Limits...
The effects of voxel localization and time of echo on the diagnostic accuracy of cystic brain tumors in 3 tesla magnetic resonance spectroscopy.

Rezvanizadeh A, Firouznia K, Salehi-Sadaghiani M, Mohseni M, Gharaei D, Ghanaati H, Saligheh Rad H, Masoudnia M - Iran J Radiol (2012)

Bottom Line: There was no statistically significant difference among the compared TEs.The percentages of ratios above the cut-off point at all TEs were more in the rim compared to the center and in the union of both compared to the rim or center.If not accessible, the use of the union of peripheral and central voxels enhances the sensitivity when compared to usage of peripheral or central voxels solely.

View Article: PubMed Central - PubMed

Affiliation: Department of Radiology, Faculty of Medicine, Tehran University of Medical Sciences, Tehran, Iran.

ABSTRACT

Background: Although magnetic resonance spectroscopy (MRS) has been shown as an effective diagnostic tool in distinguishing inflammation from neoplasm in cystic brain lesions, the optimum approach in selecting the portions of lesions in MRS and the possible effects of different times of echoes (TEs) remains unknown.

Objectives: To determine the most effective TE in diagnosing neoplastic lesions based on detecting choline (Cho), N acetyl aspartate (NAA) and creatinine (Cr). Moreover, the role of voxel localization on the diagnosis of the neoplastic nature of the lesions is assessed through comparing the abovementioned metabolite ratios in the rim and center of each lesion with the same TE.

Patients and methods: In 16 patients with brain cystic tumors, MRS was performed at TEs of 30, 135 and 270 ms for detection of Cho, NAA and Cr metabolites using a 3 tesla MRI unit. The percentage of analyzed ratios greater than a cut-off point of 1.3 for Cho/Cr and 1.6 for Cho/NAA were calculated.

Results: Cho/Cr and Cho/NAA ratio means at all TEs were more at the central area in comparison with the periphery, although none of the differences were statistically significant. There was no statistically significant difference among the compared TEs. The percentages of ratios above the cut-off point at all TEs were more in the rim compared to the center and in the union of both compared to the rim or center. All the patients had at least one voxel with a Cho/Cr ratio of more than 1.3 when the voxel was chosen according to the hotspots shown in the chemical shift imaging map, regardless of their location at all examined TEs.

Conclusions: Selection of voxels with the guide of chemical shift imaging map yields to 100% diagnostic sensitivity. If not accessible, the use of the union of peripheral and central voxels enhances the sensitivity when compared to usage of peripheral or central voxels solely.

No MeSH data available.


Related in: MedlinePlus

MR spectroscopy of parietooccipital brain lesion on the left hemisphere is depicted along with the hotspot localization. With the guidance of chemical shift imaging map hotspot a more reliable data is gathered.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3569551&req=5

fig937: MR spectroscopy of parietooccipital brain lesion on the left hemisphere is depicted along with the hotspot localization. With the guidance of chemical shift imaging map hotspot a more reliable data is gathered.

Mentions: The effect of voxel localization was also tested in another way. In practice, Cho/Cr and Cho/NAA values above or below an assumptive threshold can be an acceptable indicator of the neoplastic or nonneoplastic nature of a lesion. However, there is no global consensus on a unique threshold and the sensitivity and specificity can vary with the utilized threshold amount. According to previous studies and acceptable reported sensitivity of 1.3 for Cho/Cr and 1.6 for Cho/NAA as a threshold (17, 22), we used the abovementioned thresholds as the diagnostic cut-off point in the present study. On TE = 135, 25% of all the patients showed ratios less than 1.3 in the central regions. While for the peripheral voxels, this number decreased to 12.5%. The union of both the periphery and the center resulted in higher percentages of ratios above the cut-off point (93.75%). Similarly, the percentages of ratios among other TEs, which were above the threshold in the center, rim and union of both locations, showed an increasing manner, although the differences were not significant, except for Cho/Cr in the center (62.5%) and union of the center and the rim (93.75%) at TE = 30. The smaller mean of metabolite ratio values in combination with the greater fraction of ratio values above the threshold in the periphery in comparison with the center, indicates the heterogeneous nature of central voxels. In other words, some of the central voxels have greater metabolite ratio values relative to peripheral voxels (the ones much more than the threshold) that result in a greater mean, while there are many other central voxels with ratio values under the defined threshold. On the other hand, the peripheral voxels show a smaller ratio mean, with possible aforementioned reasons, but with a greater fraction above the defined threshold. In essence, it can be interpreted that the peripheral values show a more homogeneous pattern. We also found that selection of voxels with the guide of chemical shift imaging map hotspots (Figure 4) results in 100% sensitivity in all TEs. Hence, it is wise to select the voxels according to the chemical shift imaging map, instead of restriction to the rim or cystic parts. If not accessible, the best approach not to miss any high Cho/Cr or Cho/NAA ratio is selection of both the rim and center. Furthermore, peripheral voxels showed increased computed ratios in comparison with the central ones, when they are solely assessed, although the observed difference was not statistically significant.


