MRI Volumes Research Articles

Brain Age Estimation on a Dementia Cohort Using FLAIR MRI Biomarkers.

The prodromal stage of Alzheimer's disease presents an imperative intervention window. This work focuses on using brain age prediction models and biomarkers from FLAIR MR imaging to identify subjects who progress to Alzheimer's disease (converting mild cognitive impairment) or those who remain stable (stable mild cognitive impairment). A machine learning model was trained to predict the age of normal control subjects on the basis of volume, intensity, and texture features from 3239 FLAIR MRI volumes. The brain age gap estimation (BrainAGE) was computed as the difference between the predicted and true age, and it was used as a biomarker for both cross-sectional and longitudinal analyses. Differences in biomarker means, slopes, and intercepts were investigated using ANOVA and Tukey post hoc test. Correlation analysis was performed between brain age gap estimation and established Alzheimer's disease indicators. The brain age prediction model showed accurate results (mean absolute error = 2.46 years) when testing on held out normal control data. The computed BrainAGE metric showed significant differences between the stable mild cognitive impairment and converting mild cognitive impairment groups in cross-sectional and longitudinal analyses, most notably showing significant differences up to 4 years before conversion to Alzheimer's disease. A significant correlation was found between BrainAGE and previously established Alzheimer's disease conversion biomarkers. The BrainAGE metric can allow clinicians to consider a single explainable value that summarizes all the biomarkers because it considers many dimensions of disease and can determine whether the subject has normal aging patterns or if he or she is trending into a high-risk category using a single value.

How does it feel? An exploration of neurobiological and clinical correlates of alexithymia in trauma-exposed police-officers with and without PTSD

ABSTRACT Background: Alexithymia, an inability to recognise one’s emotions, has been associated with trauma-exposure and posttraumatic stress disorder (PTSD). Previous research suggests involvement of the oxytocin system, and socio-emotional neural processes. However, the paucity of neurobiological research on alexithymia, particularly in trauma-exposed populations, warrants further investigation. Objective: Explore associations between alexithymia, endogenous oxytocin levels, and socio-emotional brain function and morphometry in a trauma-exposed sample. Method: Dutch trauma-exposed police officers with (n = 38; 18 females) and without PTSD (n = 40; 20 females) were included. Alexithymia was assessed with the Toronto Alexithymia Scale (TAS-20). Endogenous salivary oxytocin was assessed during rest, using radioimmunoassay. Amygdala and insula reactivity to socio-emotional stimuli were assessed with functional MRI, amygdala and insula grey matter volume were derived using Freesurfer. Results: Alexithymia was higher in PTSD patients compared to trauma-exposed controls (F(1,70) = 54.031, p < .001). Within PTSD patients, alexithymia was positively associated with PTSD severity (ρ(36) = 0.497, p = .002). Alexithymia was not associated with childhood trauma exposure (β = 0.076, p = .509), police work-related trauma exposure (β = −0.107, p = .355), oxytocin levels (β = −0.164, p = .161), insula (β = −0.170, p = .158) or amygdala (β = −0.175, p = .135) reactivity, or amygdala volume (β = 0.146, p = .209). Insula volume was positively associated with alexithymia (β = 0.222, p = .016), though not significant after multiple testing corrections. Bayesian analyses supported a lack of associations. Conclusions: No convincing neurobiological correlates of alexithymia were observed with any of the markers included in the current study. Yet, the current study confirmed high levels of alexithymia in PTSD patients, independent of trauma-exposure, substantiating alexithymia’s relevance in the clinical phenotype of PTSD.

Robust U-Net-Based Approach for Accurate Brain Tumor Segmentation Using Multimodal MRI Data

Detecting and quantifying the extent of brain tumors poses a formidable challenge in medical centers. Magnetic Resonance Imaging (MRI) has developed as a non-invasive brain cancers' primary diagnostic tool, offering the crucial advantage of avoiding ionizing radiation. Brain tumor manually segmented boundaries within 3D MRI volumes is an exceedingly time-intensive task, heavily reliant on operator expertise. Among brain tumors, gliomas stand out as the prevalent and highly malignant, significantly impacting patients' life expectancy, particularly at their highest grade. Recognizing the pressing need for a reliable, completely automatic segmentation technique to efficiently assess tumor extent, this study introduces a robust approach. A completely automated brain tumor segmentation method is proposed, leveraging U-Net-based deep convolutional networks. This approach underwent rigorous evaluation on the Multimodal Brain Tumor Image Segmentation BraTS-19 dataset a widely recognized medical image analysis dataset featuring multimodal MRI scans of brain tumors, including glioblastoma, anaplastic astrocytoma, and lower-grade glioma, coupled with corresponding manual tumor segmentations. This dataset serves as a pivotal resource for advancing automatic brain tumor segmentation techniques and assessing their performance using metrics like the Dice score, which achieved 92% for entire tumor. Cross-validation results affirm the efficiency and promise of our method in achieving accurate segmentation.

