first-order respiratory bronchiole or bronchioles [22, 23]. Acini size ranges from 6 to 10 mm in diameter [24]. Therefore, on thin-section CT images, the secondary pulmonary lobule can be divided into three components: the interlobular septa; the centrilobular region; and the lobular parenchyma.

Interlobular septa

Secondary pulmonary lobule (SPL) is surrounded by the connective-tissue interlobular septa (ILS) that extend from the pleural surface of the lung inward [21]. The ILS consists of connective tissue, house pulmonary veins, and lymphatics and belongs to the peripheral interstitial fiber system. Interestingly, the pulmonary veins traverse within ILS and not parallel to the segmental or sub-segmental pulmonary artery branches and bronchi [22]. ILS are well matured in the lung periphery and most abundant in the apices of the lungs. They are concentrated near the anterior, lower, mediastinal, and diaphragmatic surfaces of the lung. Interestingly, the key structures for identifying SARS-CoV-2 have demonstrated thickening and increased visualization of these ILS at these anatomical locations [1, 4, 25].

Centrilobular region

The centrilobular region consists of the pulmonary artery and bronchioles that supply the SPL. Due to the complexity of the branches of the terminal bronchioles, branching generation is almost difficult to define and trace which branch of the pulmonary arteries and veins are aligned with the bronchioles that supply the SPL. Anatomically, lobular bronchioles are rarely seen under CT in normal individuals since their lumen measures approximately 1 mm in diameter, and their wall 0.15 mm, respectively [23]. However, the centrilobular region is well visualized through thickened walls, peribronchiolar inflammation, and/or intrabronchiolar fluid and mucus accumulations. Radiologically, they are best seen in the lung periphery which is more evident in patients with known SARS-CoV-2 [6, 10, 26].

Lobular parenchyma

Lobular parenchymal consists of the structures between outer septa and the centrilobular structures: which include the acini and intralobular septa which contain the pulmonary capillary bed [21]. These structures like others in the SPL are not visible by CT in healthy patients; however, in patients with known SARS-CoV-2, increased attenuation is seen [27].

Imaging characteristics

Ground glass opacity and SARS-CoV-2

Detection and management of GGO is an important imaging characteristic in SARS-CoV-2 screening, because persistent GGO are often a sign of poor progression in disease treatment [9, 14]. GGO is an area of hazy increased lung density with preservation of underlying bronchovascular margins that when obscured, they then form to become consolidation. However, imaging pitfalls exist by the following: 1: normally seen with exhalation, 2: volume averaging with thick collimation greater than 5 mm, 3: window settings too narrow, 4: independent from lung atelectasis, and finally 5: motion artifact due to respiration [28]. However, when reviewing the radiology-pathology correlation firstly, there is partial airspace filling by edema, hemorrhage and infection and secondly, interstitial thickening by inflammation edema or fibrosis. Therefore, the presence of consolidation can suggest that radiological characteristics of GGO demonstrate alveolar filling with the presence of reticular opacities or traction bronchiectasis suggests and interstitial thickening (and with traction bronchiectasis, fibrosis).

With the outbreak, a wide array literature has flooded the research environment with SARS-CoV-2 imaging patterns. The earliest description of GGO occurred with bilateral lung involvement on initial chest CT and then changed with time to become a consolidative pattern [29] (Fig. 2). This demonstrated a series of abnormal findings in 86% of patients, with a majority involving the lung parenchyma bilaterally [6] (Fig. 3).

Fig. 2

Top row demonstrates coronal reformats, and the bottom row axial sections: a, b ground glass opacity, c, d consolidation w air bronchogram and ground glass shadows and e, f multifocal interstitial and alveolar shadows

Fig. 3

63-year-old male presented with fever, dry cough, and chest pain for 4 days. He had a positive PCR and confirmed peripheral and central GGO bilaterally which was confirmed it was SARS-CoV-2

Degree of confidence

When the radiologist is reporting he/she shall choose the degree of confidence based on the CT findings. The most discriminating features for SARS-CoV-2 pneumonia included a peripheral distribution (80% vs. 57%, p < 0.001), ground-glass opacity (91% vs. 68%, p < 0.001) and vascular thickening (58% vs. 22%, p < 0.001) [30]. This was further validated between Chinese and US radiologists sensitivities (CHN:72–95%; USA: 73–93%) and specificities (CHN:24–94%; USA:93–100%) [30] with mixed results. Therefore, there are four degrees of confidence which are as follows:

High confidence will involve obvious peripheral, bilateral (multilobular), GGO with or without consolidation or intermixed septal thickening and multifocal GGO of rounded morphology with or without consolidation.

Intermediate confidence will involve GGO with or without consolidation that is not with a clear distribution and non-rounded and is also unilateral with or without consolidation.

Low confidence will demonstrate low patterns of consolidation without GGO and is very few in small GGO with non-rounded and non-peripheral distribution.

