Intravoxel incoherent motion (IVIM) for distinguishing glioma and solitary brain metastasis: A pilot multi-parameter magnetic resonance imaging histogram and texture analysis

doi:10.21203/rs.3.rs-2376391/v1

Download PDF

Research Article

Intravoxel incoherent motion (IVIM) for distinguishing glioma and solitary brain metastasis: A pilot multi-parameter magnetic resonance imaging histogram and texture analysis

https://doi.org/10.21203/rs.3.rs-2376391/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background: To investigate the value of diffusion-weighted imaging (DWI), intravoxel incoherent motion (IVIM), and amide proton transfer-weighted (APTW) imaging in differentiating the primary glioma from solitary brain metastases (SBM) with the histogram and texture features.

Methods: A total of 65 patients with newly diagnosed glioma or metastases were enrolled. All patients underwent DWI, IVIM, and APTW imaging, as well as the T1W, T2W, T2FLAIR, and enhanced T1W imaging. The first-order texture and histogram features of apparent diffusion coefficient (ADC) from DWI, diffusion coefficient (D), perfusion fraction (f), pseudo-diffusion coefficient (D*) from IVIM, and MTRasym (3.5 ppm) from APTW were extracted from the tumor parenchyma and compared between glioma and SBM. Parameters with significant differences were analyzed with the logistics regression and receiver operator curves to explore the optimal model and compare the differentiation performance.

Results: Higher ADC_kurtosis (p=0.02), f₉₀ (p＜0.01), and f_kurtosis (p=0.01) were found for glioma, while higher MTRasym (3.5ppm)₁₀ (p= 0.05), f_entropy (p＜0.01), and f_mean (p＜0.01) were observed for SBM. f_kurtosis (OR =0.66, 95%CI 0.48-0.92, p=0.02), and f_mean (OR=1.44, 95%CI 1.16-1.18, p<0.01) were independent factors for SBM differentiation. The model combining age, f_kurtosis, and f_mean showed an AUC of 0.83, while the model combined with f_kurtosis and f_meanwith improved sensitivity had an AUC of 0.81.

Conclusion: The f_meanand f_kurtosis in enhanced tumor region could be used to differentiate glioma and SBM.

Intravoxel incoherent motion

glioma

solitary brain metastasis

MRI

Glioma and solitary brain metastasis (SBM) are neoplastic diseases with high morbidity and mortality worldwide [1, 2]. Different treatment strategies and clinical management are used for these two types of tumors. Maximum-safe-resection followed by chemoradiotherapy is recommended for glioma to reduce tumor size and acquire tumor tissue for identifying the grade [3]. For SBM, surgery is considered only when the number of lesions is less than 4 or there are high-risk or life-threatening clinical symptoms, such as intracranial hypertension, tumor apoplexy, increased edema, obstructive hydrocephalus, and so on. Moreover, brain radiotherapy is also recommended to improve the patients’ life quality [4]. Therefore, accurate differentiation of glioma and SBM is of crucial importance. Yet, clinically, patients with glioma and SBM may present with similar symptoms, including secondary epilepsy, dysfunction, and intracranial hypertension; thus, differentiation remains challenging.

Magnetic resonance imaging (MRI) is commonly used to diagnose and differentiate brain neoplasms. MRI provides excellent tissue contrast. Conventional MRIs such as T1, T2 weighted (T1W, T2W), and enhanced T1W imaging (T1WI + C) can identify well structural abnormalities such as mass size, shape and location, edema, ring-enhancing, and necrosis, as well as the degree of blood-brain barrier damage and so on. Yet, the conventional MRI approach may not be accurate enough when differentiating SBM from glioma, especially high-grade glioma (HGG), which presents as well-defined spacing occupying lesions with an enhancing rim in T1WI + C accompanied by hyperintensity peritumoral edema in T2WI [5]. Moreover, the morphological analysis suggested that the volume of tumor parenchyma or peritumoral edema, midline shift, and rim pattern in different conventional weighted MR sequences are not optimal approaches for differentiating glioma from SBM [6, 7].

