QQ-NET - using deep learning to solve quantitative susceptibility mapping and quantitative blood oxygen level dependent magnitude (QSM+qBOLD or QQ) based oxygen extraction fraction (OEF) mapping

Quantitative BOLD (qBOLD) imaging of oxygen metabolism and blood oxygenation in the human body: A scoping review

PURPOSE

There are many approaches to the quantitative BOLD (qBOLD) technique described in the literature, differing in pulse sequences, MRI parameters and data processing. Thus, in this review, we summarized the acquisition methods, approaches used for oxygenation quantification and clinical populations investigated.

METHODS

Three databases were systematically searched (Medline, Embase, and Web of Science) for published research that used qBOLD methods for quantification of oxygen metabolism. Data extraction and synthesis were performed by one author and reviewed by a second author.

RESULTS

A total of 93 relevant papers were identified. Acquisition strategies were summarized, and oxygenation parameters were found to have been investigated in many pathologies such as steno-occlusive diseases, stroke, glioma, and multiple sclerosis disease.

CONCLUSION

A summary of qBOLD approaches for oxygenation measurements and applications could help researchers to identify good practice and provide objective information to inform the development of future consensus recommendations.

Multi-Echo Complex Quantitative Susceptibility Mapping and Quantitative Blood Oxygen Level-Dependent Magnitude (mcQSM + qBOLD or mcQQ) for Oxygen Extraction Fraction (OEF) Mapping

Oxygen extraction fraction (OEF), the fraction of oxygen that tissue extracts from blood, is an essential biomarker used to directly assess tissue viability and function in neurologic disorders. In ischemic stroke, for example, increased OEF can indicate the presence of penumbra-tissue with low perfusion yet intact cellular integrity-making it a primary therapeutic target. However, practical OEF mapping methods are not currently available in clinical settings, owing to the impractical data acquisitions in positron emission tomography (PET) and the limitations of existing MRI techniques. Recently, a novel MRI-based OEF mapping technique, termed QQ, was proposed. It shows high potential for clinical use by utilizing a routine sequence and removing the need for impractical multiple gas inhalations. However, QQ relies on the assumptions of Gaussian noise in susceptibility and multi-echo gradient echo (mGRE) magnitude signals for OEF estimation. This assumption is unreliable in low signal-to-noise ratio (SNR) regions like disease-related lesions, risking inaccurate OEF estimation and potentially impacting clinical decisions. Addressing this, our study presents a novel multi-echo complex QQ (mcQQ) that models realistic Gaussian noise in mGRE complex signals. We implemented mcQQ using a deep learning framework (mcQQ-NET) and compared it with the existing QQ-NET in simulations, ischemic stroke patients, and healthy subjects, using identical training and testing datasets and schemes. In simulations, mcQQ-NET provided more accurate OEF than QQ-NET. In the subacute stroke patients, mcQQ-NET showed a lower average OEF ratio in lesions relative to unaffected contralateral normal tissue than QQ-NET. In the healthy subjects, mcQQ-NET provided uniform OEF maps, similar to QQ-NET, but without unrealistically high OEF outliers in areas of low SNR, such as SNR ≤ 15 (dB). Therefore, mcQQ-NET improves OEF accuracy by more accurately reflecting realistic Gaussian noise in complex mGRE signals. Its enhanced sensitivity to OEF abnormalities, based on more realistic biophysics modeling, suggests that mcQQ-NET has potential for investigating tissue variability in neurologic disorders.

mcLARO: Multi-contrast learned acquisition and reconstruction optimization for simultaneous quantitative multi-parametric mapping

PURPOSE

To develop a method for rapid sub-millimeter T1 , T2 ,T 2 * $$ {\mathrm{T}}_2^{\ast } $$ , and QSM mapping in a single scan using multi-contrast learned acquisition and reconstruction optimization (mcLARO).

METHODS

A pulse sequence was developed by interleaving inversion recovery and T2 magnetization preparations and single-echo and multi-echo gradient echo acquisitions, which sensitized k-space data to T1 , T2 ,T 2 * $$ {\mathrm{T}}_2^{\ast } $$ , and magnetic susceptibility. The proposed mcLARO optimized both the multi-contrast k-space under-sampling pattern and image reconstruction based on image feature fusion using a deep learning framework. The proposed mcLARO method withR = 8 $$ R=8 $$ under-sampling was validated in a retrospective ablation study and compared with other deep learning reconstruction methods, including MoDL and Wave-MoDL, using fully sampled data as reference. Various under-sampling ratios in mcLARO were investigated. mcLARO was also evaluated in a prospective study using separately acquired conventionally sampled quantitative maps as reference standard.

