Evaluating the diagnostic performance of non-contrast magnetic resonance angiography sequences in the pre-procedural comprehensive analysis of direct carotid cavernous fistula.

PURPOSE: Endovascular treatment of direct carotid cavernous fistula (DCCF) requires invasive diagnostic cerebral angiography for diagnosis and planning; however, a less invasive modality like magnetic resonance angiography (MRA) can be useful, especially in high-risk cases. This single-centre study evaluated a newer MR angiography (MRA) sequence, silent MRA and the traditional time of flight (TOF) MRA for pre-procedural treatment planning of DCCF. METHODS: All consecutive DCCF patients who underwent TOF, silent MRA and diagnostic cerebral angiography were included in the study. Angiographic features like rent size, location, draining veins and collateral communicating arteries were analysed and compared between the two MRA sequences, with digital subtraction angiography (DSA) as the gold standard. RESULTS: Fifteen patients were included in the study. TOF MRA exhibited better sensitivity (76.9% vs 69.2%) in identifying the rent location, correctly pinpointing the location in 93.3% compared to 73.3% with silent MRA. Both MRA sequences showed good agreement with DSA for primary sac and rent size. TOF MRA correctly identified 86.2% of 210 total venous structures compared to 96% by silent MRA. Silent MRA demonstrated higher sensitivity (90% vs 76%) and accuracy (87.69 vs 94.36) in visualisation of involved veins compared to TOF MRA. CONCLUSION: Arterial characteristics of DCCF like rent location and rent size were better assessed by TOF MRA. Although both MRA identified venous features, silent MRA correlated better with DSA irrespective of the size and proximity to the site of the fistula. Combining both sequences can evaluate various angioarchitectural features of DCCF useful for therapeutic planning.

Comparative study of non-contrast silent and time-of-flight magnetic resonance angiographic sequences in the evaluation of intracranial dural arteriovenous fistula.

AIM: To compare the performance of two non-contrast magnetic resonance angiography (MRA) sequences, silent MRA and time of flight (TOF) MRA, in the evaluation of intracranial dural arteriovenous fistula (DAVF). MATERIALS AND METHODS: Forty consecutive patients with DAVF were enrolled and evaluated prospectively using silent MRA, TOF MRA, and digital subtraction angiography (DSA). The location, Cognard classification, arterial feeders, and venous drainage were evaluated. The therapeutic strategy and possible route were predicted on both silent and TOF MRA and these were compared with DSA during subsequent endovascular treatment. RESULTS: Sensitivity and accuracy of silent and TOF MRA for localisation (96.4% versus 96% and 96% versus 95%, respectively) and classification (96% versus 94% and 96% versus 93.5%, respectively) were high. Silent MRA showed higher sensitivity than TOF MRA for arterial feeders and draining veins (87% versus 79% and 81.6% versus 67%). This improved to a sensitivity of 96.4% and 89% when prominent feeders were considered. The sensitivity and accuracy were 92.6% and 85.8% for immediate draining veins. Both silent and TOF MRA were accurate for therapeutic planning (96% versus 85%), although silent MRA was more accurate. CONCLUSION: Silent MRA can more reliably evaluate the various angioarchtectural components of DAVF compared to TOF MRA.

Utility of zero echo time (ZTE) sequence for assessing bony lesions of skull base and calvarium.

BACKGROUND: The emergence of zero echo time (ZTE) imaging has transformed bone imaging, overcoming historical limitations in capturing detailed bone structures. By minimizing the time gap between radiofrequency excitation and data acquisition, ZTE generates CT-like images. While ZTE has shown promise in various applications, its potential in assessing skull base and calvarium lesions remains unexplored. Hence we aim to introduce a novel perspective by investigating the utility of inverted ZTE images (iZTE) and pseudoCT (pCT) images for studying lesions in the skull base and calvarium. MATERIALS AND METHODS: A total of 35 eligible patients, with an average age of 42 years and a male/female ratio of 1:4, underwent ZTE MRI and images are processed to generate iZTE and pCT images were generated through a series of steps including intensity equalization, thresholding, and deep learning-based pCT generation. These images were then compared to CT scans using a rating scale; inter-rater kappa coefficient evaluated observer consensus while statistical metrics like sensitivity and specificity assessed their performance in capturing bone-related characteristics. RESULTS: The study revealed excellent interobserver agreement for lesion assessment using both pCT and iZTE imaging modalities, with kappa coefficient of 0.91 (P < 0.0001) and 0.92 respectively (P < 0.0001). Also, pCT and iZTE accurately predicted various lesion characteristics with sensitivity ranging from 84.3% to 95.1% and 82.6%-94.2% (95% CI) with a diagnostic accuracy of 95.56% and 94.44% respectively. Although both of them encountered challenges with ground glassing, hyperostosis, and intralesional bony fragments, they showed good performance in other bony lesion assessments. CONCLUSIONS: The pilot study suggests strong potential for integrating the ZTE imaging into standard care for skull base and calvarial bony lesions assessment. Additionally, larger-scale studies are needed for comprehensive assessment of its efficacy.

