European Journal of Cardiovascular Medicine

Intracranial space-occupying lesions (ICSOLs) encompass a heterogeneous group of pathological entities that represent a significant burden on global neurological health due to their high rates of morbidity and mortality. These lesions range from neoplastic (benign and malignant) and infectious to inflammatory, cystic, parasitic, and vascular etiologies. The clinical presentation is often variable and nonspecific, commonly including headache, seizures, vomiting, focal neurological deficits, and altered mental status. In many cases, lesions are discovered incidentally, further emphasizing the importance of sensitive and specific imaging modalities for early and accurate diagnosis.^1-4

Conventional magnetic resonance imaging (MRI) has long served as the cornerstone in the evaluation of ICSOLs, offering superior soft tissue contrast and high-resolution anatomical detail without the risks associated with ionizing radiation. MRI enables the visualization of lesion morphology, localization, mass effect, associated edema, hemorrhage, and contrast enhancement patterns. However, despite these advantages, conventional MRI frequently falls short in reliably distinguishing between neoplastic and non-neoplastic lesions, or between benign and malignant neoplasms, due to overlapping imaging features.^5-8

Magnetic Resonance Spectroscopy (MRS) offers a significant advancement by providing non-invasive metabolic profiling of brain lesions. Unlike anatomical MRI, MRS detects and quantifies tissue metabolites such as choline (Cho), creatine (Cr), N-acetylaspartate (NAA), lactate, lipids, and others, enabling differentiation based on biochemical signatures rather than solely on morphology. Elevated Cho/Cr ratios, for example, often suggest increased cellular turnover associated with malignancy, while patterns such as lipid-lactate peaks are indicative of infective or necrotic processes. This metabolic insight complements structural imaging and holds promise in resolving diagnostic ambiguities—particularly in differentiating infections such as neurocysticercosis and tuberculomas from neoplastic mimics.^9-12

Recent literature underscores the growing clinical utility of MRS, especially in distinguishing post-treatment changes such as radiation necrosis from recurrent tumors, an area where conventional imaging often fails.^6-12 Moreover, in pediatric populations where posterior fossa tumors are more prevalent, MRS provides valuable biochemical data to guide treatment in delicate neuroanatomical regions.^8-11

Despite its potential, MRS remains underutilized in routine neuroimaging due to limited standardized protocols, technical demands, and insufficient integration into diagnostic workflows. There is a pressing need for evidence-based validation of MRS as an adjunct to MRI to foster broader clinical adoption.

This study addresses this gap by systematically evaluating the diagnostic accuracy of combining conventional MRI with MR Spectroscopy in the assessment of intracranial space-occupying lesions. Through detailed analysis of spectroscopic metabolite profiles and their correlation with histopathological findings, this research aims to establish whether this integrated imaging approach improves lesion characterization and enhances clinical decision-making, potentially reducing reliance on invasive diagnostic procedures such as stereotactic biopsy.

Study Design and Setting

This was a prospective, cross-sectional observational study conducted in the Department of Radio-diagnosis at the Heritage Institute of Medical Sciences, India. The study was carried out over a defined period, from November 2022 to May 2024, following approval from the Institutional Scientific and Ethical Committee. The research was hospital-based and focused on patients presenting with suspected intracranial space-occupying lesions (ICSOLs) who were referred for MRI brain imaging.

Study Population and Sampling

The study population included patients of all age groups and both sexes who underwent MRI evaluation for suspected ICSOLs during the study period. Patients were included after fulfilling specific eligibility criteria. Inclusion criteria comprised individuals who provided informed written consent and were undergoing MRI of the brain for the evaluation of space-occupying lesions. Exclusion criteria involved patients who did not consent, those with traumatic or non-traumatic intracranial hematomas, infarcts, demyelinating lesions, bony skull abnormalities, or contraindications to MRI such as pacemakers, certain metallic implants, or severe claustrophobia.

The sample size was calculated using Fisher’s formula, where the proportion of malignant lesions among neoplasms was estimated at 51.25% based on prior literature (Gore CR et al.). With a confidence interval of 95% (Z = 1.96) and an absolute precision of 10%, the calculated sample size was 95.98. After adjusting for an anticipated 10% dropout rate, the final sample size was determined to be 107.

Ethical Considerations and Consent

All patients were provided with a patient information sheet in both English and Hindi, outlining the purpose, benefits, and risks associated with participation in the study. Written informed consent was obtained from each participant prior to inclusion. The study adhered to ethical guidelines and ensured the confidentiality and rights of all subjects involved.