Over the years, advanced MRI technologies such as diffusion-weighted imaging (DWI), intravoxel incoherent motion (IVIM), and amide proton transfer-weighted (APTW) imaging have been developed and validated, and studies have shown that features extracted from those images can differentiate glioma from SBM [8–10]. DWI applies the diffusion gradient to characterize the mobility of water molecules in tissue when there is an inhibitory effect on cell membranes [11]. Intravoxel incoherent motion (IVIM) is a multi-b value DWI imaging that simultaneously measures the perfusion-associated microcirculation of blood capillaries at low-b values (b < 200 s/mm²) and the molecular water diffusion at high-b values. [12]. Amide proton transfer-weighted (APTW) imaging, as a relatively new noninvasive and endogenous contrast molecular imaging, utilizes the effect of chemical exchange saturation transfer (CEST) between amide protons and water protons to measure the concentration of mobile proteins/peptides and tissue pH [13]. Various studies have shown that DWI, IVIM, APTW, or the combination of quantitative parameters have a higher diagnostic value than conventional MRI when assessing malignant tissue; they can quantify tumor regions and preform differentiation of different tumor types or even different tumor types or subytpes [14–17]. Similarly, texture and histogram analysis based on MRI imaging, which could extract the characteristics based on the distribution of quantitative tumor data, were also investigated and showed potential in glioma and SBM differentiation [18].

However, thus far, most studies focused on the analysis of peritumoral or edema region for glioma and SBM [15, 17, 19] with histogram and texture analysis dedicated to the structural weighted MRI image [20, 21] or quantile of functional image [22, 23]. Yet, it remains unclear whether the texture and histogram features of the enhanced tumor area in advanced MRI imaging can be helpful when assessing this type of tumor. Therefore, in this study, we used DWI, IVIM, and APTW imaging to investigate the differentiation of glioma and SBM with the mean parametric quantification and the histogram and first-order texture features.

Participants

The study was approved by the Ethics Committee of Shanxi Provincial People's Hospital (Fifth Hospital of Shanxi Medical University). A total of 83 patients with brain lesions were enrolled from December 2020 to May 2022. Inclusion criteria were: (1) patients with a definite diagnosis; (2) patients eligible for MRI, and who underwent MRI imaging preoperatively. Those who underwent chemotherapy or radiation therapy were excluded. All gliomas were diagnosed by surgical pathology (n = 54) according to the WHO 2016 Classification [24]; the diagnosis of SBM was obtained from histology (n = 29), imaging follow-up of malignant tumor metastasis (n = 2), or tumor markers from laboratory tests (n = 4) [25]. In addition, 18 patients were excluded prior to the analysis due to the following reasons: incomplete imaging data (n = 10) and/or unsatisfied image quality (i.e., significant cystic, hemorrhagic, or massive tumor necrosis; n = 8). The remaining 65 patients were included in the study, and the study flowchart is shown in Fig. 1.

Data Acquisition

MR sequences including T1WI, T2WI, T2 fluid-attenuated inversion recovery (FLAIR), IVIM, DWI, APT weighted imaging (APTWI), and enhanced T1WI were conducted in a 3.0T MR scanner (Discovery MR 750W, GE Healthcare, Waukesha, WI, USA) with a 24-channel head neck unit coil. Axial IVIM used 12 b-values (0, 20, 40, 80, 110, 150, 200, 400, 800, 1200, 1500, and 2000s/mm²). The following parameters were applied: repetition time/echo time (TR/TE) = 5400ms/90ms, the field of view (FOV) = 220×220mm, matrix size = 110×110, slice thickness = 4.0mm, intersection gap = 1.0mm. The parameters for axial DWI with b-values of 0 and 1000s/mm² were: TR/TE = 6000ms/76ms, FOV = 240×240mm, matrix size = 120×120, slice thickness = 4.5mm, intersection gap = 1.0mm. Diffusion gradients were applied in three orthogonal directions for both IVIM and DWI, and the scanning slices covered the area of the whole lesion. The APTWI was acquired using a 2D single-shot fast spin-echo-based sequence (TR/TE = 2950ms/27ms, FOV = 256mm×256mm, matrix size = 120×120, slice thickness = 8mm) with phased cycle pulses for saturation and the water saturation sift reference (WASSR) for B₀ correction. The total duration for phase cycle pulses is 2000ms under B1 of 2\(\mu\)T. The Z-spectra includes 52 frequencies, 49 of which offset from 600 to -600 Hz at an interval of 25 Hz, and three unsaturated images at 5000 Hz for signal normalization. The maximum slice without or with a minimum area of hemorrhage and cyst in axial T2FLAIR was selected for ATPWI data acquisition.

Imaging Processing

The DWI, IVIM, and APTWI images were imported to the iQuant software (GE Healthcare, Beijing, China) to calculate the parametric maps. The DWI apparent diffusion coefficient (DWI-ADC) was generated with the formula:

S_b/S₀ = exp (-b × ADC) (Eq. 1)

where S_b and S₀ were the signal intensities for b = 1000 s/mm² and b = 0 s/mm^2,. The IVIM data were processed with the bi-exponential model as defined by the equation:

S_b/S₀ = f × exp (-b × D^*) + (1-f) × exp (-b × D) (Eq. 2)

where S_b is the signal intensity under b > 0 s/mm², D* is the pseudo-diffusion coefficient, D is the true-diffusion coefficient, and f represents the perfusion fraction. The APTWI asymmetric magnetization transfer ratio at 3.5 ppm is represented as MTRasym (3.5 ppm) and calculated by:

MTRasym (3.5 ppm) = [S_sat (− 3.5 ppm) − S_sat (+ 3.5 ppm)]/S₀ (Eq. 3)

where the S₀ is non-saturation intensity while S_sat is the signal intensity after saturation.