High-resolution MR vessel wall imaging in determining the stroke aetiology and risk stratification in isolated middle cerebral artery disease.

PURPOSE: High-resolution MR vessel wall imaging (HRVWI) can characterise vessel wall pathology affecting intracranial circulation and helps in differentiating intracranial vasculopathies. The aim was to differentiate intracranial pathologies involving middle cerebral artery (MCA) in patients with ischemic stroke and characterise the high-risk plaques in intracranial atherosclerotic disease (ICAD) using HRVWI. METHODS: Patients with ischemic stroke with isolated MCA disease with ≥ 50% luminal narrowing by vascular imaging were enrolled within 2 weeks of onset and underwent high-resolution (3 T) intracranial vessel wall imaging (VWI). The pattern of vessel wall thickening, high signal on T1-weighted images, juxtaluminal hyperintensity, pattern and grade of enhancement were studied. The TOAST classification before and after HRVWI and the correlation of the recurrence of ischemic events at 3 months with imaging characteristics were analysed. RESULTS: Of the 36 patients, the mean age was 49.53 ± 15.61 years. After luminal imaging, by TOAST classification, 12 of 36 patients had stroke of undetermined aetiology. After vessel wall imaging, lesions in MCA were analysed. Of them, 23 patients had ICAD, 8 had vasculitis, and 2 had partially occlusive thrombus in MCA. The ability of HRVWI to bring a change in diagnosis was significant (p = 0.031). Of the 23 patients with ICAD, 12 patients had recurrent strokes within 3 months. The presence of grade 2 contrast enhancement (p = 0.02) and type 2 wall thickening (p = 0.03) showed a statistically significant association with recurrent ischemic events. CONCLUSION: High-resolution MRVWI can help in identifying the aetiology of stroke. The HRVWI characteristics in ICAD can help in risk stratification.

An optimal brain tumor segmentation algorithm for clinical MRI dataset with low resolution and non-contiguous slices.

BACKGROUND: Segmenting brain tumor and its constituent regions from magnetic resonance images (MRI) is important for planning diagnosis and treatment. In clinical routine often an experienced radiologist delineates the tumor regions using multimodal MRI. But this manual segmentation is prone to poor reproducibility and is time consuming. Also, routine clinical scans are usually of low resolution. To overcome these limitations an automated and precise segmentation algorithm based on computer vision is needed. METHODS: We investigated the performance of three widely used segmentation methods namely region growing, fuzzy C means and deep neural networks (deepmedic). We evaluated these algorithms on the BRATS 2018 dataset by choosing randomly 48 patients data (high grade, n = 24 and low grade, n = 24) and on our routine clinical MRI brain tumor dataset (high grade, n = 15 and low grade, n = 28). We measured their performance using dice similarity coefficient, Hausdorff distance and volume measures. RESULTS: Region growing method performed very poorly when compared to fuzzy C means (fcm) and deepmedic network. Dice similarity coefficient scores for FCM and deepmedic algorithms were close to each other for BRATS and clinical dataset. The accuracy was below 70% for both these methods in general. CONCLUSION: Even though the deepmedic network showed very high accuracy in BRATS challenge for brain tumor segmentation, it has to be custom trained for the low resolution routine clinical scans. It also requires large training data to be used as a stand-alone algorithm for clinical applications. Nevertheless deepmedic may be a better algorithm for brain tumor segmentation when compared to region growing or FCM.