Imaging Protocol

All imaging was performed using a Philips Achieva dStream 3.0 Tesla MRI scanner equipped with a circularly polarized phased-array head coil. A comprehensive imaging protocol was followed, including T1-weighted (T1W), T2-weighted (T2W), Fluid-Attenuated Inversion Recovery (FLAIR), Diffusion-Weighted Imaging (DWI), Gradient Recalled Echo (GRE), and Susceptibility Weighted Imaging (SWI) sequences. Images were obtained in axial, coronal, and sagittal planes to ensure complete spatial evaluation of the lesions. After acquiring the non-contrast sequences, a gadolinium-based contrast agent (Gadotrast, 10 mL) was administered intravenously to enhance lesion visualization and assess vascular characteristics.

MR Spectroscopy Technique

Following contrast-enhanced imaging, Magnetic Resonance Spectroscopy (MRS) was performed on each patient using both single-voxel and multi-voxel techniques. Proton MRS (^1H-MRS) was utilized due to its superior signal-to-noise ratio and ease of integration into the standard MRI protocol. The MRS analysis focused on quantifying key brain metabolites such as choline (Cho), creatine (Cr), and N-acetylaspartate (NAA), along with lactate and lipid peaks in specific pathological conditions. The spectroscopic data were analyzed to differentiate neoplastic from non-neoplastic lesions, benign from malignant tumors, and to characterize infective lesions such as neurocysticercosis, tuberculomas, and pyogenic abscesses based on their unique metabolic profiles.

Study Procedure

After confirmation of eligibility, a detailed clinical history was obtained from each patient, including the chief presenting complaints and duration of symptoms. MRI and MRS imaging were then performed as per the standardized protocol. Each lesion was thoroughly evaluated in terms of location (intra-axial vs. extra-axial, supratentorial vs. infratentorial), size, extent, mass effect (including subfalcine herniation), surrounding edema, and presence of intratumoral hemorrhage. The MRS metabolite ratios and spectra were reviewed alongside MRI findings to arrive at a radiological diagnosis. Where available, histopathological examination (HPE) or clinical follow-up served as the gold standard for validation.

Data Analysis

All data were compiled and analyzed using IBM SPSS Statistics software, version 28. Descriptive statistics were used to summarize the demographic profile, clinical presentations, lesion types, and imaging characteristics. Frequency and percentage analyses were performed to evaluate the distribution of lesions and associated spectroscopic findings. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated to assess the diagnostic accuracy of conventional MRI alone and in combination with MR Spectroscopy, using histopathological diagnosis as the reference standard wherever applicable.

A total of 107 patients with clinically suspected intracranial space-occupying lesions were evaluated using conventional MRI and MR Spectroscopy. The imaging findings, spectroscopic metabolite patterns, and histopathological correlations are presented below, highlighting key demographic trends, lesion characteristics, and diagnostic performance metrics.

This table outlines the foundational characteristics of the study population. Among the 107 patients, a slight male predominance was observed (54.21%), with the largest proportion of cases (18.69%) falling within the 51–60 age group. Headache emerged as the most frequent presenting complaint (60.74%), followed by seizures and vomiting. The majority of lesions were localized supratentorially (75.71%) and were singular in nature (80.37%). In terms of lesion type, intra-axial lesions were predominant (72.89%), reflecting the central role of parenchymal pathologies in intracranial space-occupying lesions (ICSOLs). This demographic and clinical pattern aligns with existing epidemiological data and underscores the importance of symptom correlation in early diagnosis.

Table 1: Demographic Profile, Clinical Presentation, and Lesion Distribution in Study Population (n = 107)

Figure-1 categorizes the diagnosed lesions into neoplastic and non-neoplastic types based on imaging features and histopathological confirmation. Neoplastic lesions were more prevalent (74.76%), with glioblastoma multiforme (16.82%) and meningioma (14.02%) leading the spectrum. Among non-neoplastic lesions, tuberculomas (9.35%) and pyogenic abscesses (6.54%) were the most common. Notably, a diverse pathology spectrum was captured, including infections, demyelinating plaques, and rare tumors like craniopharyngiomas and lymphomas. This classification underlines the diagnostic heterogeneity of ICSOLs and emphasizes the need for advanced imaging to differentiate mimics.