Y.H (a neuro-radiologist with 30 years of work experience) delineated the region of interest (ROI) on tumor parenchyma with the enhanced T1W image as the reference where the cerebrospinal fluid-filled, calcification, hemorrhagic, necrotic, and cystic areas were avoided wherever possible in 3D-Slicer (https://www.slicer.org/, version 4.10). For IVIM and DWI, ROIs were drawn slice by slide, and a 3-dimension ROI was eventually obtained, while for APTWI, a 2-dimension ROI was acquired since only one slice of imaging was acquired. The first-order and histogram features, including mean, 10th percentile, 90th percentile, entropy, kurtosis, and skewness for all parametric maps in tumor ROIs were extracted with the pyradiomics plugin (https://pyradiomics.readthedocs.io/).

Statistical analysis

Quantitative variables were expressed as the mean ± standard deviation and were compared with the Student's t-test or the Wilcoxon test (Mann-Whitney U test) after the normality and homogeneity of variance were confirmed. The binary data for clinical information were compared using the Chi-Squared test. Next, the variables with significant differences were assessed using univariate and multivariate logistics regression. The receiver operator curves (ROC) analysis was applied to explore the performance of independent factors and multi-parameter combined models for differentiation of glioma and SBM. The area under the ROC curve (AUC), sensitivity (sen.), specificity (spe.), positive predictive value (PPV), and negative predictive value (NPV) were calculated and compared. Nomogram and bootstrap resampling methods were used for the evaluation of the multivariate logistics regression. R package (version 4.0.0) was used for all the statistics. A P value < 0.05 indicated a statistically significant difference.

Clinical characteristics

A number of 65 patients (44 glioma cases with age 51.27 ± 13.09 and 21 SBM cases with age 59.05 ± 12.33) were finally included in this study. Demographics and clinical manifestation are summarized in Table 1. The age in the SBM group was significantly greater than that of the glioma group, while no significant difference was found for gender and clinical manifestations (secondary epilepsy, intracranial hypertension, and dysfunction).

Table 1

Clinical characteristics
	Glioma group (n = 44)	SBM group (n = 21)	P
Demographics
Male (n, %)	25 (56.8%)	10 (47.6%)	0.609
Age at diagnosis (year)	51.27 ± 13.09	59.05 ± 12.33	0.019*
Clinical manifestations
Secondary epilepsy (n, %)	18 (40.9%)	4 (19.5%)	0.119
Intra-cranial hypertension (n, %)	22 (50.0%)	11 (52.4%)	1
Dysfunction (n, %)	18 (40.9%)	14 (66.7%)	0.062

* P < 0.05

Comparison of first-order and histogram features between glioma and SBM

Differences for ADC_kurtosis (6.43±5.63 vs. 4.54±3.23, p= 0.02), APTWI--MTRasym (3.5ppm)₁₀ (0.53±0.97 vs. 1.00±0.75, p= 0.05), f₉₀ (0.23±0.05 vs. 0.29±0.06, p＜0.01), f_entropy (0.56±0.34 vs. 0.85±0.33, p＜0.01), f_kurtosis (6.76±4.08 vs. 3.91±1.43, p＜0.01), and f_mean (0.14±0.02 vs. 0.17±0.04, p＜0.01) were observed between glioma and SBM, while other quantitative values showed no significant differences (Table 2).

