Abstract
Background and aim: Tuberculous meningitis (TBM) and cryptococcal meningitis (CM) are easily misdiagnosed due to atypical clinical symptoms. It is difficult for physcians to achieve a rapid and accurate differential diagnosis of TBM in the early stages of disease onset. The aim of this study was to construct a diagnostic prediction model for TBM and CM.
Methods: In this retrospective study, 194 patients with TBM and CM were divided into two groups: training group (163 patients) and validation group (31 patients). Univariate and multivariate analyses were performed on the training group patients. The diagnostic factors were selected to construct the differential diagnostic prediction model for TBM and CM, and the prediction model was verified. A receiver operating characteristics curve (ROC) was constructed and used to evaluate the diagnostic value of the novel model.
Results: Univariate analysis of clinical characteristics revealed differences in eight parameters (P<0.05) between tuberculous meningitis and cryptococcal meningitis. The multivariate analyses showed there were five independent differential factors including age, disease course, albumin-to-globulin ratio, cerebrospinal fluid protein, and cerebrospinal fluid sugar to blood sugar ratio in this study between TBM and CM, while there was no significant difference in the number of nucleated cells in CSF (P=0.088). A differential diagnosis model for TBM and CM was constructed based on those factors. A ROC was constructed with an area under curve [AUC] of 94.5%, a sensitivity of 85.71%, and specificity of 94.59% in the training group.
Conclusion: The novel diagnostic scoring model for TBM and CM has greater differential diagnosis potential than previous reports, which can provide more reliable preliminary diagnosis results for primary hospitals, effectively reduce misdiagnosis, and provide references for early treatment.
- Clinical analysis
- Cryptococcal meningitis
- Diagnostic model
- Differential diagnosis
- Tuberculous meningitis
Meningitis is a common central nervous system infectious disease with high incidence in developing countries and high fatality and disability rates.1 Tuberculous meningitis (TBM) and cryptococcal meningitis (CM) are common types of infectious meningitis in clinical practice.2,3 In approximately 25% of TBM cases, delayed diagnosis and inadequate treatment may lead to poor prognosis and sequelae.4,5 Despite the introduction of new antimicrobial medications, techniques for diagnosis, and treatment strategies in recent years, mortality rates remain high.6-9 TBM and CM are easily misdiagnosed due to atypical clinical symptoms;10,11 thus, it is difficult for physicians to achieve a rapid and accurate differential diagnosis of TBM in the early stages of disease onset.
In 2010, Cohen et al.12 developed a scoring system to differentiate between CM and TBM in HIV-positive people. The scoring system included five differential variables: cerebrospinal fluid white blood cell count, neck stiffness, cerebrospinal fluid initial pressure, Glasgow Coma Scale score, and body temperature. Based on these, a differential diagnosis model was constructed with a sensitivity of 83% and a specificity of 79%. However, this model12 has some limitations: first, all the study participants were HIV-positive; second, more than 100 patients with cerebrospinal fluid abnormalities evaluated in the article were receiving antiretroviral therapy, suggesting that some cases of meningitis may be attributable to immune reconstitution inflammatory syndrome. Therefore, if this model is applied to HIV-negative people, it may show great differences.
Previous studies13,14 have shown that serum albumin-to-globulin ratio (AGR), as a combined measurement of two indicators (ie., albumin and globulin), has been validated to more accurately reflect the nutritional and inflammatory status of the body. The neutrophil-to-lymphocyte ratio in peripheral blood is associated with neurological and infectious diseases.13,15 However, Cohen et al.’s model12 did not include AGR and the neutrophil-to-lymphocyte ratio. Using these new variables, this study aims to establish a better TBM and CM differential diagnosis model in HIV-negative patients.
Materials and Methods
Participants
This study retrospectively analyzed the clinical data of patients with TBM and CM from the Department of Neurology and Department of Infectious Diseases of the First Affiliated Hospital of Chongqing Medical University from January 2013 to December 2021.
