Limits...
MR Vascular Fingerprinting in Stroke and Brain Tumors Models

View Article: PubMed Central - PubMed

ABSTRACT

In this study, we evaluated an MRI fingerprinting approach (MRvF) designed to provide high-resolution parametric maps of the microvascular architecture (i.e., blood volume fraction, vessel diameter) and function (blood oxygenation) simultaneously. The method was tested in rats (n = 115), divided in 3 models: brain tumors (9 L, C6, F98), permanent stroke, and a control group of healthy animals. We showed that fingerprinting can robustly distinguish between healthy and pathological brain tissues with different behaviors in tumor and stroke models. In particular, fingerprinting revealed that C6 and F98 glioma models have similar signatures while 9 L present a distinct evolution. We also showed that it is possible to improve the results of MRvF and obtain supplemental information by changing the numerical representation of the vascular network. Finally, good agreement was found between MRvF and conventional MR approaches in healthy tissues and in the C6, F98, and permanent stroke models. For the 9 L glioma model, fingerprinting showed blood oxygenation measurements that contradict results obtained with a quantitative BOLD approach. In conclusion, MR vascular fingerprinting seems to be an efficient technique to study microvascular properties in vivo. Multiple technical improvements are feasible and might improve diagnosis and management of brain diseases.

No MeSH data available.


Related in: MedlinePlus

MR images of representative rats from tumor groups F98 (top) and 9 L (bottom).For each animal T2w and ADC images acquired are presented as well as parametric maps computed using the steady-state or the fingerprinting (with Dictionary C) approach. Healthy striatum and lesion ROIs are overlaid on the T2w images in blue and green, respectively. The color-coded parametric maps (BVf, VSI/radius, and StO2) are overlaid on the T2w images.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC5121626&req=5

f5: MR images of representative rats from tumor groups F98 (top) and 9 L (bottom).For each animal T2w and ADC images acquired are presented as well as parametric maps computed using the steady-state or the fingerprinting (with Dictionary C) approach. Healthy striatum and lesion ROIs are overlaid on the T2w images in blue and green, respectively. The color-coded parametric maps (BVf, VSI/radius, and StO2) are overlaid on the T2w images.

Mentions: We compared MRvF (with dictionary C) and steady-state approaches in Fig. 5. Visually, the F98 glioma has a higher ADC than in the contralateral striatum. In this tumor model, the steady-state approach shows an increase in VSI and a decrease in BVf and StO2 in the lesion compared to healthy tissues (radius = 12.0 ± 1.4 vs. 6.5 ± 0.5 μm; BVf = 2.8 ± 0.4 vs. 3.3 ± 0.4%; and StO2 = 54.8 ± 3.5 vs. 70.5 ± 4.5%; p < 0.05; Figs 4 and 5). These trends are also found with the fingerprinting approach except for BVf (no statistical difference was observed between the lesion and the healthy tissue). However, the oxygen values found with MRvF are globally higher than the steady-state estimates. These findings can also be seen in the graphs of Fig. 4 for the entire F98 group and for the C6 and stroke models. In the 9 L animal in Fig. 5, BVf and radius follow the same trend as the steady-state BVf and VSI. However, StO2 is higher in the tumor than in the contralateral striatum for the steady-state approach (81.1 ± 5.5 vs. 72.5 ± 6.9%, p < 0.05), while StO2 estimates from MRvF are lower in the tumor than in the contralateral striatum (62.4 ± 6.3 vs 83.6 ± 3.4%, p < 0.05). This can also be seen in the graphs of Fig. 4 for the entire 9 L group. A correlation analysis was also performed between MRvF (Dictionary C) and steady-state estimates on an animal level (healthy striatum and lesion ROIs). The results are presented in Supplementary Figure 2. Different colors and symbols are used to represent the different groups of animals. A high linear correlation coefficient (r2 = 0.95) was found between the BVf estimates. It has to be noted that the same analysis performed using Dictionary A provided the same correlation coefficient but a trendline with a slope close to one (y = 0.9x+0.2 (Dictionary A) vs y = 0.5x+1.4 (Dictionary C)). This can be understood by the fact that Dictionary C contains medium-to-large blood vessels and larger blood volume fractions, which are not included in the steady-state approach. The results for VSI and vessel radius estimates had lower correlation coefficient (R2 = 0.5) and a trendline different from unity. A poor correlation was found between the StO2 estimates (R2 = 0). However, one can clearly see that the results are heavily influenced by the estimates in the 9 L group (short red bars). When removing the data from the 9 L animals, a larger coefficient (R2 = 0.3) was found. In this case, the intercept in the trendline equation (y = 1.1x-27) suggests a bias in the estimates.