Table 2

Comparison of parametric values in the region of interest between glioma and SBM
	10^th percentile	90^th percentile	entropy	kurtosis	mean	skewness
ADC (×10^− 4mm²/s)
Glioma	8.47 ± 1.55	15.14 ± 3.35	5.23 ± 0.52	6.43 ± 5.63	11.65 ± 2.23	0.98 ± 0.93
SBM	7.96 ± 1.87	15.79 ± 2.83	5.37 ± 0.50	4.54 ± 3.23	11.48 ± 2.12	0.87 ± 0.74
P	0.252	0.453	0.164	0.021*	0.783	0.563
MTRasym (3.5 ppm)
Glioma	0.53 ± 0.97	2.23 ± 1.31	0.58 ± 0.51	3.08 ± 1.13	1.40 ± 0.98	-0.10 ± 0.42
SBM	1.00 ± 0.75	2.39 ± 0.85	0.58 ± 0.47	3.39 ± 1.14	1.66 ± 0.70	0.20 ± 0.62
P	0.044*	0.621	0.724	0.237	0.464	0.056
IVIM-D* (×10^− 4mm²/s)
Glioma	18.84 ± 6.24	31.77 ± 2.49	0.74 ± 0.25	8.10 ± 4.09	26.30 ± 2.60	-1.65 ± 0.66
SBM	19.24 ± 8.19	32.95 ± 3.97	0.67 ± 0.32	7.16 ± 3.88	26.42 ± 4.66	-1.34 ± 0.91
P	0.513	0.232	0.621	0.439	0.892	0.137
IVIM-f
Glioma	0.05 ± 0.03	0.23 ± 0.05	0.56 ± 0.34	6.76 ± 4.08	0.14 ± 0.02	0.00 ± 0.00
SBM	0.07 ± 0.05	0.29 ± 0.06	0.85 ± 0.33	3.91 ± 1.43	0.17 ± 0.04	0.00 ± 0.00
P	0.262	< 0.001*	0.002*	0.002*	< 0.001*	0.084
IVIM-D (×10^− 4mm²/s)
Glioma	7.00 ± 1.40	12.02 ± 2.40	4.87 ± 0.45	5.67 ± 4.32	9.45 ± 1.87	0.86 ± 0.81
SBM	6.46 ± 1.43	11.75 ± 2.53	4.80 ± 0.59	4.34 ± 1.93	8.90 ± 1.75	0.85 ± 0.66
P	0.160	0.690	0.972	0.305	0.285	0.545

* P < 0.05

Representative images and the histogram of IVIM fraction in lesion ROI of one patient with glioma and another with SBM are shown in Fig. 2.

The diagnostic performance with clinical, first-order, and histogram features

The regression analysis showed that age (OR = 1.05, 95%CI 1.00-1.10, p = 0.04), f_kurtosis (OR = 0.66, 95% CI 0.48–0.92, p = 0.02), f_mean (OR = 1.44, 95% CI 1.16–1.18, p < 0.01), f_entropy (OR = 11.01, 95% CI 2.09–57.99, p = 0.01) and f₉₀ (OR = 1.21, 95% CI 1.08–1.35, p < 0.01) were the associated factors for glioma and SBM differentiation, but only the f_mean was an independent factor (Table 3, Table S1).

Table 3

Univariate and multivariate regression analysis
	Univariate			Multivariate-1^†			Multivariate-2^‡
	OR	(95% CI)	P	OR	(95% CI)	P	OR	(95% CI)	P
age	1.05	(1.00-1.10)	0.035*	1.04	(0.99–1.11)	0.095
f_kurtosis	0.66	(0.48–0.92)	0.015*	0.75	(0.53–1.06)	0.106	0.76	(0.55–1.04)	0.089
f_mean	1.44	(1.16–1.18)	0.001*	1.33	(1.07–1.67)	0.012*	1.34	(1.08–1.68)	0.010*
f_entropy	11.01	(2.09–57.99)	0.005*
f₉₀	1.21	(1.08–1.35)	0.001*
ADC_kurtosis	0.89	(0.74–1.06)	0.193
MTRasym(3.5ppm) ₁₀	1.89	(0.95–3.75)	0.068
^† Multivariate-1(Multivariate regression model 1): age + f_kurtosis + f_mean
^‡Multivariate-2(Multivariate regression model 2): f_kurtosis + f_mean

The optimal model, defined by multivariate logistic regression for differentiation, includes age, f_kurtosis, and f_mean. Better specificity of f_mean (0.96) and better sensitivity of f_kurtosis (0.85) were found by ROC analysis compared with other associated factors (Table 4, Fig. 3A). Also, f_mean and f_kurtosis demonstrated the best AUC (0.73 and 0.78, respectively). There was improved diagnostic performance when combining the age, f_kurtosis, and f_mean, and the AUC of the model was 0.83 (accuracy = 0.82, sen. = 0.57, spe. = 0.93, NPV = 0.82, PPV = 0.8). The model integrating the f_kurtosis and f_mean was also evaluated, and the AUC was 0.81 (accuracy = 0.8, sen. = 0.62, spe. = 0.89, NPV = 0.83, PPV = 0.72) (Table 4, Fig. 3B).