NCMP-15. AMYLOID-BETA AND TAU-PATHOLOGY IN THE GLIOMA VICINITY

Abstract With aging societies the incidences of brain cancer and neurodegenerative disease are increasing. Although biological associations were suggested between glioma and Alzheimer's disease (AD), the rate of comorbidity is ill-defined. We here screened the tumor-adjacent cortex of a representative patient cohort for tau and Amyloid-beta deposits and correlated the results with imaging biomarkers. A total of 290 patients (37 astrocytoma, IDH-mutant, CNS-WHO grades 2-4, median-age: 39; 48 oligodendroglioma, IDH-mutant and 1p-19q-co-deleted, CNS-WHO grades 2-3, median-age: 50; 205 glioblastoma, IDH-wildtype, median-age: 65) were included. Whole-slide quantifications were performed using a digital pathology pipeline based on Matlab and QuPath codes. For a subset of 48 patients with glioblastoma 280 longitudinal MRI volumes were available for segmentation and volumetric analysis of tumor ipsi- and contralateral hippocampuses. Overall, 160 patients (55%) displayed Amyloid-beta and/or tau-pathology in tumor-adjacent cortex. AD protein deposits correlated significantly with age (p&amp;lt; 0.001), and showed topographic variability across brain regions. Amyloid-beta plaques were commonly observed in the occipital cortex in glioblastoma (50%, n= 10/20), while tau-pathology was commonest in the temporal lobe in both IDH-mutant and -wildtype tumors (71%, n= 10/14; 56% n= 44/79). Remarkably, isolated tau-pathology was more prevalent in IDH-mutant glioma (48% vs. 18%, p&amp;lt; 0.001, median age: 45). Median cell density in tumor-infiltrated cortex was highest in glioblastoma (1895 cells/mm²), and associated with reduced neuronal counts and AD protein deposits (p&amp;lt; 0.001). In glioblastoma, hippocampal atrophy correlated with AD-type pathology at time of diagnosis and remained stable during follow-up in the majority of patients. In sum, our data suggests frequent co-morbidity of glioma and AD-type pathology, which increases with age. Isolated tau-pathology was more pronounced in IDH-mutant tumors. In tumor-adjacent cortex, increased tumor cell infiltration paralleled neuronal loss and diminished AD-type pathology.

Association of MRI Volume Parameters in Predicting Patient Outcome at Time of Initial Diagnosis of Glioblastoma.

Patients with glioblastoma (GBM) may demonstrate varying patterns of infiltration and relapse. Improving the ability to predict these patterns may influence the management strategies at the time of initial diagnosis. This study aims to examine the impact of the ratio (T2/T1) of the non-enhancing volume in T2-weighted images (T2) to the enhancing volume in MRI T1-weighted gadolinium-enhanced images (T1gad) on patient outcome. A retrospective audit was performed from established prospective databases in patients managed consecutively with radiation therapy (RT) for GBM between 2016 and 2019. Patient, tumour and treatment-related factors were assessed in relation to outcome. Volumetric data from the initial diagnostic MRI were obtained via the manual segmentation of the T1gd and T2 abnormalities. A T2/T1 ratio was calculated from these volumes. The initial relapse site was assessed on MRI in relation to the site of the original T1gad volume and surgical cavity. The major endpoints were median relapse-free survival (RFS) from the date of diagnosis and site of initial relapse (defined as either local at the initial surgical site or any distance more than 20 mm from initial T1gad abnormality). The analysis was performed for association between known prognostic factors as well as the radiological factors using log-rank tests for subgroup comparisons, with correction for multiple comparisons. One hundred and seventy-seven patients with GBM were managed consecutively with RT between 2016 and 2019 and were eligible for the analysis. The median age was 62 years. Seventy-four percent were managed under a 60Gy (Stupp) protocol, whilst 26% were on a 40Gy (Elderly) protocol. Major neuroanatomical subsites were Lateral Temporal (18%), Anterior Temporal (13%) and Medial Frontal (10%). Median volumes on T1gd and T2 were 20 cm3 (q1-3:8-43) and 37 cm3 (q1-3: 17-70), respectively. The median T2/T1 ratio was 2.1. For the whole cohort, the median OS was 16.0 months (95%CI:14.1-18.0). One hundred and forty-eight patients have relapsed with a median RFS of 11.4 months (95%CI:10.4-12.5). A component of distant relapse was evident in 43.9% of relapses, with 23.6% isolated relapse. Better ECOG performance Status (p = 0.007), greater extent of resection (p = 0.020), MGMT methylation (p < 0.001) and RT60Gy Dose (p = 0.050) were associated with improved RFS. Although the continuous variable of initial T1gd volume (p = 0.39) and T2 volume (p = 0.23) were not associated with RFS, the lowest T2/T1 quartile (reflecting a relatively lower T2 volume compared to T1gd volume) was significantly associated with improved RFS (p = 0.016) compared with the highest quartile. The lowest T2/T1 ratio quartile was also associated with a lower risk of distant relapse (p = 0.031). In patients diagnosed with GBM, the volumetric parameters of the diagnostic MRI with a ratio of T2 and T1gad abnormality may assist in the prediction of relapse-free survival and patterns of relapse. A further understanding of these relationships has the potential to impact the design of future radiation therapy target volume delineation protocols.

