Multi-detector computed tomography (MDCT), which allows a thin-slice scan and has been used emergently in patients in various clinical settings, is useful in the detection of small amounts of vascular gas. A post-processing algorithm using MDCT data that can generate multiplanar images allows assessment of anatomical location of vascular gas [1]. Unlike arterial gas, small venous gas produces no clinical signs and symptoms. However, the embolism can easily occur when gas enters the cerebral vein because the air blood interaction causes the development of a network comprising air bubbles and fibrin strands that intersperses with aggregates of the platelets, red blood cells, and fat globules [2]. We encountered vascular gas in various clinical situations, including iatrogenic complications. The morbidity and mortality associated with vascular gas resulting in embolism can be prevented through a better understanding of its pathophysiology and increased awareness of various clinical settings where they occur. Emergency departments are especially high-risk locations for venous air embolism because of frequent use of venous access (central and peripheral) and intravenous drug and fluid infusions [3].

We present this article analyzing the etiology of air entry into the veins on emergent MDCT scanning in our experienced cases as well as extracted published cases by systematic literature reviews.

Approval for this retrospective study was obtained from our institutional review board, and a waiver for informed consent was obtained.

Local cases
One of the authors reviewed our MDCT reports database between 2008 and 2015 to select patients with vascular gas excluding contaminated air at intravenous injection of contrast material. Clinical characteristics, imaging findings, treatment and outcome were collected. MDCT scan of the trunk or the head were performed using a 16-row MDCT unit (Aquilion16; Toshiba Medical Systems, Ohtawara, Japan). The trunk from the neck to the pelvis was scanned with 16×1 mm collimation (16 detectors with 1-mm section thickness), a beam pitch of 0.9375, a rotation speed of 0.5 second, a table speed of 15.0 mm per rotation and a tube current determined by automated tube modulation. The head images were acquired with 16x0.5 mm collimation, a beam pitch of 0.6875, a rotation speed, of 1 second, a table speed of 5.5 mm per rotation and 120-kVp tube current. Images were reconstructed at the MDCT console to a section thickness of 5 mm. Each image was sent to a picture archiving and communication system (PACS, Synapse ver. 3.1, Fuji-Film Medical Co., Tokyo, Japan). The area of each axial scan of venous gas was determined on the monitor by tracing the margin of the gas with a built-in cursor controlled by a track ball. Section volumes were then calculated by multiplying by section thickness (1 mm in the truncal scan and 0.5 mm in head scan) and were added to determine the total gas volume. Coronal reconstructed images were obtained with a 1.0 mm thickness and a reconstruction interval of 0.8 mm. Areas of air-containing pixels of axial images were multiplied by the reconstruction interval of 0.8 mm between contiguous images and all of these subvolumes were added.

Literature search
Cases were extracted from English literature in PubMed (NCBI) for the following search term: "venous gas", "venous gas embolism", "arterial gas", and "arterial gas embolism" in combination with specific organs in our local cases. The case report papers and our local cases were summarized according to causes. In addition, we analyzed the etiology of air entry into the vessel, hypothesis of traveling gas to the vessels and its clinical significance to increase awareness of the clinical settings where the vascular gas occurred.

Our local series demonstrated single or multiple collections of vascular gas in 15 patients who were three with iatrogenic vascular gas (Table 1) and 12 with non-iatrogenic gas (Table 2); the most frequent was cerebral vascular gas (CVG, n = 11, 0.8–12 mL) followed by hepatic vascular gas (n = 10, 0.4–256 mL). A total of 20 case reports, including one review article with 165 included cases, were extracted by PubMed search terms related to vascular gas in combination with "cerebral" or "hepatic". The majority of literature reported on single patients. MDCT or computed tomography scan was performed in 129 of the 165 patients. We selected the 129 patients, including 59 with iatrogenic-caused vascular gas (Table 3) and 70 with non-iatrogenic-caused vascular gas (Table 4).

Of the accumulative 144 cases which were 129 published reporting cases and 15 local cases, 62 (43.1%) including 3 local cases were iatrogenic and 82 (56.9%) including 12 local cases were non-iatrogenic. The most frequent entity of iatrogenic vascular gas in the central venous catheter-related CVG (48 cases, 77.4%) with 39.5% mortality. Our local series had a patient with CVG-related cerebral embolism in which no intracranial gas but 1.2 mL of the left ventricular gas was shown in Figure 1. In this patient, the presence of the left ventricular gas supported the paradoxical embolism although no intra-cardiac right to left shunt was shown in the color Doppler imaging.