Alternative diagnosis the CT features demonstrate discrete nodules such as centrilobular, tree in bud, rounded small nodules, cavitation, perihilar GGO (with no peripheral GGO), smooth septal thickening with pleural effusions and fibrosis.

Principles of CT radiation dose reduction

The visualization of structures during low-dose CT (LDCT), one must consider the variable exposure that can be used in CT imaging and the fact that CT as an imaging modality is capable of yielding high radiation dose to the patient [31]. The mantra of ALARA (As Low as Reasonably Achievable) with CT scanning dictates that lung screening for SARS-CoV-2 may only be feasible if low-dose scans are employed. In this section, we will discuss the importance of LDCT in the imaging of GGO and consolidation as part of the imaging considerations in detecting disease.

There are a wide array of factors that can influence the radiation dose either directly or indirectly which can result in safe dose reduction without affecting image quality. First and foremost, it should be stated that radiation dose to the patient can be significantly reduced by carefully following proper techniques such as 1: correct patient centering by placing the chest in the center of the field of view [32], 2: reduce scan unnecessary scan length without affecting image quality [33], 3: shielding radiosensitive organs such as the mammary gland employing bismuth shields [34, 35], 4: organ-based tube current modulation [36] and 5: the radiologist screening unnecessary requests for CT scans [37].

The scanning parameters employed in the detection of GGO and consolidation involve: 1: modification of tube current is the simplest method of radiation dose reduction and has been a mainstay of radiation dose reduction methods [32], 2: employing 100 kVp protocol can reduce radiation dose by 44% while maintaining low-contrast detectability compared with a 120 kVp protocol [38], 3: helical pitch (< 1.3 mm) and collimation (40–80 mm) has no effect on image quality and in-plane resolution [39], 4: correct patient centering [40] in order to have optimal performance of the automatic exposure control with tailored according to patient weight, patients are potentially exposed to a 17–43% higher radiation dose from a chest CT [41] and 5: employing iterative reconstruction with low kVp is that of those when scanning with 120 kVp with the sensitivity to detect ground-glass opacities, ground-glass nodules and interstitial opacities decreased significantly, from 89% to 77%, 86% to 68% and 91% to 71%, respectively (all p values < 0.00001) [42] (Table 1).

Table 1

Summation of the factors that affect dose and image quality in CT scanning

Parameters affecting dose reduction	Effect on image quality	Remedies
Tube current (mAs)	Image noise will increase as radiation dose decreases	Weight-based protocols Noise Index threshold
Tube voltage (kVp)	A lower kVp setting can result in a substantial decrease in radiation dose and increase in image noise and low contrast resolution structures	100 kVp
Helical pitch	Increasing the pitch shortens the scanning time, therefore, decreasing dose, which in turn increases noise	Collimation 40–80 mm section thickness < 1.5 mm [20] 1 mm/rot > Pitch < 1.3 mm/rot
Scan range	The dose increases linearly, as scan length is increased	From apices to costophrenic recess
Automatic exposure control	Significant dose reductions by exposure adjustment to the overall size of a patient’s body, without significant change in image quality	Center the patient in the isocenter to prevent mA modulation over- or under-compensation of the airgap from the tube to the detector
Temporal resolution	Respiratory artifacts will reduce significantly due to the patients’ respiratory condition	Temporal resolution < 0.5 s
Iterative reconstruction	Can alter the texture pattern which is dependent on tube potential and current	Employ a gradual increase between hybrid and model-based iterative reconstruction and always compare it to filtered back projection as the gold standard

The detectability of lesions in CT chest scans (standard and low dose) can be variable between readers. Firstly, the performance of expert readers in lesion detection, including studies carried out on anthropomorphic phantoms [31, 43] and on lung trials with human participants [44, 45], will be discussed. Secondly, highlighting the scanning parameters of LDCT, the effect of a change in mA and image quality, and how those changes affect reader’s performance.

The reduction of radiation doses is remarkable between standard-dose CT and 50%-reduced-dose CT were 15.6–21.4 mSv and 7.8–10.7 mSv, respectively [46]. Zhu et al. (2008) also found that there was no significant compromise to image quality between CT chest scans that were obtained at 25 mAs and 115 mAs [47]. On the other spectrum, Li et al. (2002) report that lung opacities were missed on LDCT screening due overlapping structures, small faint nodules and opacities in the anatomical background. Furthermore, in this study, out of the 39 CT scans, 32 lung cancers were missed: 23 scans owing to detection errors and 16 scans owing to interpretation errors [48]. To reconcile findings between Zhu et al. (2008) and Li et al. (2002), Kubo et al. (2008) have reported in a review of five studies that a low–dose range of 20–50 mAs was sufficient for the detection of pulmonary nodules which can be those of GGO also [40, 47, 48].

