The effects of normobaric and hyperbaric oxygenation on MRI signal intensities in T1-weighted, T2-weighted and FLAIR images in human brain

Abstract Background Dissolved oxygen has known paramagnetic effects in magnetic resonance imaging (MRI). The aim of this study was to compare the effects of normobaric oxygenation (NBO) and hyperbaric oxygenation (HBO) on human brain MRI signal intensities. Patients and methods Baseline brain MRI was performed in 17 healthy subjects (mean age 27.8 ± 3.2). MRI was repeated after exposure to the NBO and HBO at different time points (0 min, 25 min, 50 min). Signal intensities in T1-weighted, T2-weighted images and fluid attenuated inversion recovery (FLAIR) signal intensities of several intracranial structures were compared between NBO and HBO. Results Increased T1-weighted signal intensities were observed in white and deep grey brain matter, cerebrospinal fluid (CSF), venous blood and vitreous body after exposure to NBO as well as to HBO compared to baseline (Dunnett's test, p < 0.05) without significant differences between both protocols. There was also no significant difference in T2-weighted signal intensities between NBO and HBO. FLAIR signal intensities were increased only in the vitreous body after NBO and HBO and FLAIR signal of caudate nucleus was decreased after NBO (Dunnett's test, p < 0.05). The statistically significant differences in FLAIR signal intensities were found between NBO and HBO (paired t-test, p < 0.05) in most observed brain structures (paired t-test, p < 0.05). Conclusions Our results show that NBO and HBO alters signal intensities T1-weighted and FLAIR images of human brain. The differences between NBO and HBO are most pronounced in FLAIR imaging.


Introduction
Magnetic resonance imaging (MRI) of brain is a superior soft-tissue contrast method that is used for the assessment of a numerous neurological conditions such as multiple sclerosis and headaches, and used to characterize strokes and space-occupying lesions.Basic MRI brain screen protocol is a sim-ple non-contrast MRI comprising a group of basic MRI sequences when imaging the brain in cases of no particular condition is being sought (e.g.headache).The protocol is designed to obtain a good general overview of the brain.A standard screening protocol might include T 1 weighted imaging for anatomical overview, T 2 weighted imaging to evaluate basal cisterns, ventricular system and subdural spaces, and good visualization of flow voids in vessels, fluid attenuated inversion recovery imaging (FLAIR) to assess white-matter, diffusion weighted imaging (DWI) for multiple possible purposes (from the identification of ischemic stroke to the assessment of active demyelination).
Dissolved oxygen can be used as a contrast agent in MRI due to the paramagnetic properties of the dioxygen molecule O 2 1 .3][4][5][6] T 1 relaxation times shortening under the influence of increased partial pressure of oxygen (pO 2 ) in the inspired gas mixture is called tissue-oxygen-level-dependent effect (TOLD).1][12][13][14][15][16] In addition, pO 2 increase also affects spin-spin relaxation time (T 2 ) 13,17 ; however, results of available studies on the effect of pO 2 on T 2 relaxation times are controversial. 7,12FLAIR images of healthy volunteers also showed increased CSF signal intensity during 100% oxygen breathing. 18oncentration of dissolved oxygen is directly proportional to its partial pressure, pO 2 . 19High pO 2 values in arterial blood as well as in brain parenchyma can be achieved with normobaric 100% oxygenation (NBO) compared to breathing normobaric air (NBA).1][22] In few animal studies, it has been already observed that HBO had a more pronounced effect on T 1 and T 2 relaxation times compared to breathing NBO or NBA. 11,23o our knowledge, no human studies were performed studying the effect of HBO on MRI signal intensities.The aim of this study was to compare the effects of HBO and NBO on MRI signal intensities (e.g.T 1 , T 2 and FLAIR).