13 Reduced Left Orbitofrontal Volume Correlates with Semantic Verbal Fluency Performance Among Veterans with TBI

Objective:Semantic verbal fluency (SVF) has traditionally been correlated with activity in the left anterior temporal lobe. Damage to the anterior temporal lobe, such as from a traumatic brain injury (TBI) or neurodegenerative disease, can result in impairments in semantic fluency and language expression. Although the orbitofrontal cortex (OFC) is not well studied as being correlated to this process, its functional connectivity to language-related brain regions, such as the inferior frontal gyrus (Broca’s area), suggests it may also play a role in SVF. In fact, preliminary research has shown that lesions in the OFC are associated with semantic fluency deficits (Almairac et al. 2015). As such, the goal of this study was to investigate whether OFC volume correlates with SVF in a population of Veterans with a prior TBI.Participants and Methods:Thirty-five Veterans who sustained TBIs were included in this study (11% female, age M = 41.77, SD = 11.27; years of education M =14.94, SD = 1.62). All participants underwent a magnetic resonance imaging (MRI) and regional normative volumes were standardized to account for differences in brain size (volume of brain area/total brain volume). Participants were given the Delis Kaplan Executive Function System (D-KEFS) verbal fluency subtest to measure verbal generativity. A Pearson correlation was conducted to investigate the relationship between OFC volume and SVF performance. Post hoc analysis was conducted with Veterans who met the criteria for a TBI sustained by a blast to the head (n = 25).Results:A significant positive correlation emerged between left OFC volume and semantic fluency (category subtest; p = .03, r = .35). Additional analyses with Veterans with blast-related TBIs indicated a significant correlation between the volume of the OFC and performance on the category (p = .02, r = .45), and switching (p = .02, r = .43) subtests of the D-KEFS verbal fluency test.Conclusions:These findings demonstrate a correlation between the volume of the left OFC and SVF performance. Specifically, Veterans with TBIs had decreased volume of the OFC which correlated with deficits on a SVF task. Among Veterans with blast-related TBIs, both category and category switching subtests correlated with OFC volume. Given the functional connectivity between the OFC and language areas of the brain, this study highlights the importance of analyzing associated cortical regions beyond the anterior temporal lobe when studying SVF performance.

Time-series visual explainability for Alzheimer’s disease progression detection for smart healthcare

Artificial intelligence (AI)-based diagnostic systems provide less error-prone and safer support to clinicians, enhancing the medical decision-making process. This study presents a smart and reliable healthcare framework for detecting Alzheimer's disease (AD) progression. Early detection of AD before the onset of clinical symptoms is the most crucial step in starting timely treatment. To predict the conversion of cognitively normal patients to those with AD, three-dimensional (3D) magnetic resonance imaging (MRI) whole-brain neuroimaging methods have been extensively studied. However, depending on the 3D volume, this method is computationally expensive. To solve this problem, we used an approximate rank pooling method originally designed for video action recognition with a 3D MRI volume to obtain a compressed representation of multiple two-dimensional (2D) MRI slices. This study proposes a hybrid multimodal CNN-Bi-LSTM deep model for AD progression detection, in which the resulting dynamic 2D images are fused with cognitive features. Moreover, a novel explainable AI approach is proposed to provide visual explanations using the resulting longitudinal 2D dynamic images. Temporal explanations were provided by visualizing the affected brain regions captured using longitudinal 2D MRIs. By utilizing a sample of 1,692 subjects with multimodal data from the Alzheimer’s Disease Neuroimaging Initiative dataset, our method was assessed using a 10-fold cross-validation process. The model achieved an area under the receiver operating characteristics curve (AUC) of 94% using longitudinal 2D three-time-step dynamic image data. The fusion of 2D dynamic images with cognitive features enhanced the performance by 2% in terms of the AUC. Patients who gradually develop AD, show changes in various brain regions. For such patients, our system highlights the critical role of the hippocampus, medial amygdala, caudal hippocampus, and lateral amygdala at the initial time steps. In the late stages of AD, the system detects abnormalities in extra brain regions such as the medial temporal gyrus, superior temporal gyrus, fusiform gyrus, and caudal hippocampus; indicating that patients have completely progressed to AD.

A fully convolutional neural network for comprehensive compartmentalization of abdominal adipose tissue compartments in MRI

BackgroundExisting literature has highlighted structural, physiological, and pathological disparities among abdominal adipose tissue (AAT) sub-depots. Accurate separation and quantification of these sub-depots are crucial for advancing our understanding of obesity and its comorbidities. However, the absence of clear boundaries between the sub-depots in medical imaging data has challenged their separation, particularly for internal adipose tissue (IAT) sub-depots. To date, the quantification of AAT sub-depots remains challenging, marked by a time-consuming, costly, and complex process. PurposeTo implement and evaluate a convolutional neural network to enable granular assessment of AAT by compartmentalization of subcutaneous adipose tissue (SAT) into superficial subcutaneous (SSAT) and deep subcutaneous (DSAT) adipose tissue, and IAT into intraperitoneal (IPAT), retroperitoneal (RPAT), and paraspinal (PSAT) adipose tissue. Material and methodsMRI datasets were retrospectively collected from Singapore Preconception Study for Long-Term Maternal and Child Outcomes (S-PRESTO: 389 women aged 31.4 ± 3.9 years) and Singapore Adult Metabolism Study (SAMS: 50 men aged 28.7 ± 5.7 years). For all datasets, ground truth segmentation masks were created through manual segmentation. A Res-Net based 3D-UNet was trained and evaluated via 5-fold cross-validation on S-PRESTO data (N = 300). The model's final performance was assessed on a hold-out (N = 89) and an external test set (N = 50, SAMS). ResultsThe proposed method enabled reliable segmentation of individual AAT sub-depots in 3D MRI volumes with high mean Dice similarity scores of 98.3%, 97.2%, 96.5%, 96.3%, and 95.9% for SSAT, DSAT, IPAT, RPAT, and PSAT respectively. ConclusionConvolutional neural networks can accurately sub-divide abdominal SAT into SSAT and DSAT, and abdominal IAT into IPAT, RPAT, and PSAT with high accuracy. The presented method has the potential to significantly contribute to advancements in the field of obesity imaging and precision medicine.