The effects of voxel localization and time of echo on the diagnostic accuracy of cystic brain tumors in 3 tesla magnetic resonance spectroscopy.

Rezvanizadeh A, Firouznia K, Salehi-Sadaghiani M, Mohseni M, Gharaei D, Ghanaati H, Saligheh Rad H, Masoudnia M - Iran J Radiol (2012)

MR spectroscopy of parietooccipital brain lesion on the left hemisphere is depicted along with the hotspot localization. With the guidance of chemical shift imaging map hotspot a more reliable data is gathered.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3569551&req=5

fig937: MR spectroscopy of parietooccipital brain lesion on the left hemisphere is depicted along with the hotspot localization. With the guidance of chemical shift imaging map hotspot a more reliable data is gathered.
Mentions: The effect of voxel localization was also tested in another way. In practice, Cho/Cr and Cho/NAA values above or below an assumptive threshold can be an acceptable indicator of the neoplastic or nonneoplastic nature of a lesion. However, there is no global consensus on a unique threshold and the sensitivity and specificity can vary with the utilized threshold amount. According to previous studies and acceptable reported sensitivity of 1.3 for Cho/Cr and 1.6 for Cho/NAA as a threshold (17, 22), we used the abovementioned thresholds as the diagnostic cut-off point in the present study. On TE = 135, 25% of all the patients showed ratios less than 1.3 in the central regions. While for the peripheral voxels, this number decreased to 12.5%. The union of both the periphery and the center resulted in higher percentages of ratios above the cut-off point (93.75%). Similarly, the percentages of ratios among other TEs, which were above the threshold in the center, rim and union of both locations, showed an increasing manner, although the differences were not significant, except for Cho/Cr in the center (62.5%) and union of the center and the rim (93.75%) at TE = 30. The smaller mean of metabolite ratio values in combination with the greater fraction of ratio values above the threshold in the periphery in comparison with the center, indicates the heterogeneous nature of central voxels. In other words, some of the central voxels have greater metabolite ratio values relative to peripheral voxels (the ones much more than the threshold) that result in a greater mean, while there are many other central voxels with ratio values under the defined threshold. On the other hand, the peripheral voxels show a smaller ratio mean, with possible aforementioned reasons, but with a greater fraction above the defined threshold. In essence, it can be interpreted that the peripheral values show a more homogeneous pattern. We also found that selection of voxels with the guide of chemical shift imaging map hotspots (Figure 4) results in 100% sensitivity in all TEs. Hence, it is wise to select the voxels according to the chemical shift imaging map, instead of restriction to the rim or cystic parts. If not accessible, the best approach not to miss any high Cho/Cr or Cho/NAA ratio is selection of both the rim and center. Furthermore, peripheral voxels showed increased computed ratios in comparison with the central ones, when they are solely assessed, although the observed difference was not statistically significant.

Bottom Line: There was no statistically significant difference among the compared TEs.The percentages of ratios above the cut-off point at all TEs were more in the rim compared to the center and in the union of both compared to the rim or center.If not accessible, the use of the union of peripheral and central voxels enhances the sensitivity when compared to usage of peripheral or central voxels solely.

View Article: PubMed Central - PubMed

Affiliation: Department of Radiology, Faculty of Medicine, Tehran University of Medical Sciences, Tehran, Iran.

ABSTRACT

Background: Although magnetic resonance spectroscopy (MRS) has been shown as an effective diagnostic tool in distinguishing inflammation from neoplasm in cystic brain lesions, the optimum approach in selecting the portions of lesions in MRS and the possible effects of different times of echoes (TEs) remains unknown.

Objectives: To determine the most effective TE in diagnosing neoplastic lesions based on detecting choline (Cho), N acetyl aspartate (NAA) and creatinine (Cr). Moreover, the role of voxel localization on the diagnosis of the neoplastic nature of the lesions is assessed through comparing the abovementioned metabolite ratios in the rim and center of each lesion with the same TE.

Patients and methods: In 16 patients with brain cystic tumors, MRS was performed at TEs of 30, 135 and 270 ms for detection of Cho, NAA and Cr metabolites using a 3 tesla MRI unit. The percentage of analyzed ratios greater than a cut-off point of 1.3 for Cho/Cr and 1.6 for Cho/NAA were calculated.

Results: Cho/Cr and Cho/NAA ratio means at all TEs were more at the central area in comparison with the periphery, although none of the differences were statistically significant. There was no statistically significant difference among the compared TEs. The percentages of ratios above the cut-off point at all TEs were more in the rim compared to the center and in the union of both compared to the rim or center. All the patients had at least one voxel with a Cho/Cr ratio of more than 1.3 when the voxel was chosen according to the hotspots shown in the chemical shift imaging map, regardless of their location at all examined TEs.

Conclusions: Selection of voxels with the guide of chemical shift imaging map yields to 100% diagnostic sensitivity. If not accessible, the use of the union of peripheral and central voxels enhances the sensitivity when compared to usage of peripheral or central voxels solely.

No MeSH data available.


Related in: MedlinePlus