Table 4

The univariate and multivariate performance in the differentiation of glioma and metastases^†
	AUC (95% CI)	Sensitivity (95% CI)	Specificity (95% CI)	Accuracy (TN + TP)/(N + P)	PPV (P/TP)	NPV (N/TN)	Threshold
Univariate
MTRasym (3.5 ppm) ₁₀	0.66 (0.51–0.80)	0.62 (0.38–0.80)	0.68 (0.55–0.82)	0.66 (44)/(65)	0.48 (13/27)	0.79 (30/38)	0.91
ADC_kurtosis	0.68 (0.53–0.82)	0.71 (0.52–0.90)	0.66 (0.52–0.79)	0.68 (44)/(65)	0.50 (15/30)	0.83 (29/35)	4.02
f_kurtosis	0.73 (0.61–0.85)	0.81 (0.61–0.95)	0.59 (0.45–0.73)	0.66 (43)/(65)	0.49 (18/25)	0.87 (26/30)	4.6
f_mean	0.78 (0.63–0.92)	0.62 (0.43–0.81)	0.96 (0.88-1.00)	0.85 (55)/(65)	0.87 (13/15)	0.84 (42/50)	0.17
age	0.68	0.52	0.80	0.71	0.55	0.78	61.50
Multivariate^†
Multivariate-1^‡	0.83 (0.73–0.93)	0.57 (0.38–0.76)	0.93 (0.86-1.00)	0.82 (53)/(65)	0.80 (12/15)	0.82 (41/50)	0.55
Multivariate-2^§	0.81 (0.70–0.92)	0.62 (0.43–0.81)	0.89 (0.80–0.98)	0.80 (52)/(65)	0.72 (13/18)	0.83 (39/47)	0.49
^†DeLong‘s test showed no statistically significant differences between AUCs from two multivariate regression models. (Z = 0.553, p = 0.58)
^‡ Multivariate-1(Multivariate regression model 1): age + f_kurtosis + f_mean
^§Multivariate-2 (Multivariate regression model 2): f_kurtosis + f_mean

AUC: area under the curve; PPV: Positive predictive value; NPV: negative predictive value; P: positives; N: negatives; TP: true positives; TN: true negatives.

No significant differences were found between these two combined models by the Delong test (p = 0.58). Nomogram and calibration analysis showed a good correlation between apparent and bias-corrected multivariate logistic regression in both models (Figure S1).

Histogram and texture analysis of parameters acquired from DWI, IVIM, and APTWI were applied to differentiate glioma and SBM. Significantly lower mean and entropy of IVIM-f in the parenchyma area and higher kurtosis of ADC and IVIM-f were found in glioma compared to SBM. Furthermore, two combined models with IVIM-f mean and IVIM-f kurtosis were comparable and superior to every parameter.

ADC was well correlated with cell intensity and proliferative capacity [26, 27]. However, the same level of ADC percentile and mean value between glioma and SBM found in this study suggested that distinguishing glioma from SBM based on cell density and proliferation is sensitive enough. Yet, glioma had higher ADC kurtosis, which implied the diffusion coefficient distribution deviated more from the Gaussian distribution and greater diffusion coefficients in the tumor were around the mean values, indicating that the structure of the tumor cells hinders the diffusion of water molecules in glioma. In other words, there may be more tumor areas with higher cell heterogeneity in glioma. Since this is the first study that characterized ADC kurtosis in the enhanced regions of glioma and SBM, we searched for diffusion studies with more complex models for tissue heterogeneity. Gao et al. [23] showed that maximum fractional anisotropy (FA) was significantly higher in glioblastoma than in SBM in the enhanced tumor area. In addition, Mao et al. found significantly lower isotropic volume fraction (V_iso) and orientation dispersion index (ODI) from neurite orientation dispersion and density imaging (NODDI) of the contrast-enhancing tumors in glioma tissue in comparison with SBM [28]. These results suggest that glioma might have a more complex microstructure. Consequently, more water molecule activity is restricted, which leads to higher ADC kurtosis in glioma.

IVIM perfusion fraction is the ratio of micro-vascular diffusion to total micro-vascular and molecular water diffusion, both of which were associated with micro-structure [29]. Lower f_mean value in glioma may reflect the relatively less microvascular composition due to higher microstructural complexity. Smaller f_kurtosis value and greater f_entropy of IVIM-f in SBM were consistent with the representative histogram in Fig. 1. From the ADC kurtosis analysis, it could be inferred that the distribution of IVIM-f was more concentrated in glioma according to the more concentrated cell density and correspondingly smaller entropy. The significant difference in IVIM-f entropy but not in ADC entropy means the heterogeneity of relative change in microvascular perfusion and water diffusion may be a more sensitive feature to differentiate glioma from SBM. Furthermore, Shim et al. [14] investigated the differentiation of glioblastoma and brain metastasis with IVIM and found no significant difference in any of quantitative parameters between the two kinds of lesions. One reason for the difference may result from the adopted b values [30, 31], which ranged from 0 s/mm² to 900 s/mm² in [14] vs. 0 s/mm² to 2000 s/mm² in the present study. On the other side, small-size ROIs with a radius 2mm in five enhanced areas were used while we delineated all the enhanced areas.