Patients and methods
The study was approved by The National Ethics Committee (No. 0120-203/2019/4).Research was conducted at the Institute of Physiology (University of Ljubljana, Faculty of Medicine).Informed consent was obtained from each subject.17 healthy volunteers (12 males and 5 females), age 20-40 years (mean age 27.8 ± 3.2), were enrolled in the study.Exclusion criteria were: history of a neurological disorder, non-MRI-compatible devices, a lung disease with FEV1/FVC < 60% and/or emphysema and/or pneumothorax, history of middle ear trauma or disease, therapy with platinum complexes, doxorubicin, bleomycin, disulfiram of mafenide acetate, pregnancy or claustrophobia.

Study protocol
MRI examination was performed before oxygen breathing protocol (baseline state), after HBO and after NBO with subsequent MRI on separate visits.NBO protocol was performed using a nonrebreather oxygen mask connected to a large reservoir supplied by 100% oxygen for 70 minutes.HBO protocol was performed in multiplace hyperbaric chamber (Kovinarska P&P, Slovenia) at 2.4 ATA with breathing of 100% oxygen for 70 minutes as shown in Figure 1.After each oxygen breathing protocol (NBO or HBO), MRI examination was repeated three times, i.e. immediately after the end of HBO or NBO, after 25 min and after 50 min.

MR image acquisition
The MRI imaging was performed on a 3T MRI system (TX Achieva Philips Netherlands) with the use of a 32-channel head coil.The MR examination consisted of:

MR data and statistical analysis
The MRI images were analysed using ImageJ free image analysis software (National Institutes of Health, USA).Mean signal values in distinct regions of interest (ROI) on T 1 -weighted, T 2 -weighted and FLAIR images were obtained: frontal white matter, thalamus, caudate nucleus, putamen, hippocampus, superior sagittal sinus, vitreous body and cerebrospinal fluid (CSF).Statistical analysis was performed using SigmaPlot 14.0 (Systat Software, Inc., USA).The signal intensities after NBO or HBO were compared to the baseline values.Shapiro-Wilk test and Brown-Forsythe were used to check for normality and equal variance.One-way repeated measurements analysis of variance (RM ANOVA) was used to test for differences between signal intensities before, immediately, 25 min and 50 min after NBO or HBO.In cases when Shapiro-Wilk or Brown-Forsythe test failed, Friedman RM ANOVA on Ranks was performed.If RM ANOVA showed statistically significant differences between groups of data, Dunnett's method for multiple comparisons was used to compare signal intensities at three time

Results
The results of T 1 -weighted signal intensities before, immediately, 25 min and 50 min after NBO or HBO are presented in Table 1.After NBO there was a statistically significant increase in T 1 -weighted signal intensity in all studied structures except for vitreous body and putamen (RM ANOVA, Dunnett's test, p < 0.05).In contrast, after HBO we observed a significant increase in T 1 -weighted signal intensities except for the superior sagittal sinus and CSF (Dunnett's test, p < 0,05).T 1 -weighted signal intensity was significantly higher immediately (0 min) as well as 25 min after the end of the HBO compared to T 1 -weighted signal intensity immediately and 25 min after NBO in vitreous body (paired t-test, p < 0.05).In contrast, there was no difference in signal intensities in T 1 -weighted images between HBO and NBO after 50 min.
The results of T 2 -weighted signal intensities before, immediately, 25 min and 50 min after NBO or HBO are presented in Table 2. T 2 -weighted signal intensities were increased only in frontal white matter and thalamus after NBO and in the superior sagittal sinus and vitreous body after HBO (Dunnett's test, p < 0.05).There was also no significant difference in T 2 -weighted signal intensities between HBO and NBO.The results of FLAIR signal intensities before, immediately, 25 min and 50 min after NBO or HBO are presented in Table 3. FLAIR signal intensities were increased only in the vitreous body after NBO and HBO, signal of caudate nucleus was decreased after NBO (Dunnett's test, p < 0.05).
The statistically significant differences in FLAIR signal intensities were found between NBO and HBO (paired t-test, p < 0.05) in caudate nucleus, thalamus, hippocampus and vitreous body at each time point (0 min, 25 min, 50 min).In addition, the differences were also observed between NBO in HBO in putamen and frontal white matter at 0 min and 25 min and in superior sagittal sinus at 25 min (paired t-test, p < 0.05).