In the local case series, massive (>100 mL) hepatic portal venous gas was observed in the remaining two patients with iatrogenic complications. The patient experienced cardiac arrest during the endoscopic procedure with balloon dilatation. Post-resuscitation MDCT images Figure 2 demonstrated multiple collections of gas in the right ventricle, the pulmonary arterial trunk, infrarenal inferior vena cava, the ascending thoracic aorta and cerebral vessels. The barotrauma during the endoscopic procedure with balloon dilatation for bowel is a possible entity of massive hepatic portal venous gas, according to four published reporting cases. Massive hepatic portal venous gas associated with cystic venous gas in another local case occurred due to decompression sickness with sub-atmospheric intraperitoneal pressure and the entry of gas into injured gastric wall vein via a mal-positioned gastrostomy tube (Figure 3B) . One published case reported massive hepatic portal venous gas in decompression sickness with injured jejunum. Other etiologies of iatrogenic cerebral gas in published cases included percutaneous lung biopsy or radiofrequency ablation.

Aggressive resuscitation was the most frequent (66.7%, n = 8) cause of vascular gas in non-iatrogenic local cases: intracranial gas in three patients, hepatic vascular gas in three patients and both in two patients. MDCT images did not provide the conclusive evidence of cerebral or hepatic vascular anatomies. MDCT images of craniofacial trauma in the local series showed gas in the overt cerebral vein (Figure 4C). Non-iatrogenic hepatic portal venous gas occurred in patients with acute colonic pseudo-obstruction in two local cases. The local case series included gas replacement of systemic arteries caused by positive pressure ventilation in the patient with aortobronchial fistula secondary to the aneurysmal rupture. MDCT in the patient showed conclusive images of intrahepatic arterial gas (Figure 3A) and cerebral arterial gas (Figure 4A-B).

The entry of gas into the circulation is mostly an iatrogenic problem but may also occur in many fields of medicine including emergency medicine. The entry of gas into the blood stream requires a pressure gradient favoring the passage of gas into the blood vessel. This occurs when venous pressure is negative relative to atmospheric pressure or when gas is forced under pressure directly or indirectly into the bloodstream [3] [16]. A 14 G catheter enables the gas to flow at a rate of about 100 mL/sec, with a pressure difference of only 5 cm H₂O [16]. Entering gas into the circulation occurs by barotrauma caused by positive pressure ventilation and insufflation of gas into the peritoneal cavity during endoscopy can also cause gas to enter the circulation [13].

The most frequently iatrogenic vascular gas was central venous catheter-related cerebral gas embolisms. Approximately, 70% of the patients with gas embolisms had intracranial gas bubbles, which were most frequently located in the subarachnoid space [5]. MDCT provides no conclusive anatomy with regards to the gas in the subarachnoid space although it can depict a small gas bubble. Therefore, the gas bubble was interpreted by some authors as intra-arterial caused by paradoxical embolism and by others as intra-venous caused by retrograde embolism [6]. Hagen et al. [25] demonstrated in a large number of autopsy specimens of human hearts that 27.3% had foramen ovales of 1 to 10 mm in diameter. Increasing right atrium pressure from entering gas is suggested to cause right to left shunting in the fossa ovale which is usually occluded [26]. In the majority of published cases, there was no reference to shunt investigation. The presence of the intra-cardiac right-left shunting on transesophageal echocardiogram is not always shown in patients with paradoxical embolisms. In a local case with central venous catheter trouble, the presence of left ventricular gas supported a paradoxical embolism, although no intra-cardiac right to left shunt was shown in the echocardiogram. Paradoxical arterial embolism is an acceptable mechanism because many cases reported central venous catheter-related coronary arterial gas embolism. But Yeslaris et al. [6] claimed that retrograde gas embolism possibly occurred through the internal jugular vein into intracranial veins when massive amounts of air entered the jugular vein via a large bored central venous catheter. Retrograde gas movement through the internal jugular vein to the intracranial venous sinus is considered as the mechanism of post-resuscitation cerebral gas or hepatic gas [20] [21] Post-resuscitation cerebral and hepatic gas are explained by the retrograde gas movements; at rapid infusion of large fluid volumes, contaminated air via the jugular or femoral central venous catheter is allow to enter the cerebral vein or hepatic vein, respectively [21].