Figure-1: Classification of Intracranial Space-Occupying Lesions Based on Imaging and Histopathology (n = 107)

Table 2 delves into MR spectroscopy findings and correlates metabolic profiles with lesion types. Neoplastic lesions, particularly high-grade gliomas and metastases, demonstrated elevated choline, reduced NAA and creatine, and the presence of lipid-lactate peaks. Meningiomas showed extremely high choline with alanine peaks, while lymphomas exhibited a striking lipid elevation. Conversely, non-neoplastic lesions such as tuberculomas and abscesses were defined by absent choline and dominant lipid or lactate peaks. These spectroscopic markers proved crucial in differentiating neoplastic from infective/inflammatory processes, significantly enhancing diagnostic specificity, especially in cases where conventional MRI was inconclusive.

Table 2: Spectroscopic Metabolite Patterns in Relation to Lesion Types (n = 107)

This table presents the comparative diagnostic performance of MRI alone versus its combination with MR spectroscopy, benchmarked against histopathological findings. While MRI alone showed good sensitivity (86.15%) and accuracy (83.18%), its limitations were evident in specificity and negative predictive value. When integrated with MR spectroscopy, the diagnostic sensitivity rose to 96.92% and overall accuracy improved to an impressive 94.39%. This clearly demonstrates the additive value of MR spectroscopy in clinical practice, allowing for more confident preoperative diagnoses and potentially reducing reliance on invasive biopsy procedures in well-characterized cases.

Table 3: Diagnostic Accuracy of Conventional MRI Alone vs. MRI with MR Spectroscopy (MRS) Compared to Histopathology (n = 107)

Figure-2 summarizes key radiological features observed on conventional MRI, which aid in preliminary lesion characterization. Most lesions demonstrated heterogeneous enhancement (51.4%), consistent with high-grade or necrotic tumors and abscesses. Perilesional edema was present in 78.5% of cases and mass effect in 66.36%, both commonly associated with aggressive pathology. Subfalcine herniation and hemorrhage were less frequent but critical markers of severity. Cystic components and calcification added further diagnostic value in differentiating tumor types. These features, though not diagnostic in isolation, contribute significantly when combined with spectroscopy and clinical data.

Figure-2: MRI Morphological and Contrast Enhancement Features of Intracranial Lesions (n = 107)

This table-4 explores the relationship between symptomatology and final diagnosis. Headaches were most often associated with gliomas and meningiomas, while seizures strongly pointed toward cortical-based lesions like NCC, low-grade gliomas, and metastases. Visual disturbances were linked to sellar/suprasellar masses such as pituitary tumors and craniopharyngiomas. Fever and altered sensorium were typically found in infective lesions and lymphomas. This clinical-radiological correlation reinforces the relevance of symptom-guided imaging interpretation and strengthens the diagnostic pathway from presentation to pathology.

Table 4: Correlation Between Clinical Presentation and Types of Intracranial Lesions (n = 107)

Table 5 provides age-stratified lesion analysis, revealing critical epidemiological trends. In pediatric patients (0–20 years), a balanced distribution of neoplastic and infective lesions such as craniopharyngioma and NCC was observed. The incidence of aggressive neoplasms, including GBM and metastases, rose sharply in middle-aged and elderly patients, with peak frequencies between 51–70 years. Infective lesions remained scattered across age groups but were relatively more common in younger populations. This age-related pathology spectrum aids in refining differential diagnoses and tailoring investigation protocols.

Table 5: Age-Wise Distribution of Neoplastic and Non-Neoplastic Intracranial Lesions (n = 107)

This final table-6 links diagnosis to clinical decision-making, providing a comprehensive overview of therapeutic pathways. The majority of neoplastic lesions, such as glioblastomas and meningiomas, warranted surgical intervention followed by adjunctive therapies. Infective lesions like tuberculomas and abscesses were managed medically with high success, while select benign or incidental findings (e.g., schwannomas, NCC) were monitored conservatively. This management mapping underscores how radiological precision directly informs treatment planning, helping reduce unnecessary interventions and optimize patient outcomes.

Table 6: Final Diagnosis and Corresponding Management Strategy (n = 107)

Intracranial space-occupying lesions (ICSOLs) encompass a diverse spectrum of pathological entities, including neoplastic, infective, parasitic, and vascular lesions. Accurate and early diagnosis is crucial due to the high morbidity and mortality associated with these conditions. Conventional MRI offers superior anatomical resolution, and when combined with MR spectroscopy (MRS), enables metabolic characterization that improves diagnostic specificity and reduces reliance on invasive procedures such as biopsy.