MTRasym (3.5 ppm) from APTW was used to quantify the percent signal generated from mobile proteins and peptides [32]. A two-dimensional APTWI study showed that the mean, 10th, 25th, 50th, 75th, and 90th percentile of MTRasym (3.5 ppm) in enhancing areas of glioblastomas were significantly higher than those of SBM, and the 10th percentile was used to obtain the optimal AUC (0.85) for glioma and SBM differentiation [22]. Another study claimed that the maximum, minimum, and mean of MTRasym (3.5 ppm) in the tumor core did not significantly differ between the SBM and GBM [19]. In our study, the 10th percentile was significantly lower in glioma than in SBM, but the differentiation performance was relatively low (AUC = 0.66, sensitivity = 0.62, and specificity = 0.68). Also, the high-grade glioma with higher enrolled MTRasym (3.5ppm)[33–35] and ROI definition, which was consistent with previous studies [19, 22], could explain these differences.

In multivariate regression analysis, models with f_kurtosis and f_mean provided the optimal differentiation performance. As a related factor pointing to a higher risk of metastasis, age did not affect the abilities of combined models. Though the AUC (0.83, 0.81) is less than the quantification with ADC of high b value [17] and APTWI [19] in edema and peritumoral region, it was comparable or better than the result in [22] with APT percentiles and in [28] with multiple advanced diffusion models in enhanced tumor core area. In a further study, we plan to analyze both parts of areas as was done in some existing literature, hoping this could provide additional or complementary information to improve the identification of glioma and SBM.

There are some limitations in the present study. First, the number of enrolled subjects is relatively small, and only part of the subtypes was analyzed, i. e., the brain metastasis resulted mainly from the lung lesions, and the glioma was also not graded. Second, different ROI delineation methods may interfere with the results. We delineated tumor regions at multiple layers on the parametric maps of ADC and IVIM for analysis but only delineated the tumor region of the largest tumor parenchyma layer on 2-dimensional APTW imaging. Mismatches in tumor regions may impact the analysis results, and our ROI was drawn by one reader only, so there is a lack of external validation. In addition, detailed pathological information such as immunohistochemical staining, cell distribution density, and microvessel density was not used as the reference, which may help confirm these findings and should be considered in the future.

The f_mean and f_kurtosis from IVIM MRI in enhanced tumor region demonstrated the difference between glioma and solitary brain metastasis, and their combined model could be used as a useful imaging biomarker promoting the differentiation.

SBM: solitary brain metastasis

HGG: high grade glioma

ADC_min, ADC_kurtosis: minimum, kurtosis of ADC

APTWI: amide proton transfer-weighted imaging

D: diffusion coefficient

D_min: minimum diffusion coefficient

f: perfusion fraction

f₁₀, f₉₀, f_mean, f_kurtosis, f_entropy, f_max: 10^th percentile,90^th percentile, mean, kurtosis, entropy, maximum of perfusion fraction

D*: pseudo-diffusion coefficient

D*_max: maximum of the pseudo-diffusion

MTRasym (3.5 ppm): asymmetric magnetization transfer ratio at 3.5 ppm

MTRasym (3.5 ppm)₁₀, MTRasym (3.5 ppm)₉₀: 10^th percentile,90^th percentile of asymmetric magnetization transfer ratio at 3.5 ppm

FA: fractional anisotropy

V_iso: isotropic volume fraction

ODI: orientation dispersion index

NODDI: neurite orientation dispersion and density imaging

Ethics approval and consent to participate

The protocol was reviewed and approved by the Ethics Committee of Shanxi Provincial People's Hospital /Fifth Hospital of Shanxi Medical University (2022 Research Review No. 153). The informed consent was waived the same time. And all methods were carried out in accordance with the Declaration of Helsinki.

Consent for publication

Not applicable.

Availability of data and materials

The datasets used and analyzed during the current study are available from the corresponding author on reasonable request..

Competing interests

Two authors of this manuscript (J.G., and M.Z.) are the employees of GE Healthcare. No known competing financial interests or personal relationships to declare for the remaining authors.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Authors' contributions

Y.S., R.C. and J.G. conceived and designed the work that led to the submission. Y.S. and L.H. acquired data. Y.S., and J.G. interpreted the results. Y.S. and J.G. wrote the main manuscript text and prepared all tables and figures. J.G. and M.Z. provided the software used in the work; H.J. L.H. Y.H. and C.X. have drafted the work and substantively revised it. All authors performed manuscript revisions and approved the final version of the manuscript.