Inclusion Criteria
TBM: Mycobacterium tuberculosis is isolated from the cerebrospinal fluid (i.e., acid-fast bacilli are found in cerebrospinal fluid, M. tuberculosis is cultured in the cerebrospinal fluid, or cerebrospinal fluid microscopic nucleic acid amplification test is positive); or clinical symptoms, signs, and auxiliary examinations indicating the presence of meningitis, but Gram stain is negative, India ink stain is negative, bacterial and fungal cultures are negative, and one or more of the following conditions are combined: (1) brain magnetic resonance imaging (MRI) findings in TBM (hydrocephalus, edema, signs of basal meningeal enhancement, tuberculosis, and cerebral infarction) were consistent, (2) chest radiograph was consistent with active pulmonary tuberculosis or supported by clinical evidence of other extrapulmonary tuberculosis, and (3) anti-tuberculosis chemotherapy had a good effect.13
In addition to the symptoms and signs of meningitis, patients with CM should meet at least one of the following criteria: (1) cerebrospinal fluid culture is positive for Cryptococcus; (2) cerebrospinal fluid is positive for Indian ink-stained smear.
Exclusion Criteria
Age <16 years; presence of immunodeficiency: HIV, malignant tumor, autoimmune disease, long-term hormone, or other immunosuppressive conditions; meningitis after neurosurgery, post-traumatic meningitis, or a combination of multiple types of meningitis; incomplete medical records.
Methods
The clinical data of the patients included in this study were collected by two neurologists, and the diagnosis of all meningitis patients was reviewed by at least one professor. We double-checked and input the collected clinical data of patients to ensure the data accuracy. The collected data are as follows: (1) baseline data of patients (sex, age, clinical symptoms, signs, Glasgow Coma Scale score, etc.); (2) blood indicators (blood routine, neutrophil ratio, C reactive protein (CRP), serum sodium, albumin, etc.); (3) cerebrospinal fluid indicators (conventional, biochemical, smear, culture, etc.); (4) imaging data of patients (brain computed tomography [CT]/magnetic resonance imaging, chest CT, etc.).
Ethical Approval
This study was approved by the Ethics Committee of the First Affiliated Hospital of Chongqing Medical University. Written consent was waived, as this retrospective study does not influence the health care of included individuals. All patients’ data were anonymized.
Statistical Analysis
All statistical analyses were performed using SPSS software (version 25.0). The normally distributed measurement data are expressed as mean ± standard deviation (x±s), the skewed distribution measurement data are expressed as median (quartile) [M (P25, P75)]. Univariate analysis of the normal distribution measurement data was performed by single-factor test, and the univariate analysis of measurement data with skewed distribution was performed by the nonparametric test of two independent samples; counting data were expressed as the number of cases and percentage [n (%)], and comparison between groups was performed by Chi-square test. The variables with significance (P<0.05) were selected from the univariate analysis for multivariate analysis; the logistic stepwise forward (LR) method was used to screen the variables, and the independent difference factors (P<0.05) were selected. The nodules were constructed according to the regression coefficients. A diagnostic model for meningitis, with the best cut-off point for independent difference factors and the model, was determined according to the maximum Youden index. Finally, the receiver operating characteristics (ROC) curve of the training set was drawn, the area under the curve (AUC) was calculated to evaluate the model discrimination, and the model in the training set and validation set were tested to determine its sensitivity, specificity.
Results
Clinical Data
In total, 194 patients were included in this study. The patients were divided into a training group and a validation group according to the time of admission. Patients admitted from January 2013 to December 2019 were in the training group (163 patients). There were 126 cases of TBM and 37 cases of CM in the training group. Patients admitted from January 2020 to December 2021 were in the validation group (31 patients), including 24 cases of TBM and 7 cases of CM. Most of the parameters had no difference between the training group and the validation group in the baseline data (Table 1).
Comparison of clinical data of patients in the training group and validation group
Analysis of Clinical Features in Patients with Tuberculous Meningitis and Cryptococcal Meningitis
The clinical indicators of the patients included in the training group were subjected to univariate analysis, and the results showed the following: age, fever, disease duration, total protein, AGR, number of nucleated cells in the CSF, protein in the cerebrospinal fluid, the ratio of sugar in the CSF and blood sugar. The differences in these 8 variables were statistically significant (P<0.05; Table 2).
Univariate analysis of clinical features in patients with tuberculous meningitis and cryptococcal meningitis
The multivariate logistic regression analysis included the above eight variables. and the results showed that age (OR=1.076, 95%CI: 0.026-0.074), disease duration (OR=1.079, 95%CI: 0.028-0.076), AGR (OR=1380.731, 95%CI: 1.697-7.23), protein in CSF(OR=0.522, 95%CI: −0.65-0.312) and the ratio of CSF sugar to blood sugar (OR=0.003, 95%CI: −5.681-2.184) were the differential factors for differentiating TBM from CM. (Table 3.)