Interpreting deep learning models for glioma survival classification using visualization and textual explanations

BackgroundSaliency-based algorithms are able to explain the relationship between input image pixels and deep-learning model predictions. However, it may be difficult to assess the clinical value of the most important image features and the model predictions derived from the raw saliency map. This study proposes to enhance the interpretability of saliency-based deep learning model for survival classification of patients with gliomas, by extracting domain knowledge-based information from the raw saliency maps.Materials and methodsOur study includes presurgical T1-weighted (pre- and post-contrast), T2-weighted and T2-FLAIR MRIs of 147 glioma patients from the BraTs 2020 challenge dataset aligned to the SRI 24 anatomical atlas. Each image exam includes a segmentation mask and the information of overall survival (OS) from time of diagnosis (in days). This dataset was divided into training (n=118) and validation (n=29) datasets. The extent of surgical resection for all patients was gross total resection. We categorized the data into 42 short (mean mu =157 days), 30 medium (mu =369 days), and 46 long (mu =761 days) survivors. A 3D convolutional neural network (CNN) trained on brain tumour MRI volumes classified all patients based on expected prognosis of either short-term, medium-term, or long-term survival. We extend the popular 2D Gradient-weighted Class Activation Mapping (Grad-CAM), for the generation of saliency map, to 3D and combined it with the anatomical atlas, to extract brain regions, brain volume and probability map that reveal domain knowledge-based information.ResultsFor each OS class, a larger tumor volume was associated with a shorter OS. There were 10, 7 and 27 tumor locations in brain regions that uniquely associate with the short-term, medium-term, and long-term survival, respectively. Tumors located in the transverse temporal gyrus, fusiform, and palladium are associated with short, medium and long-term survival, respectively. The visual and textual information displayed during OS prediction highlights tumor location and the contribution of different brain regions to the prediction of OS. This algorithm design feature assists the physician in analyzing and understanding different model prediction stages.ConclusionsDomain knowledge-based information extracted from the saliency map can enhance the interpretability of deep learning models. Our findings show that tumors overlapping eloquent brain regions are associated with short patient survival.

Non-abstinent recovery in alcohol use disorder is associated with greater regional cortical volumes than heavy drinking.

Harm-reduction (i.e., non-abstinent recovery) approaches to substance use treatment have garnered increasing attention. Reduced levels of alcohol consumption post-treatment have been associated with better psychosocial functioning and physical health, yet less is known regarding differences in brain structures associated with varying levels of alcohol consumption. This study investigated regional cortical volumes after alcohol use disorder (AUD) treatment among individuals who achieved complete abstinence and those who returned to lower and higher levels of consumption. Data were collected from individuals with AUD (n = 68) approximately 8 months after the initiation of treatment. Using risk drinking levels defined by the World Health Organization, participants were classified as abstaining (AB) or relapsing with low (RL) or higher (RH) levels. Data were also obtained from 34 age-matched light/non-drinking controls (LN). All participants completed a 1.5 T magnetic resonance imaging session and volumes for 34 bilateral cortical regions of interest were quantitated with FreeSurfer. Generalized linear models were used to examine group differences in cortical volume. All group findings are significant at an FDR-corrected value of 0.018. Adjusting for age and intracranial volume, significant group differences were found in 13/34 cortical regions. AB showed greater volumes than RL in 2/13 regions and RH in 6/13 regions. RH demonstrated significantly smaller volumes than LN in 12/13 ROIs, whereas RL differed from LN in 9/13 regions. RH and RL differed in only two cortical regions. Individuals who consumed low-risk levels of alcohol post-treatment exhibited regional cortical volumes more similar to abstainers than individuals who returned to higher-risk levels. This suggests that low-risk levels of alcohol consumption are associated with brain integrity that is comparable to that seen with complete abstinence. Given the previously demonstrated improvement in psychosocial and physical health with reduced levels of alcohol consumption post-treatment, harm reduction may be a beneficial and more attainable goal for some individuals with AUD who are seeking treatment.