When considering the effect of mA upon image quality, there is some evidence to show little significant statistical difference in the determination of image quality between CT chest scans that were obtained at 25 mAs and 115 mAs [47]. In a study by Prasad et al. (2002), it was reported that a 50%-reduction in radiation dose for chest CT patients still allowed for adequate visualization of the central and peripheral lung parenchyma. Another similar study found that the reduction of radiation dose between standard-dose CT and 50%-reduced-dose CT that could be achieved with acceptable image quality was up to 50%, from 15.6–21.4 mSv down to 7.8–10.7 mSv, respectively [46]. Gunnar (2013) has also reported that there was no impact upon the mean sensitivity of readers when viewing scans of lower image quality (those taken at 150 mAs compared with 40 mAs) [49]. Finally, image noise is key to determine the tube current needed to obtain the desired image quality at the prescribed noise index. Higher image noise implies the need for lower radiation dose. Lower image noise, on the other hand, needs increased radiation dose to achieve lower image noise or higher image quality [50]. This can be adjusted based on optimal patient centering in the center of the gantry and the mA modulation in both x, y and z will adjust to provide both minimum and maximum mA relative to the patient habitus.

Gordic et al. (2014) reported that dose levels achieved using LDCT (0.06 mSv) [31] were comparable to conventional CXR study (0.02 mSv for a PA CXR to 0.1 mSv for the lateral CXR) making LDCT an ethical choice on the radiation dose front [51]. Although Gordic et al. (2014) used an anthropomorphic phantom of an average size adult in their study, the results suggest that dose delivered to a human in the course of detection for pulmonary lesions/opacities can be reduced to 0.06 mSv when using a single-energy scanner at 100 kVp.

Zhu et al. (2004) on the other hand found that quality of a CT chest deteriorates when the mA is reduced, causing the noise levels to increase, and thereby affecting image quality. Various mAs (25, 40 and 115) levels were analyzed for image quality in chest CT and found that 25mAs or greater was a sufficient exposure parameter for the detection of pulmonary nodules. Exposure factors below 25 mAs were considered unsatisfactory, when examining nodules on lung window settings, due to the compromise of image quality and the change in noise levels [47].

GGO detection using LDCT

The fundamentals of clinical scanning should be to assist in the diagnosis (or exclusion) of disease, and therefore if the same clinical questions can be answered with a LDCT scan as a standard-dose CT scan, then the imperative to reduce the dose is more important than the overall image quality of the scan per say (24). In the early millennia, Kubo et al. (2008) attempted to study the visual search patterns when viewing CT images and possible disturbances when reducing image quality. Kubo et al. (2008) also undertook a review of studies that compared the lowest acceptable diagnostic mA for a range of chest CT examinations, including specific clinical indications such as GGO, pulmonary nodules, lymphoma, emboli and asbestos. The review looked at five CT studies conducted between 1998 and 2004, and the studies suggested that a current–time product of 20–50 mAs was sufficient for the detection of pulmonary nodules. In another part of Kubo’s review, it was also noted that two other studies conducted between 2000 and 2003 reported decreased nodule detectability in images obtained with less than 20 mAs, as an exposure parameter; hence, this may be an important imaging threshold to evaluate [40]. Finally, when comparing LDCT with standard dose CT, there was no statistical significance in nodule detection (LDCT vs. Standard; 518 vs. 533), with varying nodule sizes: < 3, 3–4.9, 5–6.9, 7–9.9 and ≥ 10 mm used in this and other studies [52-54].

Ground glass opacities and iterative reconstruction

Iterative reconstruction (IR) techniques has the ability to improve the image quality relative to the filtered back projection (FBP) techniques when undergoing low-dose chest CT [55]. The technical details of different proprietary IR techniques can be broad, but usually they are categorized into hybrid-based and model-based IR [56, 57]. Both their benefits and weakness are poorly understood and can lead to glossy images depending on which type of IR is employed [58]. In a recent study, it demonstrated that LDCT images demonstrated nonuniform noise in lung textured backgrounds and had somewhat degraded low-contrast spatial resolution [59]. In the current status, hybrid IR technique and LDCT imaging with 50 mAs enables noise and to maintain the detectability of GGO which is equivalent to the reference acquisitions of 200 mAs with filtered back projection [55]. While, at very low tube current at 10–20 mAs, the GGO attenuation is higher than the actual CT value with filtered back projection, with the potential of small GGO nodules could not be visualized because it was buried in increased noise that could be due to electronic noise because of the influence of the streak artifact. Equally, IR has been known to show that lower image noise, but did not provide any real improvement for radiologist’s evaluation of thin-section (< 2 mm) LDCT of the lung [60, 61]. Therefore, it is more important to employ the contrast to noise ratio with filtered back projection relative to the IR employed and the level of de-noising in order to maintain the actual CT value at use of use of low effective mAs [56, 57].

In summary, this article attempts to highlight some of the common chest CT signs that radiologists and emergency physicians are likely to encounter. With the increasing trend of using chest CT scans as an imaging tool to diagnose SAR-CoV-2, we emphasize that pattern recognition is the key, and this pictorial essay should serve as a guide to help establish a correct diagnosis coupled with correct scanner parameters to reduce radiation dose without affecting imaging quality in this tragic pandemic the world is facing.