Utility of silent magnetic resonance angiography in the evaluation and characterisation of intracranial dural arteriovenous fistula.

AIM: To evaluate the utility of silent magnetic resonance angiography (MRA) in the diagnosis, characterisation, and therapeutic planning of intracranial dural arteriovenous fistula (DAVF). MATERIALS AND METHODS: Twenty consecutive patients with DAVF were enrolled prospectively and were evaluated using silent MRA and digital subtraction angiography (DSA) as a part of routine work-up. The diagnosis and location of fistula, Borden and Cognard classification, entire arterial feeders, and venous drainage were analysed. A therapeutic strategy was formulated, and the accessible route and vessel were predicted, which was confirmed on endovascular treatment. RESULTS: Silent MRA was 100% sensitive and accurate for location and classification of fistulas. Silent MRA showed a sensitivity of 82% and 76.5% for entire arterial feeders and draining veins, which improved to a sensitivity of 90% and 94% when prominent feeders and immediate venous drainage was considered. Among the missed veins, thrombosed sinus, slow sinus flow, small calibre, reduced image quality were the causes. The therapeutic decision matched with DSA in all cases and silent MRA accurately identified the potential accessible feeder in 94% cases. CONCLUSION: Silent MRA is a promising MR technique that can provide both diagnostic and therapeutic information similar to that obtained from DSA.

Resting fMRI as an alternative for task-based fMRI for language lateralization in temporal lobe epilepsy patients: a study using independent component analysis.

PURPOSE: Our aim is to investigate whether rs-fMRI can be used as an effective technique to study language lateralization. We aim to find out the most appropriate language network among different networks identified using ICA. METHODS: Fifteen healthy right-handed subjects, sixteen left, and sixteen right temporal lobe epilepsy patients prospectively underwent MR scanning in 3T MRI (GE Discovery™ MR750w), using optimized imaging protocol. We obtained task-fMRI data using a visual-verb generation paradigm. Rs-fMRI and language-fMRI analysis were conducted using FSL software. Independent component analysis (ICA) was used to estimate rs-fMRI networks. Dice coefficient was calculated to examine the similarity in activated voxels of a common language template and the rs-fMRI language networks. Laterality index (LI) was calculated from the task-based language activation and rs-fMRI language network, for a range of LI thresholds at different z scores. RESULTS: Measurement of hemispheric language dominance with rs-fMRI was highly concordant with task-fMRI results. Among the evaluated z scores for a range of LI thresholds, rs-fMRI yielded a maximum accuracy of 95%, a sensitivity of 83%, and specificity of 92.8% for z = 2 at 0.05 LI threshold. CONCLUSION: The present study suggests that rs-fMRI networks obtained using ICA technique can be used as an alternative for task-fMRI language laterality. The novel aspect of the work is suggestive of optimal thresholds while applying rs-fMRI, is an important endeavor given that many patients with epilepsy have co-morbid cognitive deficits. Thus, an accurate method to determine language laterality without requiring a patient to complete the language task would be advantageous.

Structural correlates of mild cognitive impairment: A clinicovolumetric study.

CONTEXT:: Annually 10-12% of patients with mild cognitive impairment (MCI) are likely to progress to Alzheimer's Disease (AD). The morphometric profile in stable non-converters has not been adequately characterized. AIMS: To determine the structural differences between amnestic MCI and early AD using volumetric magnetic resonance imaging (MRI) and its correlation with neuropsychological test performances. SETTINGS AND DESIGN:: This was a hospital-based case-control study. MATERIALS AND METHODS:: Twenty-four patients classified as having "non-progressor" MCI, 13 as having an early AD, and 25 controls, and assessed using neuropsychological evaluation, and three-dimensional T1-weighted 1.5T magnetic resonance maging (MRI) were included in the study. We used both voxel-based morphometry and automated regional volumetry to assess the topographical patterns of volume loss. STATISTICAL ANALYSIS USED: Post-hoc analysis of variance was done for comparison between means, and partial correlation analysis was done for correlating volumetric and cognitive measures. RESULTS:: Consistently, significant atrophy of the superior temporal gyrus, left hippocampus, and mesial frontoparietal regions were identified in patients with MCI in comparison to controls. Increased atrophy in the limbic regions, temporal neocortex, and precuneus was identified in patients with early AD in comparison to patients with MCI. While differences in retention and recall scores between the groups were independent of age and volumetric variables, significant correlations were observed between the learning and recall scores and the volume of hippocampus in patients with MCI as well as temporal neocortex in patients with AD. Atrophy of the superior temporal gyrus and mesial neocortical regions represents the structural correlate of amnestic MCI parallel to the development of hippocampal atrophy. CONCLUSIONS:: Identification of the pattern of volumetric abnormalities in patients with amnestic MCI in addition to atrophy of the medial temporal lobes necessitates a close follow up to continuously assess these patients for their progression to early AD.