Acknowledgements

Not applicable

Miller KD, Ostrom QT, Kruchko C, Patil N, Tihan T, Cioffi G, et al: Brain and other central nervous system tumor statistics, 2021. CA Cancer J Clin 2021, 71(5):381-406.
Sacks P, Rahman M: Epidemiology of Brain Metastases. Neurosurg Clin N Am 2020, 31(4):481-488.
Weller M, van den Bent M, Preusser M, Le Rhun E, Tonn JC, Minniti G, et al: EANO guidelines on the diagnosis and treatment of diffuse gliomas of adulthood. Nat Rev Clin Oncol 2021, 18(3):170-186.
Vogelbaum MA, Brown PD, Messersmith H, Brastianos PK, Burri S, Cahill D, et al: Treatment for Brain Metastases: ASCO-SNO-ASTRO Guideline. J Clin Oncol 2022, 40(5):492-516.
Faehndrich J, Weidauer S, Pilatus U, Oszvald A, Zanella FE, Hattingen E: Neuroradiological viewpoint on the diagnostics of space-occupying brain lesions. Clin Neuroradiol 2011, 21(3):123-139.
Schwartz KM, Erickson BJ, Lucchinetti C: Pattern of T2 hypointensity associated with ring-enhancing brain lesions can help to differentiate pathology. Neuroradiology 2006, 48(3):143-149.
Baris MM, Celik AO, Gezer NS, Ada E: Role of mass effect, tumor volume and peritumoral edema volume in the differential diagnosis of primary brain tumor and metastasis. Clin Neurol Neurosurg 2016, 148:67-71.
Jones KM, Pollard AC, Pagel MD: Clinical applications of chemical exchange saturation transfer (CEST) MRI. J Magn Reson Imaging 2018, 47(1):11-27.
Mehrabian H, Detsky J, Soliman H, Sahgal A, Stanisz GJ: Advanced Magnetic Resonance Imaging Techniques in Management of Brain Metastases. Front Oncol 2019, 9:440.
Paschoal AM, Leoni RF, Dos Santos AC, Paiva FF: Intravoxel incoherent motion MRI in neurological and cerebrovascular diseases. Neuroimage Clin 2018, 20:705-714.
Padhani AR, Liu G, Koh DM, Chenevert TL, Thoeny HC, Takahara T, et al: Diffusion-weighted magnetic resonance imaging as a cancer biomarker: consensus and recommendations. Neoplasia 2009, 11(2):102-125.
Le Bihan D, Turner R: The capillary network: a link between IVIM and classical perfusion. Magn Reson Med 1992, 27(1):171-178.
Zhou J, Lal B, Wilson DA, Laterra J, van Zijl PC: Amide proton transfer (APT) contrast for imaging of brain tumors. Magn Reson Med 2003, 50(6):1120-1126.
Shim WH, Kim HS, Choi CG, Kim SJ: Comparison of Apparent Diffusion Coefficient and Intravoxel Incoherent Motion for Differentiating among Glioblastoma, Metastasis, and Lymphoma Focusing on Diffusion-Related Parameter. PLoS One 2015, 10(7):e0134761.
Caravan I, Ciortea CA, Contis A, Lebovici A: Diagnostic value of apparent diffusion coefficient in differentiating between high-grade gliomas and brain metastases. Acta Radiol 2018, 59(5):599-605.
Yang Y, Qu X, Huang Y, Afsar K, Yan G, Guo G, et al: Preliminary application of 3.0 T magnetic resonance chemical exchange saturation transfer imaging in brain metastasis of lung cancer. BMC Med Imaging 2020, 20(1):4.
Han C, Huang S, Guo J, Zhuang X, Han H: Use of a high b-value for diffusion weighted imaging of peritumoral regions to differentiate high-grade gliomas and solitary metastases. J Magn Reson Imaging 2015, 42(1):80-86.
Lambin P, Rios-Velazquez E, Leijenaar R, Carvalho S, van Stiphout RG, Granton P, et al: Radiomics: extracting more information from medical images using advanced feature analysis. Eur J Cancer 2012, 48(4):441-446.
Yu H, Lou H, Zou T, Wang X, Jiang S, Huang Z, et al: Applying protein-based amide proton transfer MR imaging to distinguish solitary brain metastases from glioblastoma. Eur Radiol 2017, 27(11):4516-4524.
Nguyen DH, Nguyen DM, Nguyen HV, Nguyen-Thi VA, Nguyen-Thi HA, Nguyen TX, et al: Discrimination between glioblastoma and solitary brain metastasis: a quantitative analysis based on FLAIR signal intensity. Eur Rev Med Pharmacol Sci 2022, 26(10):3577-3584.
Mouthuy N, Cosnard G, Abarca-Quinones J, Michoux N: Multiparametric magnetic resonance imaging to differentiate high-grade gliomas and brain metastases. J Neuroradiol 2012, 39(5):301-307.