Multi-factor analysis of clinical features in patients with tuberculous meningitis and cryptococcal meningitis
Establishment of a Predictive Model to Distinguish Tuberculous Meningitis from Cryptococcal Meningitis
Using the differential factors selected by multivariate logistic analysis, a predictive model for distinguishing TBM from CM was constructed. The specific diagnostic score of each variable in the model was weighted according to the size of its corresponding regression coefficient. The division value of the independent predictors in the model was the cut-off value obtained when the sensitivity and specificity of the variable ROC curve were optimal. The specific diagnostic index (DI) of each factor is shown in Table 4.
Differential diagnosis model between tuberculous meningitis and cryptococcal meningitis
Total diagnostic index (TDI) is derived from the following computational model: TDI=DI (age)+DI (disease duration)+DI (AGR)+DI (CSF protein)+DI (the ratio of CSF sugar to blood sugar). The best value of the total model score was obtained by scoring the corresponding cases in the evaluation data set and constructing the ROC curve (Figure 1). When the sensitivity and specificity of the model were the best, the cutoff value was set as 3, while a score > 3 indicated TBM, while a score ≤ 3 indicated CM.
The Receiver Operating Characteristics Curve (ROC) of a predictive model for distinguishing TBM from CM with an AUC of 94.5%.
Evaluation of Model Indicators
The ROC curve of the training group is shown in Figure 1. The model showed a good degree of discrimination in the training group with an AUC of 94.5%. The model diagnosis model in the training group had a sensitivity of 85.71% and a specificity of 94.59%. The model was applied in the validation group including 24 patients with TBM and 7 patients with CM. The AUC of the diagnostic model was 85.40% and the sensitivity was 91.67%. The new diagnosis model showed good sensitivity and diagnostic efficiency in the validation test.
Discussion
This study established a novel diagnostic model for TBM and CM based on five differential factors: age, disease course, AGR, CSF protein, and the CSF glucose-to-blood glucose ratio. Compared to Cohen et al.’s model designed for HIV-positive populations,12 this model demonstrated higher specificity (94.59%) and comparable sensitivity (85.71%), making it more suitable for HIV-negative patients. Its applicability in primary healthcare settings highlights its potential for aiding early diagnosis and reducing misdiagnosis in resource-limited areas.
Consistent with previous studies,16 our findings show that CM patients are generally older than TBM patients, with mortality rates increasing significantly in individuals aged over 60 years. This phenomenon is likely linked to immunosenescence, a process characterized by a decline in immune function with age.17-19 Age-related immune dysfunction is associated with several phenotypic and functional abnormalities, such as the expansion of CMV-specific T cells17,21 and a reduced response to opportunistic pathogens like Cryptococcus spp.20 In contrast, TBM patients exhibited earlier symptom onset and shorter disease duration, corroborating prior findings that TBM often has a more acute course due to severe blood-brain barrier (BBB) disruption.22-24 M. tuberculosis invades and spreads through the subarachnoid space, triggering a significant inflammatory response characterized by fibrin deposition, exudates, and meningeal adhesion.24,31 This acute inflammation underpins the earlier manifestation of meningitis-specific symptoms, including fever, headache, and neck stiffness.
A notable finding of this study is the significantly lower AGR observed in TBM patients compared to CM patients. AGR, a composite marker of serum albumin and globulin levels, reflects the inflammatory state more comprehensively than either parameter alone.25,29 During inflammation, serum albumin levels decline due to reduced hepatic synthesis and increased capillary leakage, while globulin levels rise as immune-related proteins, such as interleukins and immunoglobulins, are produced.13,27-30 While previous studies have established AGR as a prognostic biomarker in various diseases,14 our findings suggest that TBM patients experience a more pronounced systemic inflammatory response than CM patients. The lower AGR in TBM may reflect the severity of inflammation and its impact on protein metabolism. This observation underscores the potential of AGR as a novel diagnostic indicator, particularly in differentiating TBM from CM in HIV-negative populations.
Our study also confirms that CSF protein levels are significantly higher in TBM than in CM patients, consistent with previous reports.7,11,16 BBB disruption caused by M. tuberculosis facilitates the entry of macromolecules into the CSF, resulting in elevated protein levels.24,31,32 The degree of protein elevation correlates with BBB integrity and inflammation severity, making CSF protein a valuable marker for assessing TBM pathology.