Reduced brain volumes in children of women with epilepsy: A neuropsychological and voxel based morphometric analysis in pre-adolescent children.

BACKGROUND AND PURPOSE: Children of Women with Epilepsy with antenatal exposure to antiepileptic drugs (CAED) have reduced neuropsychological functions. We aimed to explore the anatomical basis for this impairment by comparing the brain volumes of CAED with that of matched healthy children without antenatal AED exposure (COAED). MATERIALS AND METHODS: CAED aged 8-12 years were recruited from the Kerala Registry of epilepsy and pregnancy that prospectively follows up children of women with epilepsy and COAED from children attending the imaging department for minor illnesses. Maternal clinical details and the neuropsychological data including IQ of CAED and COAED were obtained. Total intracranial volume (TBV), grey matter volume (GMV), white matter volume (WMV) and volumes of deep grey matter were measured by Voxel Based Morphometry. RESULTS: We studied 30 CAED (mean age 10.8+1.11 years) and 35 COAED (mean age 10.64+1.26). The antenatal AED exposure for the CAED was monotherapy for 8 children and polytherapy for 22 children. The CAED had significantly lower (P<0.001) IQ (77.5+13.8), TBV(1259.55±169.85mL) and GMV (672.51±85.42 mL) compared to the IQ (87.0+13.5), TBV(1405.37±161mL) and GMV (745.427±86.69 mL) of COAED. CAED had lower volumes for Lt Inferior Triangular Gyrus, and hippocampi on both sides, when compared to COAED. Group analysis CAED showed less GMV (P<0.05) for left inferior and middle frontal gyri relative to COAED. CONCLUSIONS: These observations point towards an anatomical basis of lower GMV for the lower neuropsychological functions in children with antenatal AED exposure.

Is 'burned-out hippocampus' syndrome a distinct electro-clinical variant of MTLE-HS syndrome?

AIM: To study the clinical, electrophysiological and imaging characteristics of patients with unilateral mesial temporal lobe epilepsy (MTLE) with contralateral ictal onset on scalp EEG, viz. 'burned-out hippocampus' syndrome (MTLE-BHS). METHODS: MTLE-BHS was defined as TLE with unilateral hippocampal sclerosis (HS) without any dual pathology on MRI and contralateral ictal onset on scalp EEG, unlike in classical hippocampal sclerosis (HS). Consecutive "MTLE-BHS" patients evaluated at our Centre for Comprehensive Epilepsy Care from January 2005 to July 2014 were studied. Twenty-five cases of classic MTLE-HS operated during the same period were also analyzed for comparison. RESULTS: Seventeen patients were diagnosed to have MTLE-BHS. Mean age of seizure onset was 9.5±7.7years and the mean duration of epilepsy was18.2±7.3years. Epigastric aura was more common in MTLE-HS and fear, secondary generalized seizures and temporal polar changes on MRI were more prevalent in the MTLE-BHS subgroup. In the latter group, five (29%) exhibited seizure semiology and 2 (12%) had interictal discharges discordant to the side of MTS. Eight (47%) patients in the MTLE-BHS sub-group had normal medial temporal volume on Scheltens scale. Eight patients among MTLE-BHS underwent surgery (4 following intracranial monitoring that localized to the side of HS) with Engel class I outcome at 1year follow-up in 6 and Engel class II outcome in 2. CONCLUSION: Attenuation of ipsilateral fast ictal rhythms on scalp EEG as well as neocortical changes are likely to be deterministic factors for MTLE-BHS as opposed to the severity of hippocampal atrophy. Considering good post-operative outcomes, intracranial monitoring for surgical selection is not mandatory in MTLE-BHS despite discordant semiology and ictal onset, in the presence of inter-ictal, functional imaging and neuropsychology data concordant to the side of HS.