RESULTS

The retrospective ablation study showed improved image sharpness of mcLARO compared to the baseline network without the multi-contrast sampling pattern optimization or image feature fusion module. The under-sampling ratio experiment showed that higher under-sampling ratios resulted in blurrier images and lower precision of quantitative values. The prospective study showed that small or negligible bias and narrow 95% limits of agreement on regional T1 , T2 ,T 2 * $$ {\mathrm{T}}_2^{\ast } $$ , and QSM values by mcLARO (5:39 mins) compared to reference scans (40:03 mins in total). mcLARO outperformed MoDL and Wave-MoDL in terms of image sharpness, noise suppression, and artifact removal.

CONCLUSION

mcLARO enabled fast sub-millimeter T1 , T2 ,T 2 * $$ {\mathrm{T}}_2^{\ast } $$ , and QSM mapping in a single scan.

Evaluating Physiological MRI Parameters in Patients with Brain Metastases Undergoing Stereotactic Radiosurgery-A Preliminary Analysis and Case Report

Brain metastases occur in ten to thirty percent of the adult cancer population. Treatment consists of different (palliative) options, including stereotactic radiosurgery (SRS). Sensitive MRI biomarkers are needed to better understand radiotherapy-related effects on cerebral physiology and the subsequent effects on neurocognitive functioning. In the current study, we used physiological imaging techniques to assess cerebral blood flow (CBF), oxygen extraction fraction (OEF), cerebral metabolic rate of oxygen (CMRO2) and cerebrovascular reactivity (CVR) before and three months after SRS in nine patients with brain metastases. The results showed improvement in OEF, CBF and CMRO2 within brain tissue that recovered from edema (all p ≤ 0.04), while CVR remained impacted. We observed a global post-radiotherapy increase in CBF in healthy-appearing brain tissue (p = 0.02). A repeated measures correlation analysis showed larger reductions within regions exposed to higher radiotherapy doses in CBF (rrm = -0.286, p < 0.001), CMRO2 (rrm = -0.254, p < 0.001), and CVR (rrm = -0.346, p < 0.001), but not in OEF (rrm = -0.004, p = 0.954). Case analyses illustrated the impact of brain metastases progression on the post-radiotherapy changes in both physiological MRI measures and cognitive performance. Our preliminary findings suggest no radiotherapy effects on physiological parameters occurred in healthy-appearing brain tissue within 3-months post-radiotherapy. Nevertheless, as radiotherapy can have late side effects, larger patient samples allowing meaningful grouping of patients and longer follow-ups are needed.

Assessing Cerebral Oxygen Metabolism Changes in Patients With Preeclampsia Using Voxel-Based Morphometry of Oxygen Extraction Fraction Maps in Magnetic Resonance Imaging

OBJECTIVE

The objective of this study was to analyze the different brain oxygen metabolism statuses in preeclampsia using magnetic resonance imaging and investigate the factors that affect cerebral oxygen metabolism in preeclampsia.

MATERIALS AND METHODS

Forty-nine women with preeclampsia (mean age 32.4 years; range, 18-44 years), 22 pregnant healthy controls (PHCs) (mean age 30.7 years; range, 23-40 years), and 40 non-pregnant healthy controls (NPHCs) (mean age 32.5 years; range, 20-42 years) were included in this study. Brain oxygen extraction fraction (OEF) values were computed using quantitative susceptibility mapping (QSM) plus quantitative blood oxygen level-dependent magnitude-based OEF mapping (QSM + quantitative blood oxygen level-dependent imaging or QQ) obtained with a 1.5-T scanner. Voxel-based morphometry (VBM) was used to investigate the differences in OEF values in the brain regions among the groups.

RESULTS

Among the three groups, the average OEF values were significantly different in multiple brain areas, including the parahippocampus, multiple gyri of the frontal lobe, calcarine, cuneus, and precuneus (all P-values were less than 0.05, after correcting for multiple comparisons). The average OEF values of the preeclampsia group were higher than those of the PHC and NPHC groups. The bilateral superior frontal gyrus/bilateral medial superior frontal gyrus had the largest size of the aforementioned brain regions, and the OEF values in this area were 24.2 ± 4.6, 21.3 ± 2.4, and 20.6 ± 2.8 in the preeclampsia, PHC, and NPHC groups, respectively. In addition, the OEF values showed no significant differences between NPHC and PHC. Correlation analysis revealed that the OEF values of some brain regions (mainly involving the frontal, occipital, and temporal gyrus) were positively correlated with age, gestational week, body mass index, and mean blood pressure in the preeclampsia group (r = 0.361-0.812).

CONCLUSION

Using whole-brain VBM analysis, we found that patients with preeclampsia had higher OEF values than controls.