Additionally, this study found that CM patients are more likely to exhibit a lower CSF glucose-to-blood glucose ratio compared to TBM patients, aligning with prior research.11,33 The disproportionate glucose depletion in CM is likely attributable to the higher metabolic activity of Cryptococcus spp. in the CSF, which consumes glucose more rapidly than M. tuberculosis.34,35 Furthermore, Glucose Transporter Type 1 Deficiency Syndrome and increased glycolysis during infection may exacerbate glucose depletion.36 This finding reinforces the diagnostic utility of CSF glucose dynamics in distinguishing between TBM and CM. Notably, no significant differences were observed in CSF nucleated cell counts between TBM and CM groups (P=0.088). This could be attributed to the limited sample size and single-center design, warranting further investigation in larger, multi-center cohorts. As a result, CSF nucleated cell counts were excluded from the construction of the diagnostic scoring model.
The diagnostic model proposed in this study achieved an AUC of 94.5%, with superior specificity compared to the Cohen model.12 Unlike previous studies, this model focuses exclusively on HIV-negative patients while excluding other meningitis types, such as postoperative and post-traumatic meningitis. These criteria enhance its clinical relevance for primary healthcare settings, where HIV-negative cases predominate, and diagnostic resources are limited.
Limitations
Despite its promising findings, this study has several limitations. First, as a single-center retrospective study, potential baseline differences between the training and validation cohorts may affect model performance. Second, some TBM cases lacked gold-standard etiological diagnoses, which may influence the accuracy of the model. Third, the exclusion of HIV-positive patients limits the generalizability of our findings. Future prospective studies with multi-center cohorts are necessary to validate the model and refine its diagnostic efficacy across diverse populations.
Conclusion
In conclusion, the diagnostic model based on age, disease course, AGR, CSF protein, and CSF glucose-to-blood glucose ratio provides a valuable tool for distinguishing TBM and CM in HIV-negative patients. By improving specificity and maintaining sensitivity, this model offers reliable preliminary diagnostic insights for primary healthcare providers,12 facilitating early treatment and reducing misdiagnosis. Further validation in larger, multi-center studies will enhance its utility and applicability in broader clinical contexts.
Acknowledgments
We would like to thank Editage (www.editage.cn) for English language editing.
Footnotes
Disclosures: The authors have declared no competing interests. No funding was received for conducting this study. Data availability: De-identified data will be available by contacting the corresponding author.
- Received August 17, 2023.
- Revision received December 22, 2024.
- Accepted January 23, 2025.
References:
- 1.↵Bystritsky RJ, Chow FC. Infectious Meningitis and Encephalitis. Neurol Clin. 2022;40(1):77-91. doi:10.1016/j.ncl.2021.08.006.
- 2.↵Pyrgos V, Seitz AE, Steiner CA, Prevots DR, Williamson PR. Epidemiology of cryptococcal meningitis in the US: 1997-2009. PLoS One. 2013;8(2):e56269. doi:10.1371/journal.pone.0056269.
- 3.↵Thwaites GE, van Toorn R, Schoeman J. Tuberculous meningitis: more questions, still too few answers. Lancet Neurol. 2013;12(10):999-1010. doi:10.1016/S1474-4422(13)70168-6.
- 4.↵Slane VH, Unakal CG. Tuberculous Meningitis. In: StatPearls. Treasure Island (FL): StatPearls Publishing; September 2, 2024.
- 5.↵Chen Y, Wang Y, Liu X, Comparative diagnostic utility of metagenomic next-generation sequencing, GeneXpert, modified Ziehl-Neelsen staining, and culture using cerebrospinal fluid for tuberculous meningitis: A multi-center, retrospective study in China. J Clin Lab Anal. 2022;36(4):e24307. doi:10.1002/jcla.24307
- 6.↵Ellis J, Cresswell FV, Rhein J, Ssebambulidde K, Boulware DR. Cryptococcal Meningitis and Tuberculous Meningitis Co-infection in HIV-Infected Ugandan Adults. Open Forum Infect Dis. 2018;5(8):ofy193. doi:10.1093/ofid/ofy193.