Quantification of diffusion and anisotropy in intracranial epidermoids using diffusion tensor metrics and p: q tensor decomposition.

PURPOSE: To quantitatively evaluate the diffusion tensor metrics p, q, L and fractional anisotropy in intracranial epidermoids in comparison with normal white matter in the splenium of the corpus callosum. METHODS: This retrospective study included 20 consecutive patients referred to our institute. All patients had a magnetic resonance imaging (MRI) study on a 1.5-Tesla MR system. A spin-echo echo-planar DTI sequence with diffusion gradients along 30 non-collinear directions was performed. The eigen values (λ1, λ2, λ3) were computed for each voxel and, using p: q tensor decomposition, the DTI metrics p, q and L-values and fractional anositropy (FA) were calculated. The region of interest (ROI) (6 pixels each) was placed within the lesion in all the cases and in the splenium of the corpus callosum. RESULTS: The mean FA in the lesion and splenium were 0.50 and 0.88 respectively, with a statistically significant difference between them (P<0.01). On p: q tensor decomposition, the mean p-value in the epidermoid was 1.55±0.24 and 1.35±0.20 in the splenium; the mean q-values in the epidermoid was 0.67±0.13 and 1.27±0.17 in the splenium; the differences were statistically significant (P=0.01 and <0.01 respectively). The significant difference between p- and q-values in epidermoids compared with the splenium of callosum was probably due to structural and orientation differences in the keratin flakes in epidermoids and white matter bundles in the callosum. However, no significant statistical difference in L-values was noted (P=0.44). CONCLUSION: DTI metrics p and q have the potential to quantify the diffusion and anisotropy in various tissues thereby gaining information about their internal architecture. The results also suggest that significant differences of DTI metrics p and q between epidermoid and the splenium of the corpus callosum are due to the difference in structural organization within them.

Varying clinical and imaging outcomes in patients with spontaneous thrombosis of vein of Galen malformation--a report of two cases.

BACKGROUND: Vein of Galen malformations are rare congenital intracranial vascular malformations. The pathophysiological consequences usually manifest as high-output cardiac failure and neurological symptoms secondary to cerebral venous congestion and abnormal CSF flow. Management of these patients is complex. Indications and time to intervene are decided depending upon the age of child, clinical presentation, and multisystemic neonatal scores. Many of these children presenting early and who are not in cardiac failure are expectantly followed up. Spontaneous thrombosis of the vein of Galen malformation is a rare occurrence in these as such rare vascular malformations. METHODS: This review reports two cases of spontaneous thrombosis of the vein of Galen malformations with varying clinical as well as imaging outcomes. Possible pathophysiologic mechanisms are discussed with review of literature. RESULTS: This report highlights the importance of vigilant clinical and imaging follow-up even after spontaneous thrombosis in the vein of Galen malformations.

Correlation between anatomic landmarks and fMRI in detection of the sensorimotor cortex in patients with structural lesions.

BACKGROUND: Structural lesions in/near the sensorimotor cortex may cause distortion/obscuration of the anatomic landmarks. PURPOSE: To compare the localization of the sensorimotor cortex using anatomical landmarks and fMRI in the clinical setting in patients with structural lesions in/near the central sulcus. MATERIAL AND METHODS: We analyzed the anatomic and fMRI data of 68 consecutive patients (42 tumors, 15 gliotic lesions, 11 focal cortical dysplasias [FCD]) who underwent MRI to assess the relationship of these lesions to the sensorimotor cortex. Anatomical data was analyzed on conventional two- and three-dimensional sequences. BOLD fMRI was performed with block design hand/leg or lip movement paradigm and general linear model was used for detecting the activated cortex. fMRI was considered as a valid method for identifying the sensorimotor cortex based on previously reported literature. RESULTS: The sensorimotor cortex could not be identified with anatomical landmarks in 9/68 (13.2%) patients. fMRI detected activation in areas different from that predicted by anatomical landmarks in 11/68 (16.2%) cases. This occurred in 5/42 (11.9%) tumors, 6/15 (40%) gliotic lesions, and 0/11 (0%) FCDs. The kappa value for concordance between fMRI and anatomic landmarks was 0.883 overall, 1.0 for tumors, 0.721 for gliotic lesions, and in none of the patients with focal cortical dysplasias. CONCLUSION: In patients with lesions that obscure normal cerebral landmarks, fMRI supplies the information that is not available from the anatomic images. In patients with landmarks that can be recognized, the location of the rolandic cortex may be misjudged in some cases if functional imaging is not used. Anatomic landmarks may not correlate with the area of functional activation in gliotic lesions and tumors. Determining the risk of a postoperative neurologic defect from surgery is likely to be more reliable with functional imaging than with conventional anatomic imaging.