An Automated Brain Metastasis Detection and Segmentation System from MRI with a Large Multi-Institutional Dataset

Developments of automated systems for brain metastasis (BM) detection and segmentation from MRI for assisting early detection and stereotactic radiosurgery (SRS) have been reported but most based upon relatively small datasets from single institutes. This work aims to develop and evaluate a system using a large multi-institutional dataset, and to improve both identification of small/subtle BMs and segmentation accuracy of large BMs. A 3D U-Net system was trained and evaluated to detect and segment intraparenchymal BMs with a size > 2mm using 1856 MRI volumes from 1791 patients treated with SRS from seven institutions (1539 volumes for training, 183 for validation, and 134 for testing). All patients had 3D post-Gd T1w MRI scans pre-SRS. Gross tumor volumes (GTVs) of BMs for SRS were curated by each institute first. Then, additional efforts were spent to create GTVs for the untreated and/or uncontoured BMs, including central reviews by two radiologists, to improve accuracy of ground truth. The training dataset was augmented with synthetic BMs of 3773 MRIs using a 3D generative pipeline. Our system consists of two U-Nets with one using small 3D patches dedicated for detecting small BMs and another using large 3D patches for segmenting large BMs, and a random-forest based fusion module for combining the two network outputs. The first U-Net was trained with 3D patches containing at least one BM < 0.1 cm3. For detection performance, we measured BM-level sensitivity and case-level false-positive (FP) rate. For segmentation performance, we measured BM-level Dice similarity coefficient (DSC) and 95-percentile Hausdorff distance (HD95). We also stratified performances based upon BM sizes. For 739 BMs in the 134 testing cases, the overall lesion-level sensitivity was 0.870 with an average case-level FP of 1.34±1.92 (95% CI: 1.02-1.67). The sensitivity was >0.969 for the BMs >0.1 cm3, but dropped to 0.755 for the BMs < 0.1 cm3 (Table 1). The average DSC and HD95 for all detected BMs were 0.786 and 1.35mm. The worse performance for BMs > 20 cm3 was caused by a case with 83 cm3 GTV and artifacts in the MRI volume. We achieved excellent detection sensitivity and segmentation accuracy for BMs > 0.1 cm3, and promising performance for small BMs (<0.1cm3) with a controlled FP rate using a large multi-institutional dataset. Clinical utility for assisting early detection and SRS planning will be investigated. Table 1: Per-lesion detection and segmentation performance stratified by individual BM size. N is the number of BMs in each category.

Deep Learning-Based Multi-Modality Segmentation of Primary Gross Tumor Volume in CT and MRI for Nasopharyngeal Carcinoma

The delineation of primary gross tumor volume (GTV) of nasopharyngeal carcinoma (NPC) is an essential step for radiotherapy planning. In clinical practice, radiation oncologists manually delineate the GTV in planning CT with the help of diagnostic MRI. This is because NPC tumors are closely adjacent to many important anatomic structures, and CT and MRI provide complementary strength to accurately determine the tumor extension boundary. Manual delineation is time-consuming with the potential registration errors between MRI and CT decreasing the delineation accuracy. In this study, we propose a fully automated GTV segmentation method based on CT and MRI by first aligning MRI to CT, and then, segmenting the GTV using a multi-modality deep learning model. We collected 104 nasopharyngeal carcinoma patients with both planning CT and diagnostic MRI scans (T1 & T2 phases). An experienced radiation oncologists manually delineated the GTV, which was further examined by another senior radiation oncologist. Then, a coarse to fine cross-modality registration from MRI to CT was conducted as follows: (1) A rigid transformation was performed on MRI to roughly align MRI to CT with similar anatomic position. (2) Then, the region of interest (RoI) on both CT and rigid-transformed MRI were cropped. (3) A leading cross-modality deformable registration algorithm, named DEEDS, was applied on the cropped MRI and CT RoIs to find an accurate local alignment. Next, using CT and registered MRI as the combined input, a multi-modality deep segmentation network based on nnUNet was trained to generate the GTV prediction. 20% patients were randomly selected as the unseen testing set to quantitatively evaluate the performance. The quantitative NPC GTV segmentation performance is summarized in Table 1. The deep segmentation model using CT alone achieved reasonable high performance with 76.6% Dice score and 1.34mm average surface distance (ASD). When both CT and registered MRI were used, the segmentation model further improved the performance by 0.9% Dice score increase and 11% relative ASD error reduction, demonstrating the complementary strength of CT and MRI in determining NPC GTV. Notably, the achieved 77.5% Dice score and 1.19mm ASD by the multimodality model is among the top performing results reported in recent automatic NPC GTV segmentation using either CT or MRI modality. We developed a fully automated multi-modal deep-learning model for NPC GTV segmentation. The developed model can segment the NPC GTV in high accuracy. With further optimization and validation, this automated model has potential to standardize the NPC GTV segmentation and significantly decrease the workload of radiation oncologists in clinical practice.