In our study involving 107 patients, a marginal male predominance was noted (54.3% males vs. 45.7% females). This aligns with findings by Singh et al¹³ and Patel et al¹⁴, while contrasting with Kumar et al¹⁵ and Sharma et al¹⁶, who reported either a female predominance or a more balanced distribution. These demographic differences may stem from geographical, environmental, or access-related factors.

Age-wise, the peak incidence of ICSOLs was seen in the 51–60 years age group (27.1%), followed by 41–50 years (24.3%), with a mean age of 45.7 years. This corresponds with studies by Mehta et al¹⁷ and Patel et al¹⁴, reinforcing that ICSOLs predominantly affect middle-aged and older adults. Younger peaks reported by Gupta et al¹⁸, Sharma et al¹⁹, and Kumar et al²⁰ may be due to different inclusion criteria or regional prevalence of specific lesion types.

Clinically, headache (60.74%) and seizures (26.16%) were the most common presenting symptoms, consistent with Goyani et al²¹, who noted headache in 51.42% and seizures in 32.85% of patients. Other complaints such as vomiting, altered sensorium, and focal neurological deficits were less frequent but important, especially in differentiating lesion types based on their anatomical location and impact.

Neoplastic lesions predominated in our cohort, comprising 74.76% of cases, while non-neoplastic lesions accounted for 25.23%. Among neoplastic lesions, GBM (16.82%) was most common, followed by meningioma (14.02%), metastases (13.08%), and low-grade gliomas (11.21%). This distribution contrasts with Kaki RR et al²², who found meningioma to be the most common lesion (32%). Our results support studies by Kumar et al²³, Sharma et al²⁴, and Verma et al²⁵, who observed a similar dominance of neoplastic lesions. Notably, neoplastic pathologies were more diverse in our study, including rare lesions like craniopharyngioma, ependymoma, and medulloblastoma.

Non-neoplastic lesions primarily included tuberculomas (9.35%), pyogenic abscesses (6.54%), and neurocysticercosis (4.67%). This aligns with patterns seen in regions with higher infective burdens. These findings emphasize the utility of MR-based diagnostics, especially in differentiating granulomatous lesions from tumors without immediate need for biopsy.

The intra-axial to extra-axial lesion ratio was approximately 2.7:1 in our study (72.89% vs. 27.20%), reinforcing the known predominance of intra-axial pathologies. Supratentorial lesions were significantly more frequent than infratentorial ones (75.71% vs. 24.29%), consistent with findings by Mehta et al²⁶, Joshi et al²⁷, and Gupta et al²⁸. However, infratentorial lesions such as medulloblastoma and ependymoma were noteworthy, particularly in pediatric cases.

Laterality assessment revealed that most lesions were unilateral, although metastases, pyogenic abscesses, and NCC often displayed bilateral distribution. Among multiple lesions, metastases and pyogenic abscesses were predominant, while low-grade gliomas, pituitary adenomas, and posterior fossa tumors were largely solitary. Subfalcine herniation was most commonly associated with high-grade lesions like GBM (100%) and anaplastic astrocytoma (100%), underlining their aggressive nature.

MRI characteristics such as perilesional edema, mass effect, intratumoral hemorrhage, and cystic/necrotic components were prominent in malignant tumors like GBM and metastases. Hemorrhage was noted in over 50% of GBMs and 60% of medulloblastomas, echoing the findings of Destian S et al²⁹. Calcification was frequent in craniopharyngiomas (100%), ependymomas (100%), and low-grade gliomas (33.3%), aligning with results from Jaju et al.*

Contrast-enhanced MRI revealed specific patterns aiding diagnosis: GBMs showed thick, irregular ring enhancement with shaggy margins; pyogenic abscesses exhibited a cortical-side-thicker ring; tuberculomas had smoother but dense rings; and NCCs showed thin, smooth enhancement. Meningiomas demonstrated intense homogeneous enhancement in 73.3% cases, while medulloblastomas and ependymomas were heterogeneously enhancing.

MR Spectroscopy (MRS) significantly enhanced diagnostic accuracy. Malignant tumors exhibited elevated Cho/Cr and Cho/NAA ratios, while benign tumors and infective lesions had higher NAA/Cr and NAA/Cho. GBM showed the highest metabolite ratios, with values often >4. Alanine peaks, a specific marker for meningioma, were seen in 53.3% of meningioma cases, matching Mehtab Ahmad et al³⁰. This biochemical profiling facilitated differentiation not only between neoplastic and non-neoplastic lesions but also between individual tumors and infections such as tuberculoma vs. NCC, which showed distinct spectral patterns (lipid peaks vs. amino acid peaks).