Kamimura K, Nakajo M, Yoneyama T, Fukukura Y, Hirano H, Goto Y, et al: Histogram analysis of amide proton transfer-weighted imaging: comparison of glioblastoma and solitary brain metastasis in enhancing tumors and peritumoral regions. Eur Radiol 2019, 29(8):4133-4140.
Gao E, Gao A, Kit Kung W, Shi L, Bai J, Zhao G, et al: Histogram analysis based on diffusion kurtosis imaging: Differentiating glioblastoma multiforme from single brain metastasis and comparing the diagnostic performance of two region of interest placements. Eur J Radiol 2022, 147:110104.
Louis DN, Perry A, Reifenberger G, von Deimling A, Figarella-Branger D, Cavenee WK, et al: The 2016 World Health Organization Classification of Tumors of the Central Nervous System: a summary. Acta Neuropathol 2016, 131(6):803-820.
Le Rhun E, Guckenberger M, Smits M, Dummer R, Bachelot T, Sahm F, et al: EANO-ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up of patients with brain metastasis from solid tumours. Ann Oncol 2021, 32(11):1332-1347.
Mori N, Inoue C, Tamura H, Nagasaka T, Ren H, Sato S, et al: Apparent diffusion coefficient and intravoxel incoherent motion-diffusion kurtosis model parameters in invasive breast cancer: Correlation with the histological parameters of whole-slide imaging. Magn Reson Imaging 2022, 90:53-60.
Luo H, He L, Cheng W, Gao S: The diagnostic value of intravoxel incoherent motion imaging in differentiating high-grade from low-grade gliomas: a systematic review and meta-analysis. Br J Radiol 2021, 94(1121):20201321.
Mao J, Zeng W, Zhang Q, Yang Z, Yan X, Zhang H, et al: Differentiation between high-grade gliomas and solitary brain metastases: a comparison of five diffusion-weighted MRI models. BMC Med Imaging 2020, 20(1):124.
Le Bihan D: What can we see with IVIM MRI? Neuroimage 2019, 187:56-67.
Malagi AV, Das CJ, Khare K, Calamante F, Mehndiratta A: Effect of combination and number of b values in IVIM analysis with post-processing methodology: simulation and clinical study. MAGMA 2019, 32(5):519-527.
Chabert S, Verdu J, Huerta G, Montalba C, Cox P, Riveros R, et al: Impact of b-Value Sampling Scheme on Brain IVIM Parameter Estimation in Healthy Subjects. Magn Reson Med Sci 2020, 19(3):216-226.
Goerke S, Milde KS, Bukowiecki R, Kunz P, Klika KD, Wiglenda T, et al: Aggregation-induced changes in the chemical exchange saturation transfer (CEST) signals of proteins. NMR Biomed 2017, 30(1).
Kang XW, Xi YB, Liu TT, Wang N, Zhu YQ, Wang XR, et al: Grading of Glioma: combined diagnostic value of amide proton transfer weighted, arterial spin labeling and diffusion weighted magnetic resonance imaging. BMC Med Imaging 2020, 20(1):50.
Togao O, Hiwatashi A, Yamashita K, Kikuchi K, Mizoguchi M, Yoshimoto K, et al: Differentiation of high-grade and low-grade diffuse gliomas by intravoxel incoherent motion MR imaging. Neuro Oncol 2016, 18(1):132-141.
Zou T, Yu H, Jiang C, Wang X, Jiang S, Rui Q, et al: Differentiating the histologic grades of gliomas preoperatively using amide proton transfer-weighted (APTW) and intravoxel incoherent motion MRI. NMR Biomed 2018, 31(1).

Competing interest reported. Two authors of this manuscript (J.G., and M.Z.) are the employees of GE Healthcare. No known other competing financial interests or personal relationships to declare for the remaining authors.

Download PDF

Version 1

posted

You are reading this latest preprint version

Intravoxel incoherent motion (IVIM) for distinguishing glioma and solitary brain metastasis: A pilot multi-parameter magnetic resonance imaging histogram and texture analysis

Status:

Version 1

Abstract

Figures

Background

Methods

Participants

Data Acquisition

Imaging Processing

Statistical analysis

Results

Clinical characteristics

The diagnostic performance with clinical, first-order, and histogram features

Discussion

Conclusions

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1