Spectroscopic MRI Detects Occult Glioblastoma Invasion during Chemoradiation

The standard of care for glioblastoma (GBM) includes surgical resection followed by adjuvant chemoradiation (chemoRT). Treatment margins are controversial since conventional imaging does not define the extent of infiltrating tumor cells. Whole-brain spectroscopic MRI (sMRI) allows for visualization of native metabolites in normal brain and tumor cells, and the relative choline to N-acetyl-aspartate ratio greater than 2 (rChoNAA>2) strongly correlates with the presence of occult GBM cells in otherwise normal-appearing brain. With an MRI-Linac, we are performing studies of adaptive radiotherapy to measure changes in cavity size, edema, and enhancement during chemoRT. We questioned whether rChoNAA>2 would change along with anatomical changes to inform clinical target volumes for adaptive chemoRT trials. In a prospective study, 18 patients with primary GBM underwent daily MRI-guided chemoRT with standalone 3T sMRI generation of rChoNAA>2 maps at three timepoints before, during, and after chemoradiation. Conventional treatment volumes of T1 post-contrast and cavity (GTV2, i.e., boost) with or without FLAIR hyperintensity (GTV1) were compared to rChoNAA>2 volumes. DICE similarity coefficients were calculated to assess the spatial similarity of these volumes. Hausdorff distances were calculated to identify rChoNAA>2 extending beyond GTVs throughout the course of chemoradiation. The mean GTV1 was 58.1 cc (range 0-251.4 cc), the mean GTV2 was 47.9 cc (range 0-139.9 cc), and the mean rChoNAA>2 volume was 31.1 cc (range 0-103.2 cc). rChoNAA>2 volumes did not significantly change over the course of chemoRT or correlate with measurement timepoint. The mean DICE similarity coefficient between GTV1 and rChoNAA>2 volumes was 0.39 (range 0-0.80), and the mean DICE similarity coefficient between GTV2 and rChoNAA>2 volumes was 0.29 (range 0-0.77). DICE similarity coefficients were significantly different from unity indicating spatial differences between rChoNAA>2 and conventional MRI volumes. The mean Hausdorff distances of rChoNAA>2 extending beyond GTV1 was 1.3 cm (range 0.7-2.1 cm), and the mean Hausdorff distances of rChoNAA>2 extending beyond GTV2 was 1.9 cm (range 0.8-2.9 cm), suggesting high-risk disease invading beyond what is visible on conventional MRI sequences. Whole-brain sMRI with generation of rChoNAA>2 maps suggest conventional MRI does not fully capture the extent of disease in primary GBM throughout the course of chemoradiation. rChoNAA>2 maps often extend up to approximately 2 cm beyond conventional boost radiotherapy volumes. Further studies are ongoing to determine how sMRI can be used to adapt radiation target volumes during chemoradiation and escalate dose to occult disease.

Prostate MR image quality of apparent diffusion coefficient maps versus fractional intracellular volume maps from VERDICT MRI using the PI-QUAL score and a dedicated Likert scale for artefacts

PurposeThis study aimed to assess the image quality of apparent diffusion coefficient (ADC) maps derived from conventional diffusion-weighted MRI and fractional intracellular volume maps (FIC) from VERDICT MRI (Vascular, Extracellular, Restricted Diffusion for Cytometry in Tumours) in patients from the INNOVATE trial. The inter-reader agreement was also assessed. MethodsTwo readers analysed both ADC and FIC maps from 57 patients enrolled in the INNOVATE prospective trial. Image quality was assessed using the Prostate Imaging Quality (PI-QUAL) score and a subjective image quality Likert score (Likert-IQ). The image quality of FIC and ADC were compared using a Wilcoxon Signed Ranks test. The inter-reader agreement was assessed with Cohen’s kappa. ResultsThere was no statistically significant difference between the PI-QUAL score for FIC datasets compared to ADC datasets for either reader (p = 0.240 and p = 0.614). Using the Likert-IQ score, FIC image quality was higher compared to ADC (p = 0.021) as assessed by reader-1 but not for reader-2 (p = 0.663). The inter-reader agreement was ‘fair’ for PI-QUAL scoring of datasets with FIC maps at 0.27 (95% confidence interval; 0.08–0.46) and ADC datasets at 0.39 (95% confidence interval 0.22–0.57). For Likert scoring, the inter-reader agreement was also ‘fair’ for FIC maps at 0.38 (95% confidence interval; 0.10–0.65) and substantial for ADC maps at 0.62 (95% confidence interval; 0.39–0.86). ConclusionImage quality was comparable for FIC and ADC. The inter-reader agreement was similar when using PIQUAL for both FIC and ADC datasets but higher for ADC maps compared to FIC maps using the image quality Likert score.

Brain tumors segmentation using a hybrid filtering with U-Net architecture in multimodal MRI volumes

Brain tumors segmentation using a hybrid filtering with U-Net architecture in multimodal MRI volumes

Tumoral and peritumoral vascularization of brain tumours: a study comparing an intraoperative ultrasensitive Doppler and a preoperative first-pass perfusion MRI