Accurate low and high grade glioma classification using free water eliminated diffusion tensor metrics and ensemble machine learning.

Glioma, a predominant type of brain tumor, can be fatal. This necessitates an early diagnosis and effective treatment strategies. Current diagnosis is based on biopsy, prompting the need for non invasive neuroimaging alternatives. Diffusion tensor imaging (DTI) is a promising method for studying the pathophysiological impact of tumors on white matter (WM) tissue. Single-shell DTI studies in brain glioma patients have not accounted for free water (FW) contamination due to tumors. This study aimed to (a) assess the efficacy of a two-compartment DTI model that accounts for FW contamination and (b) identify DTI-based biomarkers to classify low-grade glioma (LGG) and high-grade glioma (HGG) patients. DTI data from 86 patients (LGG n = 39, HGG n = 47) were obtained using a routine clinical imaging protocol. DTI metrics of tumorous regions and normal-appearing white matter (NAWM) were evaluated. Advanced stacked-based ensemble learning was employed to classify LGG and HGG patients using both single- and two-compartment DTI model measures. The DTI metrics of the two-compartment model outperformed those of the standard single-compartment DTI model in terms of sensitivity, specificity, and area under the curve of receiver operating characteristic (AUC-ROC) score in classifying LGG and HGG patients. Four features (out of 16 features), namely fractional anisotropy (FA) of the edema and core region and FA and mean diffusivity of the NAWM region, showed superior performance (sensitivity = 92%, specificity = 90%, and AUC-ROC = 90%) in classifying LGG and HGG. This demonstrates that both tumorous and NAWM regions may be differentially affected in LGG and HGG patients. Our results demonstrate the significance of using a two-compartment DTI model that accounts for FW contamination by improving diagnostic accuracy. This improvement may eventually aid in planning treatment strategies for glioma patients.

Brain-computer interfaces for neurorehabilitation.

Brain-computer interfaces (BCIs) enable control of computers and other assistive devices, such as neuro-prostheses, which are used for communication, movement restoration, neuro-modulation, and muscle stimulation, by using only signals measured directly from the brain. A BCI creates a new output channel for the brain to a computer or a device. This requires retrieval of signals of interest from the brain, and its use for neuro-rehabilitation by means of interfacing the signals to a computerized device. Brain signals such as action potentials from single neurons or nerve fibers, extracellular local field potentials (LFPs), electrocorticograms, electroencephalogram and its components such as the event-related brain potentials, real-time functional magnetic resonance imaging, near-infrared spectroscopy, and magneto-encephalogram have been used. BCIs are envisaged to be useful for communication, control and self-regulation of brain function. BCIs employ neurofeedback to enable operant conditioning to allow the user to learn using it. Paralytic conditions arising from stroke or other diseases are being targeted for BCI application. Neurofeedback strategies ranging from sensory feedback to direct brain stimulation are being employed. Existing BCIs are limited in their throughput in terms of letters per minute or commands per minute, and need extensive training to use the BCI. Further, they can cause rapid fatigue due to use and have limited adaptability to changes in the patient's brain state. The challenge before BCI technology for neuro-rehabilitation today is to enable effective clinical use of BCIs with minimal effort to set up and operate.

Susceptibility weighted imaging in the evaluation of movement disorders.