- 7.↵Vidal JE, Peixoto de Miranda EJF, Gerhardt J, Croda M, Boulware DR. Is it possible to differentiate tuberculous and cryptococcal meningitis in HIV-infected patients using only clinical and basic cerebrospinal fluid characteristics? S Afr Med J. 2017;107(2):156-159. doi:10.7196/SAMJ.2017.v107i2.11162.
- 8.Yao Y, Zhang JT, Yan B, Voriconazole: a novel treatment option for cryptococcal meningitis. Infect Dis (Lond). 2015;47(10):694-700. doi:10.3109/23744235.2015.1044260.
- 9.↵Lee YC, Wang JT, Sun HY, Chen YC. Comparisons of clinical features and mortality of cryptococcal meningitis between patients with and without human immunodeficiency virus infection. J Microbiol Immunol Infect. 2011;44(5):338-345. doi:10.1016/j.jmii.2010.08.011.
- 10.↵Soria J, Metcalf T, Mori N, Mortality in hospitalized patients with tuberculous meningitis. BMC Infect Dis. 2019;19(1):9. doi:10.1186/s12879-018-3633-4.
- 11.↵Zhang B, Lv K, Bao J, Lu C, Lu Z. Clinical and laboratory factors in the differential diagnosis of tuberculous and cryptococcal meningitis in adult HIV-negative patients. Intern Med. 2013;52(14):1573-1578. doi:10.2169/internalmedicine.52.0168.
- 12.↵Cohen DB, Zijlstra EE, Mukaka M, Diagnosis of cryptococcal and tuberculous meningitis in a resource-limited African setting. Trop Med Int Health. 2010;15(8):910-917. doi:10.1111/j.1365-3156.2010.02565.x.
- 13.↵Li K, Tang H, Yang Y, Clinical features, long-term clinical outcomes, and prognostic factors of tuberculous meningitis in West China: a multivariate analysis of 154 adults. Expert Rev Anti Infect Ther. 2017;15(6):629-635. doi:10.1080/14787210.2017.1309974.
- 14.↵Zhou T, Yu ST, Chen WZ, Xie R, Yu JC. Pretreatment albumin globulin ratio has a superior prognostic value in laryngeal squamous cell carcinoma patients: a comparison study. J Cancer. 2019;10(3):594-601. doi:10.7150/jca.28817.
- 15.↵Bozbay M, Uyarel H. Neutrophiltolymphocyte ratio: A novel and simple prognostic marker for infective endocarditis. J Crit Care. 2015;30(4):822. doi:10.1016/j.jcrc.2015.04.115.
- 16.↵Qu J, Zhou T, Zhong C, Deng R, Lü X. Comparison of clinical features and prognostic factors in HIV-negative adults with cryptococcal meningitis and tuberculous meningitis: a retrospective study. BMC Infect Dis. 2017;17(1):51. doi:10.1186/s12879-016-2126-6.
- 17.↵Almanzar G, Schwaiger S, Jenewein B, Long-term cytomegalovirus infection leads to significant changes in the composition of the CD8+ T-cell repertoire, which may be the basis for an imbalance in the cytokine production profile in elderly persons. J Virol. 2005;79(6):3675-3683. doi:10.1128/JVI.79.6.3675-3683.2005
- 18.Asanuma H, Sharp M, Maecker HT, Maino VC, Arvin AM. Frequencies of memory T cells specific for varicella-zoster virus, herpes simplex virus, and cytomegalovirus by intracellular detection of cytokine expression. J Infect Dis. 2000;181(3):859-866. doi:10.1086/315347.
- 19.↵Castle SC. Impact of age-related immune dysfunction on risk of infections. Z Gerontol Geriatr. 2000;33(5):341-349. doi:10.1007/s003910070030.
- 20.↵Zheng H, Li M, Luo Y, A retrospective study of contributing factors for prognosis and survival length of cryptococcal meningoencephalitis in Southern part of China (1998–2013). BMC Infect Dis. 2015;15(1):77. doi:10.1186/s12879-015-0826-y.
- 21.↵Pourgheysari B, Khan N, Best D, Bruton R, Nayak L, Moss PA. The cytomegalovirus-specific CD4+ T-cell response expands with age and markedly alters the CD4+ T-cell repertoire. J Virol. 2007;81(14):7759-7765. doi:10.1128/JVI.01262-06
- 22.↵Thwaites GE, Chau TTH, Stepniewska K, Diagnosis of adult tuberculous meningitis by use of clinical and laboratory features. Lancet. 2002;360(9342):1287-1292. doi:10.1016/S0140-6736(02)11318-3.