IntroductionSurgery for gliomas can be guided by neuronavigation using magnetic resonance imaging (MRI) and intraoperative B-mode ultrasound. An ultrasensitive Doppler (USD) using plane waves is a new method of microvascularization imaging which can be used intraoperatively and could identify tumoral and peritumoral areas with neoangiogenesis but its value requires evaluation. The aim of this pilot study then was to evaluate the correlations between ultrasound measurements of glioma vascularization (tumoral and peritumoral region) obtained by a USD intraoperatively and first-pass perfusion measurements obtained on preoperative MRI. Methods18 patients with proven glial tumors were selected for the analysis. They underwent preoperative MRI and intraoperative USD acquisition. The MRI scans were re-aligned to the reference ultrasound slice plane, and for each patient a segmentation of the tumoral and peritumoral zone was performed. Two perfusion parameters were studied: relative cerebral tumor blood volume (rCCBV) in MRI and fractional moving blood volume (FMBV) in a USD. We studied the correlations between mean rCCBV and mean FMBV measured in the tumoral and peritumoral zones in the reference ultrasound slice plane. ResultsThe mean rCCBV and mean FMBV measured in the tumoral zone were significantly and strongly correlated (r = 0.80; p < 0.001). The mean rCCBV and mean FMBV measured in the peritumoral zone were not statistically correlated, although a tendency towards a correlation was noted (r = 0.45; p = 0.067). ConclusionThere was a good correlation between a tumor FMBV obtained by a USD intraoperatively and rCCBV on a preoperative MRI validating the reliability of USD for intraoperative analyses of tumor microvascularization in gliomas.

Duration of fetoscopic spina bifida repair does not affect the central nervous system in fetal lambs

Prenatal spina bifida aperta (SBA) repair improves neurologic outcomes, yet comes with a significant risk of prematurity and uterine scar-related complications. To reduce those, different fetoscopic techniques, e.g. with varying number of ports, are being explored. This has an impact on the duration of the procedure, potentially affecting central nervous system (CNS) development. Both the condition and anesthesia can impact the CNS, particularly the hippocampus, a region vital for prospective and episodic memory. Previous animal studies have shown the potential influence of anesthesia, premature delivery, and maternal surgery during pregnancy on this area. To compare the effects of 2- versus 3-port fetoscopic SBA repair in the fetal lamb model using neuron count of the hippocampus as primary outcome. Based on the hippocampal neuron count from previous lamb experiments, we calculated that we required five animals per group to achieve a statistical power of ≥ 80%. A SBA defect was created in fetal lambs at 75 days of gestation (term: 145 days). At 100 days, fetuses underwent either a 2-port or 3-port fetoscopic repair. At 143 days, all surviving fetuses were delivered by cesarean section, anesthetized and transcardially perfused with a mixture of formaldehyde and gadolinium. Next, they underwent neonatal brain and spine MRI following which these organs were harvested for histology. Hippocampus, frontal cortex, caudate nucleus, and cerebellum samples were immunostained to identify neurons, astrocytes, microglia, and markers associated with cell proliferation, myelination, and synapses. The degree of hindbrain herniation and the ventricular diameter were measured on MRI images and volumes of relevant brain and medulla areas were segmented. Both treatment groups included five fetuses and nine unoperated littermates served as normal controls. The durations for both skin-to-skin (341 ± 31 vs 287 ± 40 min; P=0.04) and fetal surgery (183 ± 30 vs 128 ± 22; P=0.01) were longer for the 2-port than for the 3-port approach. There were no significant differences in neuron density in the hippocampus, frontal cortex and cerebellum. In the caudate nucleus, the neuron count was higher in the 2-port group (965 ± 156 vs. 767 ± 92 neurons/mm2, P=0.04). There were neither differences in proliferation, astrogliosis, synaptophysin, or myelin. The tip of the cerebellar vermis was closer to the foramen magnum in 2-port animals compared to 3-port animals (-0.72 ± 0.67 mm vs. -2.47 ± 0.91 mm, P=0.009). There were no significant differences in the ratio of hippocampus, caudate nucleus, or cerebellar volume to body weight. For the spine, no differences were noted in spine volume/body weight ratio for the lower (L1/L2), mid (L3/L4), and higher (L5/6) levels. Compared to controls, in repaired animals the cerebellar vermis tip lays closer to the foramen magnum, parietal ventricles were enlarged and medulla volumes were reduced. In the experimental spina bifida fetal lamb model, a 2-port repair took 40% longer than a 3-port repair. There was however no indication of any relevant morphologic differences in the fetal brain.

P14.12.A PLASMA CFDNA LIQUID BIOPSY IN THE FOLLOW-UP OF HIGH-GRADE GLIOMAS

Abstract BACKGROUND As highlighted by the 2021 WHO classification of brain tumors, molecular characterization of gliomas is now critical to complement histopathological evaluation. However, intratumor heterogeneity may lead to surgical specimens not being fully representative of the whole tumor. Moreover, given the invasiveness of surgery, tissue analyses are not repeatable over time, thus limiting the possibility to monitor the response to therapies. In this context, we investigated the potential value of plasmatic cell-free DNA (PcfDNA) as a non-invasive ‘liquid biopsy’ marker of disease evolution. MATERIAL AND METHODS Patients’ blood was collected at 4 time points (TP), in parallel to clinical follow-up. A reproducible pipeline was optimized for blood withdrawal, centrifugation, plasma storage, cfDNA extraction from plasma by QIAamp MinElute ccfDNA Midi Kit (Quiagen) and cfDNA fluorometric quantification by Qubit (Thermo Fisher Scientific). MRI volumes were segmented in T1+contrast and T2-FLAIR sequences. RESULTS Fifty-four high-grade glioma patients and 14 healthy controls were enrolled. Blood samples were collected from all 54 patients at radiological diagnosis before surgery (TP0), from 33 patients after surgery (TP1), from 21 patients after radio-chemotherapy(TP2), from 10 patients at the first radiological progression (TP3). At T0, a significantly higher concentration of PcfDNA was detected in patients than in controls. IDH-wild-type gliomas showed a trend toward higher PcfDNA when compared with IDH-mutant gliomas. Digital-droplet PCR, performed at T0on PcfDNA of 4 patients with IDH-mutations in tumor tissues, identified the same IDH1-R132H mutation in all cases. When PcfDNA at TP0 was compared to that at TP1 of the same patient (n=33), there was a significant reduction of plasmatic cfDNA after surgery, mirroring the extensive tumor-resection shown by MRI. Finally, PcfDNA was evaluated in 10 patients until their radiological progression. Interestingly, PcfDNA levels not only followed the same trend of the volumes of T2-FLAIR hyperintensities and T1 contrast-enhancements, but were also mirrored by the evolution of KPS and NANO scores. CONCLUSION Preliminary results suggest that PcfDNA content may be an informative tool to complement MRI during the follow-up of high-grade gliomas.