Discussion
In the present study we observed increased signal intensity in T 1 -weighted imaging in frontal white matter, thalamus, caudate nucleus and hippocampus after NBO as well as HBO, in superior sagittal sinus and CSF after NBO and in vitreous body and putamen after HBO.Additionally, signal intensity was increased in T 2 -weighted imaging in frontal white matter and thalamus after NBO as well as in superior sagittal sinus and vitreous body after HBO.FLAIR signal intensities were increased only in the vitreous body after NBO and HBO.In contrast, FLAIR signal of caudate nucleus was decreased after NBO.Statistically significant differences between HBO and NBO were observed in FLAIR signal intensities of caudate nucleus, vitreous body, putamen, frontal white matter, hippocampus and thalamus and also in T 1 -weighted signal intensity of vitreous body.
In our study, T 1 -weighted signal intensity of brain structures increased progressively with time after NBO/HBO and was the highest 50 min after the end of both, HBO and NBO.This finding is in agreement with the paramagnetic effect of O 2 .Increased level of dissolved paramagnetic molecular O 2 shortens T 1 -relaxation times due to dipoldipol interactions and increases signal intensity on T 1 -weighted images. 4,7-16 23 24Relaxation rate (R 1 = 1/T 1 ) increases proportionally with increasing pO 2 in inspired gas mixture, the increase being linear or logarithmic when in normobaric or hyperbaric conditions, respectively. 23,24The various increase of T 1 -weighted signal intensities in the observed tissues might be explained by increased microvascular pO 2 as well as by differences in tissue oxygenation. 7The sustained increase in T 1 -weighted signal intensity is further supported by a study of Rockswold et al. which showed significantly elevated brain tissue pO 2 30 min after the end of HBO and NBO. 25 In contrast to Rockswold et al., we failed to observe a peak in T 1 -weighted signal intensity immediately after the end of oxygen therapy.A possible explanation is that the time delay between HBO/NBO and MRI was too long to detect the peak.
We observed progressive increase in T 1weighted signal intensities after both NBO and HBO along with MRI imaging time, with the highest signal increase at the end of imaging protocol.This phenomenon could not be attributed solely to changes in pO 2 , but also to the effect of ROS on T 1 and T 2 -weighted images.Since ROS such as hydroxyl and superoxide radicals contain unpaired electrons, they also exhibit strong paramagnetic effect (a strong T 1 relaxation times shortening) and only a small, statistically insignificant reduction of T 2 relaxation times. 5,6Additional point to consider is that distinct neurons respond to oxidative stress differently 26,27 , which leads us to presumption that the effect of HBO-induced oxidative stress would lead to different levels of ROS and thus different effect on T 1 and T 2 weighted signal intensities in various brain regions.
The increase of T 1 -weighted signal intensities was more pronounced in frontal white matter and thalamus after HBO compared to NBO.This could be explained by altered O 2 diffusion after HBO.We observed increased signal intensity in superior sagittal sinus and CSF only after NBO, but not after HBO.Longer time delay between HBO and MRI most likely lowered pO 2 in the aforementioned fluids before the beginning of MRI.We showed that T 1 -weighted signal intensity of vitreous body was significantly increased immediately after the end of HBO exposure and then decreased in subsequent imaging blocks.This is in accordance with expected pO 2 dynamics in vitreous body, described by Shui et al. 28 Surprisingly, after the exposure to NBO, no increase in vitreous T 1 -weighted signal intensity was observed.A possible explanation is that lower vitreous pO 2 (as achieved during NBO compared to HBO) dropped to baseline level before the beginning of the MRI.