MR Vascular Fingerprinting in Stroke and Brain Tumors Models
MR images of representative rats from tumor groups F98 (top) and 9 L (bottom).For each animal T2w and ADC images acquired are presented as well as parametric maps computed using the steady-state or the fingerprinting (with Dictionary C) approach. Healthy striatum and lesion ROIs are overlaid on the T2w images in blue and green, respectively. The color-coded parametric maps (BVf, VSI/radius, and StO2) are overlaid on the T2w images.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC5121626&req=5

f5: MR images of representative rats from tumor groups F98 (top) and 9 L (bottom).For each animal T2w and ADC images acquired are presented as well as parametric maps computed using the steady-state or the fingerprinting (with Dictionary C) approach. Healthy striatum and lesion ROIs are overlaid on the T2w images in blue and green, respectively. The color-coded parametric maps (BVf, VSI/radius, and StO2) are overlaid on the T2w images.
Mentions: We compared MRvF (with dictionary C) and steady-state approaches in Fig. 5. Visually, the F98 glioma has a higher ADC than in the contralateral striatum. In this tumor model, the steady-state approach shows an increase in VSI and a decrease in BVf and StO2 in the lesion compared to healthy tissues (radius = 12.0 ± 1.4 vs. 6.5 ± 0.5 μm; BVf = 2.8 ± 0.4 vs. 3.3 ± 0.4%; and StO2 = 54.8 ± 3.5 vs. 70.5 ± 4.5%; p < 0.05; Figs 4 and 5). These trends are also found with the fingerprinting approach except for BVf (no statistical difference was observed between the lesion and the healthy tissue). However, the oxygen values found with MRvF are globally higher than the steady-state estimates. These findings can also be seen in the graphs of Fig. 4 for the entire F98 group and for the C6 and stroke models. In the 9 L animal in Fig. 5, BVf and radius follow the same trend as the steady-state BVf and VSI. However, StO2 is higher in the tumor than in the contralateral striatum for the steady-state approach (81.1 ± 5.5 vs. 72.5 ± 6.9%, p < 0.05), while StO2 estimates from MRvF are lower in the tumor than in the contralateral striatum (62.4 ± 6.3 vs 83.6 ± 3.4%, p < 0.05). This can also be seen in the graphs of Fig. 4 for the entire 9 L group. A correlation analysis was also performed between MRvF (Dictionary C) and steady-state estimates on an animal level (healthy striatum and lesion ROIs). The results are presented in Supplementary Figure 2. Different colors and symbols are used to represent the different groups of animals. A high linear correlation coefficient (r2 = 0.95) was found between the BVf estimates. It has to be noted that the same analysis performed using Dictionary A provided the same correlation coefficient but a trendline with a slope close to one (y = 0.9x+0.2 (Dictionary A) vs y = 0.5x+1.4 (Dictionary C)). This can be understood by the fact that Dictionary C contains medium-to-large blood vessels and larger blood volume fractions, which are not included in the steady-state approach. The results for VSI and vessel radius estimates had lower correlation coefficient (R2 = 0.5) and a trendline different from unity. A poor correlation was found between the StO2 estimates (R2 = 0). However, one can clearly see that the results are heavily influenced by the estimates in the 9 L group (short red bars). When removing the data from the 9 L animals, a larger coefficient (R2 = 0.3) was found. In this case, the intercept in the trendline equation (y = 1.1x-27) suggests a bias in the estimates.

View Article: PubMed Central - PubMed

ABSTRACT

In this study, we evaluated an MRI fingerprinting approach (MRvF) designed to provide high-resolution parametric maps of the microvascular architecture (i.e., blood volume fraction, vessel diameter) and function (blood oxygenation) simultaneously. The method was tested in rats (n&thinsp;=&thinsp;115), divided in 3 models: brain tumors (9&thinsp;L, C6, F98), permanent stroke, and a control group of healthy animals. We showed that fingerprinting can robustly distinguish between healthy and pathological brain tissues with different behaviors in tumor and stroke models. In particular, fingerprinting revealed that C6 and F98 glioma models have similar signatures while 9&thinsp;L present a distinct evolution. We also showed that it is possible to improve the results of MRvF and obtain supplemental information by changing the numerical representation of the vascular network. Finally, good agreement was found between MRvF and conventional MR approaches in healthy tissues and in the C6, F98, and permanent stroke models. For the 9&thinsp;L glioma model, fingerprinting showed blood oxygenation measurements that contradict results obtained with a quantitative BOLD approach. In conclusion, MR vascular fingerprinting seems to be an efficient technique to study microvascular properties in vivo. Multiple technical improvements are feasible and might improve diagnosis and management of brain diseases.

No MeSH data available.


Related in: MedlinePlus