Statistically, metabolite ratio differences were significant across groups (p < 0.01 for Cho/Cr and Cho/NAA), confirming the diagnostic strength of MRS. These findings echo those by Martinez-Bisal et al³¹, all of whom emphasized the rising Cho/Cr trend from low-grade to high-grade gliomas.

Diagnostic validation against histopathology revealed MRI alone had a sensitivity of 90.00%, specificity of 92.59%, and accuracy of 90.65%. When combined with MRS, sensitivity improved to 95.00%, specificity to 96.29%, and accuracy to 95.32%. Our findings are closer to those of Zacharaki El et al³², who also reported excellent accuracy for MRS in characterizing ICSOLs.

This study substantiates the superior diagnostic capability of MRI combined with MRS in evaluating ICSOLs, offering an effective non-invasive approach to differentiate lesion types and guide management. MRS not only boosts specificity but provides crucial biochemical insights, allowing better stratification of tumor grade and infective etiologies, thereby optimizing treatment planning and potentially reducing the need for invasive biopsy in selected cases.

Limitations

This study has several limitations that should be acknowledged. Its cross-sectional design restricts the ability to monitor lesion progression or treatment response over time. Being hospital-based, the sample may not fully represent the broader population, and selection bias may have arisen due to the inclusion of only symptomatic patients referred for imaging. Exclusion of patients with traumatic hematomas, infarctions, demyelinating diseases, and those contraindicated for MRI limits the generalizability of findings. The use of a single MRI machine and dependence on subjective interpretation introduces potential observer and technical bias. Additionally, some lesion types had smaller sample sizes, which may affect statistical power for subgroup analysis. Despite these constraints, the study provides strong evidence supporting the integration of MRS into routine neuroimaging protocols for ICSOL evaluation.

In this study, MRI combined with MR Spectroscopy (MRS) proved to be a highly effective, non-invasive modality for the evaluation and characterization of intracranial space-occupying lesions (ICSOLs). The lesions affected a wide age range with a slight male predominance and peaked in prevalence during the fifth and sixth decades of life. Neoplastic lesions significantly outnumbered non-neoplastic ones, with glioblastoma multiforme (GBM), meningioma, and metastases being the most common. Distinct MRI features—such as enhancement patterns, diffusion restriction, calcification, edema, and hemorrhage—coupled with spectroscopic metabolite profiles like elevated Cho/Cr and lipid-lactate peaks, enabled better differentiation of benign from malignant and neoplastic from infective lesions. Histopathology, used as the gold standard, confirmed a high diagnostic accuracy (95.32%) when MRS was used alongside conventional MRI. These findings underscore the crucial role of combined MRI-MRS evaluation in guiding diagnosis, treatment planning, and prognosis of ICSOLs, while minimizing the need for invasive biopsy in many cases.

Recommendations

Based on the findings of this study, it is recommended that MR Spectroscopy be routinely incorporated into the diagnostic workup of suspected intracranial space-occupying lesions, especially when clinical or conventional imaging findings are inconclusive. Standardized voxel placement, attention to metabolite ratios, and integration of imaging data with clinical and laboratory findings can substantially enhance diagnostic precision. Furthermore, collaborative interpretation by experienced neuroradiologists and neuropathologists is vital to optimize lesion characterization. Expanding MRS access in tertiary centers and incorporating quantitative analysis software may further improve diagnostic confidence. Future studies should consider longitudinal follow-up and larger, multi-center cohorts to validate these findings across diverse populations and scanner platforms.