In our study, there were statistical differences in T 2 -weighted signal intensities between baseline and after NBO in frontal white matter and thalamus.This is in accordance with Wu et al. who observed significant differences in T 1 and T 2 between grey and white matter after inhalation of NBO. 12 According to Wu et al., T 2 relaxation time increases in rat brain with hyperoxia.In contrast, Tadamura et al. did not observe this effect in human "nonbrain" tissues (myocardium, spleen, liver, subcutaneous fat, skeletal muscle and bone marrow). 7herefore, it is possible that the effect of hyperoxia on T 2 -weighted signal intensities appears to vary in different tissues.In the present study, a signifi- T 2 shortening analogous to effect on T 1 (although the effect on T 2 is much smaller) and T 2 lengthening due to diffusion of water protons through field inhomogeneities induced by deoxyhemoglobin generated field gradients (blood-oxygen-leveldependent (BOLD) effect). 13,17An increased T 2weighted signal intensity after NBO and HBO in our study suggests that in human brain structures and vitreous body the paramagnetic effect of oxygen on T 2 relaxation times shortening prevails over BOLD effect.Our results show statistically significant differences between HBO and NBO were observed in FLAIR signal intensities in different brain structures particularly those that are close to CSF spaces.These results are in accordance with previous studies which showed that in patients receiving 100% NBO elevated pO 2 leads to incomplete signal suppression of CSF in FLAIR imaging. 29,30The elevated pO 2 most likely favors O 2 entry into the CSF not through the choroid plexus but directly through the walls of arteries and arterioles on the brain surface. 30Since in HBO there is up to 2.5 times higher pO 2, this effect in FLAIR imaging is more pronounced.We observed increase in FLAIR signal of vitreous body immediately after HBO/ NBO exposure and then a subsequent decrease in time -again, this is in accordance with expected pO 2 dynamics in vitreous body, as described by Shui et al. 28 The results of the present study could have also some clinical implications.Namely, the prolonged intubation induces changes of signal intensities in T 1 -weighted and FLAIR images of brain MRI 31 similar as those observed in our study after NBO.Knowing that prolonged oxygenation induces paramagnetic effects in brain tissues as observed in our study, it is important to take this into account when interpreting brain MRI in intubated patients or in patients after HBO therapy.
The present study has several limitations.First, we failed to show significant differences in MRI signal intensities in brain structures after HBO compared to NBO.It would be expected that brain tissue pO 2 is significantly higher after HBO compared to NBO 21 due to higher concentration of dissolved O 2 during HBO. 19The only exception was T 1 -weighted signal intensity of vitreous body immediately and 25 min after HBO compared to NBO.One possible explanation is that MRI was performed with time delay of 15 minutes after the end of HBO due to logistics.Perhaps with shorter time delay the peak in T 1 -weighted signal intensities could be observed similarly as in the study of Rockswold et al. 25 Additionally, the present study was semiquantitative using clinical head MRI protocol and the next step would be more quantitative approach using T 1 mapping and T 2 mapping.Furthermore, we did not measure brain tissue pO 2 nor levels of ROS, which would help to explain the observed changes in signal intensities in T 1weighted and T 2 -weighted images.Since our study was performed in vivo in a group of volunteers measuring of brain tissue pO 2 seems rather controversial.We could only measure pO 2 in arterial blood, however these results do not reflect brain tissue pO 2 directly.However, according to the reference, at 3 ATA pO 2 in arterial blood increases to nearly 270 kPa and in tissue to above 53 kPa. 32In contrast, in NBO conditions, partial pressure of pO 2 in the brain is expected to be only between 4 -6.4 kPa according to study of Meixensberger et al. 33 These values are much lower than during HBO.Therefore, we expected similar tissue pO 2 differences between HBO and NBO in the present study protocol.
In conclusion, the increased T 1 -weighted signal intensities were observed in white and grey brain tissues, brain fluids and vitreous body after NBO as well as HBO, without significant differences between both protocols.In addition, the structure limited and diverse signal intensity increase was observed in T 2 -weighted imaging and FLAIR after NBO and HBO.However, the prospective quantitative studies are needed to further clarify the effects of NBO and HBO breathing on MRI in human brain.

FIGURE 2 .
FIGURE 2. Representative MRI images in healthy subject at baseline, immediately after the end, after 25 min and after 50 min of NBO or HBO.