Movement disorders are neurodegenerative disorders associated with abnormalities of brain iron deposition. In this presentation, we aim to describe the role of susceptibility weighted imaging (SWI) in the imaging of patients with movement disorders and differentiate between the various disorders. SWI is a high-resolution, fully velocity-encoded gradient-echo magnetic resonance imaging (MRI) sequence that consists of using both magnitude and phase information. We describe briefly the physics behind this sequence and the post-processing techniques used. The anatomy of the midbrain and basal ganglia in normal subjects on SWI is covered. A number of neurodegenerative disorders are associated with abnormal iron deposition, which can be detected due to the susceptibility effects.

Fractal dimension: analyzing its potential as a neuroimaging biomarker for brain tumor diagnosis using machine learning.

Purpose: The main purpose of this study was to comprehensively investigate the potential of fractal dimension (FD) measures in discriminating brain gliomas into low-grade glioma (LGG) and high-grade glioma (HGG) by examining tumor constituents and non-tumorous gray matter (GM) and white matter (WM) regions. Methods: Retrospective magnetic resonance imaging (MRI) data of 42 glioma patients (LGG, n = 27 and HGG, n = 15) were used in this study. Using MRI, we calculated different FD measures based on the general structure, boundary, and skeleton aspects of the tumorous and non-tumorous brain GM and WM regions. Texture features, namely, angular second moment, contrast, inverse difference moment, correlation, and entropy, were also measured in the tumorous and non-tumorous regions. The efficacy of FD features was assessed by comparing them with texture features. Statistical inference and machine learning approaches were used on the aforementioned measures to distinguish LGG and HGG patients. Results: FD measures from tumorous and non-tumorous regions were able to distinguish LGG and HGG patients. Among the 15 different FD measures, the general structure FD values of enhanced tumor regions yielded high accuracy (93%), sensitivity (97%), specificity (98%), and area under the receiver operating characteristic curve (AUC) score (98%). Non-tumorous GM skeleton FD values also yielded good accuracy (83.3%), sensitivity (100%), specificity (60%), and AUC score (80%) in classifying the tumor grades. These measures were also found to be significantly (p < 0.05) different between LGG and HGG patients. On the other hand, among the 25 texture features, enhanced tumor region features, namely, contrast, correlation, and entropy, revealed significant differences between LGG and HGG. In machine learning, the enhanced tumor region texture features yielded high accuracy, sensitivity, specificity, and AUC score. Conclusion: A comparison between texture and FD features revealed that FD analysis on different aspects of the tumorous and non-tumorous components not only distinguished LGG and HGG patients with high statistical significance and classification accuracy but also provided better insights into glioma grade classification. Therefore, FD features can serve as potential neuroimaging biomarkers for glioma.

Experimenting with ASL-based arterialized cerebral blood volume as a novel imaging biomarker in grading glial neoplasms.

BACKGROUND: Perfusion imaging is one of the methods used to grade glial neoplasms, and in this study we evaluated the role of ASL perfusion in grading brain glioma. PURPOSE: The aim is to evaluate the role of arterialized cerebral blood volume (aCBV) of multi-delay ASL perfusion for grading glial neoplasm. MATERIALS AND METHODS: This study is a prospective observational study of 56 patients with glial neoplasms of the brain who underwent surgery, and only cases with positive diagnosis of glioma are included to evaluate the novel diagnostic parameter. RESULTS: In the study, ASL-derived normalized aCBV (naCBV) and T2*DSC-derived normalized CBV (nCBV) are showing very high correlation (Pearson's correlation coefficient value of 0.94) in grading glial neoplasms. naCBV and nCBF are also showing very high correlation (Pearson's correlation coefficient value of 0.876). The study also provides cutoff values for differentiating LGG from HGG for normalized aCBV(naCBV) of ASL, normalized CBV (nCBV), and normalized nCBF derived from T2* DCS as 1.12, 1.254, and 1.31, respectively. ASL-derived aCBV also shows better diagnostic accuracy than ASL-derived CBF. CONCLUSION: This study is one of its kind to the best of our knowledge where multi-delay ASL perfusion-derived aCBV is used as a novel imaging biomarker for grading glial neoplasms, and it has shown high statistical correlation with T2* DSC-derived perfusion parameters.