Hepatic portal venous gas has been described in non-obstructive dilatation of bowels in acute colonic pseudo-obstruction (Ogilvie's syndrome) and Crohn's disease [23] [24]. Massive amounts of hepatic portal venous gas occurred due to the barotrauma during endoscopic procedure with balloon dilatation. The increased intra-luminal pressure causes mucosal tears within the bowel, which allows gas to enter submucosal veins and flow to the hepatic portal vein. A large amount of hepatic portal venous gas is not dangerous because the gas is not rapidly or easily transmitted to the central circulation. However, it may enter the inferior vena cava via the portal vein-hepatic artery capillary anastomosis when the gas is forced to be infused at a rapid and massive flow. The mechanical obstruction of the right ventricular pulmonary outflow tract and pulmonary vasculature occurs when a large amount of gas enters the right ventricle. Hepatic portal gas was observed in 20 cases with a 15% mortality rate, in the published case reports and our local case series.

Gas embolus does produce clinical signs and symptoms which may mimic an acute cardiopulmonary or cerebrovascular event. The spectrum of vascular gas causes detected with MDCT is widening. Radiologists as well as emergency department physicians, should be familiar with the increasing number of vascular gas causes in addition to knowledge of the patient's clinical history. A careful search for clues on MDCT images was useful in discussing the etiology of vascular gas entry points and increased awareness of the emergent clinical settings where the vascular gas occurred.