Correlation between Perfusion-Weighted Magnetic Resonance Imaging (MRI) Features with Histopathological Grading of Glioma

Background: Gliomas are the most common primary neoplasm of the brain. It histopathologically grades I to IV. I and II are low grade while III, IV are high grade. Low-grade gliomas are usually subject to either strict follow-up or surgery. The treatment for high grade gliomas is typically surgery, followed by concomitant radiation therapy and chemotherapy. Conventional MRI has limitations for the grading of gliomas. In addition to conventional MRI techniques, a variety of new techniques such as perfusion-weighted MRI which noninvasively measures cerebral perfusion that can be used for better assessment of glioma. This will be helpful for therapeutic approach and counseling about prognosis of the patient. Methods: This cross sectional observational study was conducted in the Department of Neurosurgery, Bangabandhu Sheikh Mujib Medical University from IRB clearance to 18 months. The patients who fulfill the selection criteria were enrolled in this study. Written informed consent was taken from the patients and/or the legal guardian/ responsible family members after completely explaining to them the procedure and the purpose of the study. Patient’s data was collected in questionnaire/data collection sheet. The privacy of the patient was strictly maintained and the patient’s information will not be disclosed to any source. The study data will only be used for the purpose of this scientific study. This study was not causing any additional harm to the patients. All the patients were evaluated by a standard conventional contrast-enhanced study with perfusion-weighted image on Siemens 3 Tesla MRI. The histopathological grading of the tumor was done as per the WHO classification of 2007. Then correlation was done between PWI finding and WHO histopathological grading. Results: Measurement of relative cerebral blood volume in glioma patients was done in PW-MRI preoperatively. Out of the 35 tumors evaluated, 17 were found as low rCBV and 18 as high. Histopathological examination of glioma was done postoperatively. Among 35 tumors, 19 are low grade gliomas and 16 are high grade gliomas. For the correlation between relative cerebral blood volume in preoperative PW MRI and histopathological grading of glioma, Spearman’s Rank Correlation Coefficient Test was done. We found a significant positive coefficient value of r = 0.572 with a significant p-value of p = &lt;0.001. A single cutoff relative cerebral blood volume value of 1.7 for low versus high grade glioma was found to be 77.8% sensitive and 88.2% specific. Conclusion: The present study ascertains that histopathological grading of glioma is positively correlated with relative cerebral blood volume in PW MRI. Bang. J Neurosurgery 2022; 12(1): 12-19

List-mode quantitative joint reconstruction of activity and attenuation maps in Time-of-Flight PET

Quantitative PET imaging requires accurate determination of patient-specific attenuation maps, which remains challenging on hybrid PET/MRI systems owing to the lack of a direct relationship between MR image intensity and attenuation coefficients. The aim of the present study is to develop a list-mode based algorithm for accurate and robust attenuation correction of PET data using time-of-flight (TOF) emission information. We analyze and address the challenges of list-mode emission-based maximum-likelihood joint estimation of activity and attenuation (LM-MLAA) in state-of-the-art PET imaging. The proposed method exploits a rapid on-the-fly system matrix calculation algorithm based on elliptic integrals while updating the attenuation map from accumulating list-mode coincidences to achieve accelerated image reconstruction. The scattering compensation is incorporated inside it using an iterative approach in such that the current estimation of attenuation map used on a course grid sampling scattering points to make an estimate of scattering. The performance of the proposed LM-MLAA approach was evaluated on Monte Carlo simulations of a phantom at different time resolutions. The contrast and noise for hot and cold regions on reconstructed images at different time resolutions were analysed. The estimated attenuation map exhibits resilience against noise, effectively eliminates high-frequency cross-talk even in the absence of prior information on attenuation coefficients, and enables discrimination among different anatomical regions in the reconstructed image. The error in the mean estimated attenuation coefficients after 50 iterations was ∼ 2% in water and ∼ -14% in Teflon regions for TOF resolutions corresponding to those of most current commercial PET systems (∼ 500 ps). The proposed LM-MLAA framework can be used for joint reconstruction of activity and attenuation maps from list-mode emission data as standalone or a complementary approach to existing in multimodality imaging such as PET/MRI, where direct measurement attenuation maps is not possible.