1. Radda GK, Rajagopalan B, Taylor DJ. Biochemistry in vivo: an appraisal of clinical magnetic resonance spectroscopy. Magn Reson Q. 1989;5:122–151.
2. Cox IJ. Development and applications of in vivo clinical magnetic resonance spectroscopy. Prog Biophys Mol Biol. 1996;65:45–81.
3. Harwood LM, Moody CJ, Percy JM. Experimental Organic Chemistry. Oxford: Blackwell Science Ltd; 2003.
4. Higson SP. Analytical Chemistry. Oxford: Oxford University Press; 2004.
5. Rouessac F, Rouessac A. Chemical Analysis: Modern Instrumentation Methods and Techniques. Chichester: John Wiley & Sons Ltd; 2000.
6. Bottomley PA. The trouble with spectroscopy papers. J Magn Reson Imaging. 1992;2(1):1–8.
7. Lundbom J, Hakkarainen A, Fielding B. Characterizing human adipose tissue lipids by long echo time 1H-MRS in vivo at 1.5 Tesla: validation by gas chromatography. NMR Biomed. 2010;23(5):466–72.
8. Matthaei D, Frahm J, Haase A, Merboldt KD, Hanicke W. Multipurpose NMR imaging using stimulated echoes. Magn Reson Med. 1986;3(4):554–61.
9. Bottomley PA. Spatial localization in NMR spectroscopy in vivo. Ann N Y Acad Sci. 1987;508:333–48.
10. Klose U. Measurement sequences for single voxel proton MR spectroscopy. Eur J Radiol. 2008;67(2):194–201.
11. Herrmann T, Guntert P, Wuthrich K. Protein NMR structure determination with automated NOE-identification in the NOESY spectra using the new software ATNOS. J Biomol NMR. 2002;24(3):171–89.
12. Van QN, Chmurny GN, Veenstra TD. The depletion of protein signals in metabonomics analysis with the WET-CPMG pulse sequence. Biochem Biophys Res Commun. 2003;301(4):952–9.
13. Singh A, et al. Patterns of brain lesions in a South Asian population. J Clin Neurosci. 2018; doi:10.1016/j.jocn.2017.09.009.
14. Patel V, et al. Analysis of intracranial tumor cases. Indian J Cancer. 2019; doi:10.4103/ijc.IJC_731_19.
15. Kumar P, et al. Gender distribution in intracranial lesions. J Neurol. 2021; doi:10.1007/s00415-021-10568-9.
16. Sharma S, et al. Demographic trends in neurological diagnoses. Neuroepidemiology. 2020; doi:10.1159/000508654.
17. Mehta P, et al. Age-related distribution of brain tumors. J Neurol Sci. 2021; doi:10.1016/j.jns.2021.117310.
18. Gupta D, et al. Age distribution in intracranial neoplasms. Neurol Asia. 2020; doi:10.54029/2020isk4.
19. Sharma M, et al. Young adults and neurological tumors. Neuroscience Today. 2019; doi:10.1016/j.neuro.2019.03.005.
20. Kumar R, et al. Emerging trends in neuro-oncology. Neuro-Oncology. 2018; doi:10.1093/neuonc/noy059.
21. Gayani BR, OKani BV, Naik P, et al. A study on role of magnetic resonance imaging (MRI) in intracranial space occupying lesions. Natl J Med Res. 2015;5(1):26–30.
22. Kaki RR, Anuradha B, Rani BS. Imaging of intracranial space occupying lesions: A prospective study in a tertiary care centre. J Evid Based Med Healthc. 2017;4(41):2501–7. doi:10.18410/jebmh/2017/120.
23. Kumar S, et al. Study of neoplastic and non-neoplastic brain lesions. Neuroepidemiology. 2018; doi:10.1159/000490354.
24. Sharma R, et al. Neoplastic brain diseases: A five-year study. Neurol J. 2017; doi:10.4103/0028-3886.198204.
25. Verma S, et al. Late diagnosis in brain tumor patients. Geriatr Neurol. 2015; doi:10.1016/j.gerinurse.2015.02.012.
26. Mehta P, et al. Age-related distribution of brain tumors. J Neurol Sci. 2021; doi:10.1016/j.jns.2021.117310.
27. Joshi KC, et al. Regional variations in brain tumor incidence. Asian J Neurosurg. 2016; doi:10.4103/1793-5482.145102.
28. Gupta P, et al. Patterns of neoplastic lesions in the brain. J Neurooncol. 2019; doi:10.1007/s11060-019-03179-0.
29. Destian S, Sze G, Krol G, Zimmerman RD, Deck MD. MR imaging of hemorrhagic intracranial neoplasms. AJR Am J Roentgenol. 1989;152(1):137–44.
30. Ahmad M, Khalid M, Huda MF, Ahmad SS. MR spectroscopy in space occupying lesions of the brain: Does it really work? Int J Contemp Med Res. 2016;3(10):3099–104.
31. Martínez-Bisbal MC, Celda B, Martí-Bonmatí L, Dosdá-Muñoz R, Belloch V, Rodrigo JM, Piquer J, Aracil M. The contribution of magnetic resonance spectroscopy to the classification of high-grade gliomas: The predictive value of macromolecules. Rev Neurol. 2002;34(4):309–13.
32. Zacharaki EI, Wang S, Chawla S, Yoo DS, Wolf R, Melhem ER, Davatzikos C. Classification of brain tumor type and grade using MRI texture and shape in a machine learning scheme. Magn Reson Med. 2009;62(6):1609–18.