LARO: Learned acquisition and reconstruction optimization to accelerate quantitative susceptibility mapping

Quantitative susceptibility mapping (QSM) involves acquisition and reconstruction of a series of images at multi-echo time points to estimate tissue field, which prolongs scan time and requires specific reconstruction technique. In this paper, we present our new framework, called Learned Acquisition and Reconstruction Optimization (LARO), which aims to accelerate the multi-echo gradient echo (mGRE) pulse sequence for QSM. Our approach involves optimizing a Cartesian multi-echo k-space sampling pattern with a deep reconstruction network. Next, this optimized sampling pattern was implemented in an mGRE sequence using Cartesian fan-beam k-space segmenting and ordering for prospective scans. Furthermore, we propose to insert a recurrent temporal feature fusion module into the reconstruction network to capture signal redundancies along echo time. Our ablation studies show that both the optimized sampling pattern and proposed reconstruction strategy help improve the quality of the multi-echo image reconstructions. Generalization experiments show that LARO is robust on the test data with new pathologies and different sequence parameters. Our code is available at https://github.com/Jinwei1209/LARO-QSM.git.

Rapid estimation of 2D relative B 1 + -maps from localizers in the human heart at 7T using deep learning

PURPOSE

Subject-tailored parallel transmission pulses for ultra-high fields body applications are typically calculated based on subject-specific B 1 + $$ {\mathrm{B}}_1^{+} $$ -maps of all transmit channels, which require lengthy adjustment times. This study investigates the feasibility of using deep learning to estimate complex, channel-wise, relative 2D B 1 + $$ {\mathrm{B}}_1^{+} $$ -maps from a single gradient echo localizer to overcome long calibration times.

METHODS

126 channel-wise, complex, relative 2D B 1 + $$ {\mathrm{B}}_1^{+} $$ -maps of the human heart from 44 subjects were acquired at 7T using a Cartesian, cardiac gradient-echo sequence obtained under breath-hold to create a library for network training and cross-validation. The deep learning predicted maps were qualitatively compared to the ground truth. Phase-only B 1 + $$ {\mathrm{B}}_1^{+} $$ -shimming was subsequently performed on the estimated B 1 + $$ {\mathrm{B}}_1^{+} $$ -maps for a region of interest covering the heart. The proposed network was applied at 7T to 3 unseen test subjects.

RESULTS

The deep learning-based B 1 + $$ {\mathrm{B}}_1^{+} $$ -maps, derived in approximately 0.2 seconds, match the ground truth for the magnitude and phase. The static, phase-only pulse design performs best when maximizing the mean transmission efficiency. In-vivo application of the proposed network to unseen subjects demonstrates the feasibility of this approach: the network yields predicted B 1 + $$ {\mathrm{B}}_1^{+} $$ -maps comparable to the acquired ground truth and anatomical scans reflect the resulting B 1 + $$ {\mathrm{B}}_1^{+} $$ -pattern using the deep learning-based maps.

CONCLUSION

The feasibility of estimating 2D relative B 1 + $$ {\mathrm{B}}_1^{+} $$ -maps from initial localizer scans of the human heart at 7T using deep learning is successfully demonstrated. Because the technique requires only sub-seconds to derive channel-wise B 1 + $$ {\mathrm{B}}_1^{+} $$ -maps, it offers high potential for advancing clinical body imaging at ultra-high fields.

Brain oxygen extraction and neural tissue susceptibility are associated with cognitive impairment in older individuals

BACKGROUND AND PURPOSE

We investigated the effects of aging, white matter hyperintensities (WMH), and cognitive impairment on brain iron levels and cerebral oxygen metabolism, known to be altered in Alzheimer's disease (AD), using quantitative susceptibility mapping and MR-based cerebral oxygen extraction fraction (OEF).

METHODS

In 100 individuals over the age of 50 (68/32 cognitively impaired/intact), OEF and neural tissue susceptibility (χ_n ) were computed retrospectively from MRI multi-echo gradient echo data, obtained on a 3 Tesla MRI scanner. The effects of age and WMH on OEF and χ_n were assessed within groups, and OEF and χ_n were assessed between groups, using multivariate regression analyses.

RESULTS

Cognitively impaired subjects were found to have 19% higher OEF and 34% higher χ_n than cognitively intact subjects in the cortical gray matter and several frontal, temporal, and parietal regions (p < .05). Increased WMH burden was significantly associated with decreased OEF in the cognitively impaired, but not in the cognitively intact. Older age had a stronger association with decreased OEF in the cognitively intact group. Both older age and increased WMH burden were significantly associated with increased χ_n in temporoparietal regions in the cognitively impaired.

CONCLUSIONS

Higher brain OEF and χ_n in cognitively impaired older individuals may reflect altered oxygen metabolism and iron in areas with underlying AD pathology. Both age and WMH have associations with OEF and χ_n but are modified by the presence of cognitive impairment.