1. Jackowski C, Thali M, Sonnenschein M, et al. Visualization and quantification of air embolism structure by processing postmortem MSCT data. J Forensic Sci 2004 Nov;49(6):1339–42.   [CrossRef]   [Pubmed]    
2. O'Quin RJ, Lakshminarayan S. Venous air embolism. Arch Intern Med 1982 Nov;142(12):2173–6.   [CrossRef]   [Pubmed]    
3. Sviri S, Woods WP, van Heerden PV. Air embolism: A case series and review. Crit Care Resusc 2004 Dec;6(4):271–6.   [Pubmed]    
4. Torimoto I, Takebayashi S, Manak H, Nakamura K, Morimura N. Systemic air-angiogram appearance caused by aortobronchial fistula in a cardiac arrest patient with ruptured aortic arch aneurysm. Edorium J Radiol 2015;1:6–9.    
5. Pinho J, Amorim JM, Araújo JM, et al. Cerebral gas embolism associated with central venous catheter: Systematic review. J Neurol Sci 2016 Mar 15;362:160–4.   [CrossRef]   [Pubmed]    
6. Yesilaras M, Atilla OD, Aksay E, Kilic TY. Retrograde cerebral air embolism. Am J Emerg Med 2014 Dec;32(12):1562.e1–2.   [CrossRef]   [Pubmed]    
7. Bou-Assaly W, Pernicano P, Hoeffner E. Systemic air embolism after transthoracic lung biopsy: A case report and review of literature. World J Radiol 2010 May 28;2(5):193–6.   [CrossRef]   [Pubmed]    
8. Jeannin A, Saignac P, Palussière J, Gékière JP, Descat E, Lakdja F. Massive systemic air embolism during percutaneous radiofrequency ablation of a primary lung tumor. Anesth Analg 2009 Aug;109(2):484–6.   [CrossRef]   [Pubmed]    
9. Mokhlesi B, Ansaarie I, Bader M, Tareen M, Boatman J. Coronary artery air embolism complicating a CT-guided transthoracic needle biopsy of the lung. Chest 2002 Mar;121(3):993–6.   [CrossRef]   [Pubmed]    
10. Rapicetta C, Lococo F, Levrini G, Ricchetti T, Sgarbi G, Paci M. Asymptomatic air-embolism following percutaneous radiofrequency ablation of lung tumor: Rare or underestimated complication? Thoracic Cancer 2015 Mar;6(2):227–9.   [CrossRef]   [Pubmed]    
11. Lee CG, Kang HW, Song MK, et al. A case of hepatic portal venous gas as a complication of endoscopic balloon dilatation. J Korean Med Sci 2011 Aug;26(8):1108–10.   [CrossRef]   [Pubmed]    
12. Eoh EJ, Derrick B, Moon R. Cerebral Arterial Gas Embolism During Upper Endoscopy. A A Case Rep 2015 Sep 15;5(6):93–4.   [CrossRef]   [Pubmed]    
13. Ma AS, Ewing I, Murray CD, Hamilton MI. Hepatic portal venous gas and portal venous thrombosis following colonoscopy in a patient with terminal ileal Crohn's disease. BMJ Case Rep 2015 May 4;2015. pii: bcr2014206854.   [CrossRef]   [Pubmed]    
14. Pandurangadu AV, Paul JA, Barawi M, Irvin CB. A case report of cerebral air embolism after esophagogastroduodenoscopy: diagnosis and management in the emergency department. J Emerg Med 2012 Dec;43(6):976–9.   [CrossRef]   [Pubmed]    
15. Schatz TP, Nassif MO, Farma JM. Extensive portal venous gas: Unlikely etiology and outcome. Int J Surg Case Rep 2015;8C:134–6.   [CrossRef]   [Pubmed]    
16. Makino Y, Shimofusa R, Hayakawa M, et al. Massive gas embolism revealed by two consecutive postmortem computed-tomography examinations. Forensic Sci Int 2013 Sep 10;231(1-3):e4–10.   [CrossRef]   [Pubmed]    
17. Yang MS, Yang TH, Ou CH, et al. Iatrogenic and fatal arterial air embolism during the CT scan. J Chin Med Assoc 2011 Apr;74(4):188–91.   [CrossRef]   [Pubmed]    
18. Pampín-Huerta FR, Mourelo-Fariña M, Galeiras-Vázquez RM, Vázquez-Vigo R, Delgado-Roel M, Martínez Muñiz A. Cerebral arterial gas embolism secondary to gunshot chest wound. [Article in Spanish]. Med Intensiva 2014 Jan-Feb;38(1):56–8.   [CrossRef]   [Pubmed]    
19. Shiotani S, Kohno M, Ohashi N, Atake S, Yamazaki K, Nakayama H. Cardiovascular gas on non-traumatic postmortem computed tomography (PMCT): the influence of cardiopulmonary resuscitation. Radiat Med 2005 Jun;23(4):225–9.   [Pubmed]    
20. Arata S, Suzuki J, Moriwaki Y, et al. Characteristics of high-echoic objects in the hepatic vessels of patients with cardiopulmonary arrest: a prospective cohort study. Emerg Med J 2012 Mar;29(3):213–8.   [CrossRef]   [Pubmed]    
21. Shah PA, Cunningham SC, Morgan TA, Daly BD. Hepatic gas: widening spectrum of causes detected at CT and US in the interventional era. Radiographics 2011 Sep-Oct;31(5):1403–13.   [CrossRef]   [Pubmed]    
22. Morita S, Yamagiwa T, Inokuchi S. Portal venous gas on computed tomography imaging in patients with decompression sickness. J Emerg Med 2013 Jul;45(1):e7–11.   [CrossRef]   [Pubmed]    
23. Inokuchi R, Fukuda T, Yahagi N, Nakamura K. Severe hepatic portal venous gas that spontaneously resolved within a day. Intensive Care Med 2014 Sep;40(9):1369.   [CrossRef]   [Pubmed]    
24. Kearns K, Tran Van D, Alberti N, Fontaine B, Fritsch N. Hepatic portal venous gas: surgery or not surgery?. [Article in French]. Ann Fr Anesth Reanim 2013 Nov;32(11):803–6.   [CrossRef]   [Pubmed]    
25. Hagen PT, Scholz DG, Edwards WD. Incidence and size of patent foramen ovale during the first 10 decades of life: an autopsy study of 965 normal hearts. Mayo Clin Proc 1984 Jan;59(1):17–20.   [CrossRef]   [Pubmed]    
26. Langholz D, Louie EK, Konstadt SN, Rao TL, Scanlon PJ. Transesophageal echocardiographic demonstration of distinct mechanisms for right to left shunting across a patent foramen ovale in the absence of pulmonary hypertension. J Am Coll Cardiol 1991 Oct;18(4):1112–7.   [CrossRef]   [Pubmed]