- 23.Yang Y, Qu XH, Zhang KN, A Diagnostic Formula for Discrimination of Tuberculous and Bacterial Meningitis Using Clinical and Laboratory Features. Front Cell Infect Microbiol. 2020;9:448. doi:10.3389/fcimb.2019.00448.
- 24.↵Marais S, Pepper DJ, Schutz C, Wilkinson RJ, Meintjes G. Presentation and outcome of tuberculous meningitis in a high HIV prevalence setting. PLoS One. 2011;6(5):e20077. doi:10.1371/journal.pone.0020077.
- 25.↵Sütcüoğlu O, Akdoğan O, Gürler F, The role of serum albumin/globulin ratio in combination with prognostic risk indexes of febrile neutropenia. Int J Clin Pract. 2021;75(7):e14185. doi:10.1111/ijcp.14185.
- 26.Zhang J, Wang T, Fang Y, Clinical Significance of Serum Albumin/Globulin Ratio in Patients With Pyogenic Liver Abscess. Front Surg. 2021;8:677799. doi:10.3389/fsurg.2021.677799.
- 27.↵Wang H, Zhou H, Jiang R, Qian Z, Wang F, Cao L. Globulin, the albumin-to-globulin ratio, and fibrinogen perform well in the diagnosis of Periprosthetic joint infection. BMC Musculoskelet Disord. 2021;22(1):583. doi:10.1186/s12891-021-04463-7.
- 28.Mirsaeidi M, Omar HR, Sweiss N. Hypoalbuminemia is related to inflammation rather than malnutrition in sarcoidosis. Eur J Intern Med. 2018;53:e14-e16. doi:10.1016/j.ejim.2018.04.016.
- 29.↵Wu PP, Hsieh YP, Kor CT, Chiu PF. Association between Albumin–Globulin Ratio and Mortality in Patients with Chronic Kidney Disease. J Clin Med. 2019;8(11):1991. doi:10.3390/jcm8111991.
- 30.↵Marais S, Meintjes G, Lesosky M, Wilkinson KA, Wilkinson RJ. Interleukin-17 mediated differences in the pathogenesis of HIV-1-associated tuberculous and cryptococcal meningitis. AIDS. 2016;30(3):395-404. doi:10.1097/QAD.0000000000000904
- 31.↵Jain SK, Paul-Satyaseela M, Lamichhane G, Kim KS, Bishai WR. Mycobacterium tuberculosis invasion and traversal across an in vitro human blood-brain barrier as a pathogenic mechanism for central nervous system tuberculosis. J Infect Dis. 2006;193(9):1287-1295. doi:10.1086/502631.
- 32.↵Gonzales Zamora JA, Espinoza LA, Nwanyanwu RN. Neurosyphilis with Concomitant Cryptococcal and Tuberculous Meningitis in a Patient with AIDS: Report of a Unique Case. Case Rep Infect Dis. 2017;2017:1-5. doi:10.1155/2017/4103858.
- 33.↵Hoen B, Varon E, Debroucker T, ; Expert and reviewing group. Management of acute community-acquired bacterial meningitis (excluding newborns). Short text. Med Mal Infect. 2019;49(6):367-398. doi:10.1016/j.medmal.2019.03.008.
- 34.↵Liu Q, Gao Y, Zhang B, Cytokine profiles in cerebrospinal fluid of patients with meningitis at a tertiary general hospital in China. J Microbiol Immunol Infect. 2020;53(2):216-224. doi:10.1016/j.jmii.2018.08.019.
- 35.↵Zhang C, Tan Z, Tian F. Impaired consciousness and decreased glucose concentration of CSF as prognostic factors in immunocompetent patients with cryptococcal meningitis. BMC Infect Dis. 2020;20(1):69. doi:10.1186/s12879-020-4794-5.
- 36.↵Knutsson L, Seidemo A, Scherman AR, Arterial Input Functions and Tissue Response Curves in Dynamic Glucose-Enhanced (DGE) Imaging: Comparison between glucoCEST and Blood Glucose Sampling in Humans. Tomography. 2018;4(4):164-171. doi:10.18383/j.tom.2018.00025.
