Brain MRI Images Research Articles

NIMG-67. THE CURRENT STATE OF LARGE LANGUAGE MODELS AND NEURO-ONCOLOGY IMAGING

Abstract INTRODUCTION Large language models (LLM), such as chatGPT by OpenAI, exist at the intersection of machine learning and natural language processing, and are poised to transform healthcare with rapid growth in recent years. The ability of these models to process visual data has also advanced in recent years, with GPT-4 being the first model to publicly process visual data, and GPT-4o adding additional capabilities. There has been significant interest in the application of GPT-4o to healthcare image processing. OBJECTIVE This study sought to examine the ability of GPT-4o in determining a final diagnosis based on a single MRI image of an intracranial neoplasm. We then compared the performance to GPT-4, as well as to humans including neurosurgery residents, fellow, and attendings. METHODS GPT-4 and GPT-4o were given a standard prompt and an accompanying image requesting the most likely diagnosis for 20 brain MRI images, including 5 gliomas, 5 meningiomas, 5 pituitary tumors, and 5 sham non-tumor images. These were compared to a survey of neurosurgery attendings, fellows, senior residents, and junior residents. RESULTS GPT-4 supplied correct diagnoses in just 40% of the cases, while GPT-4o correctly identified the pathology in 70% of cases. Neurosurgeons yielded correct responses in 92% of cases, with neurosurgery senior residents faring the best (albeit with a small sample size). CONCLUSION GPT-4 and GPT-4o underperformed neurosurgery residents, fellows, and attendings in correctly identifying CNS malignancies. Still, with no additional training, GPT-4o yielded significantly better results than GPT-4, showing rapid advancement in the LLM over is predecessor in identifying intracranial neoplasms via single-image MRIs.

Zebra based optimal deep learning for Parkinson’s disease detection using brain MRI images

Zebra based optimal deep learning for Parkinson’s disease detection using brain MRI images

Frontoparietal Structural Network Disconnections Correlate With Outcome After a Severe Stroke.

Structural disconnectome analyses have provided valuable insights into how a stroke lesion results in widespread network disturbances and how these relate to deficits, recovery patterns, and outcomes. Previous analyses have primarily focused on patients with relatively mild to moderate deficits. However, outcomes vary among survivors of severe strokes, and the mechanisms of recovery remain poorly understood. This study assesses the association between lesion-induced network disconnection and outcome after severe stroke. Thirty-eight ischaemic stroke patients underwent MRI brain imaging early after stroke and longitudinal clinical follow-up. Lesion information was integrated with normative connectome data to infer individual disconnectome profiles on a localized regional and region-to-region pathway level. Ordinal logistic regressions were computed to link disconnectome information to the modified Rankin Scale after 3-6 months. Disconnections of ipsilesional frontal, parietal, and temporal cortical brain areas were significantly associated with a worse motor outcome after a severe stroke, adjusted for the initial deficit, lesion volume, and age. The analysis of the underlying pathways mediating this association revealed location-specific results: For frontal, prefrontal, and temporal brain areas, the association was primarily driven by relatively sparse intrahemispheric disconnections. In contrast, the ipsilesional primary motor cortex, the dorsal premotor cortex, and various parietal brain regions showed a remarkable involvement of either frontoparietal intrahemispheric or additionally interhemispheric disconnections. These results indicate that localized disconnection of multiple regions embedded in the structural frontoparietal network correlates with worse outcomes after severe stroke. Specifically, primary motor and parietal cortices might gain particular importance as they structurally link frontoparietal networks of both hemispheres. These data shed novel light on the significance of distinct brain networks for recovery after a severe stroke.

Pushing the limits of MRI brain imaging.

Pushing the limits of MRI brain imaging.

A 3D CNN Model Framework for Early Identification and Classification of Alzheimer’s in Brain MRI Images

A 3D CNN Model Framework for Early Identification and Classification of Alzheimer’s in Brain MRI Images

Generation and discrimination of autism MRI images based on autoencoder.

This study aims to explore an autoencoder-based method for generating brain MRI images of patients with Autism Spectrum Disorder (ASD) and non-ASD individuals, and to discriminate ASD based on the generated images. Initially, we introduce the research background of ASD and related work, as well as the application of deep learning in the field of medical imaging. Subsequently, we detail the architecture and training process of the proposed autoencoder model, and present the results of generating MRI images for ASD and non-ASD patients. Following this, we designed an ASD classifier based on the generated images and elucidated its structure and training methods. Finally, through analysis and discussion of experimental results, we validated the effectiveness of the proposed method and explored future research directions and potential clinical applications. This research offers new insights and methodologies for addressing challenges in ASD studies using deep learning technology, potentially contributing to the automated diagnosis and research of ASD.

12363 A Unique Case Of hüRthle Cell Carcinoma of the Thyroid Gland and Meningioma

Abstract Disclosure: I. Elsayed: None. R. Khan: None. H. Dhaliwal: None. Introduction: MEN1 gene is a tumor suppressor gene on chromosome 11q13 that encodes the nuclear protein menin, which regulates cell growth and cycle. When menin protein is deficient, MEN1 develops. MEN1 mutation usually involves tumors within the parathyroid glands, pancreas, or pituitary. Case Description: A 66-year-old male with past medical history of hypertension, hyperlipidemia, and coronary artery disease presented with sudden onset dizziness, nausea and vomiting for 1 hour. He had similar symptoms on and off for several months, however on the day of presentation, these symptoms were much more acute and severe. Stroke work-up was done including CT and CTA head which revealed a 1.7cm enhancing lesion to R temporal lobe with surrounding edema with an incidental 3 x 3 cm heterogeneous mass in the large portion of the right thyroid gland. Steroids initiated. MRI brain imaging revealed findings consistent with meningioma with local mass effect and pronounced parenchymal edema. Patient underwent successful resection of the mass and pathology confirmed diagnosis of Meningioma, WHO Grade 1. Ear Nose Throat specialist was consulted for incidental thyroid mass. Ultrasound guided Fine needle aspiration biopsy was, Pathology report returned suspicious for follicular neoplasm, Hürthle cell-type. Patient initially underwent right substernal semi-thyroidectomy. Pathology confirmed diagnosis of Oncocytic (Hürthle cell) carcinoma, minimally invasive, 4.3 cm with a single focus of capsular invasion, negative for lymphovascular invasion Discussion: This case presents a 66-year-old male with a rare coexistence of two distinct tumors: a meningioma and a Hürthle cell carcinoma of the thyroid gland. While meningiomas and Hürthle cell carcinoma are not commonly associated with MEN1 syndrome, there have been reports of atypical cases with unusual tumor associations. MEN1 is usually characterized by abnormalities in at least two affected endocrine glands, but variations can involve other endocrine and non-endocrine tumors as well. 25% of thyroid disease can be seen in MEN 1 syndrome; typically papillary thyroid cancer. Only one case in literature found Hürthle cell carcinoma and meningioma in the setting of MEN-1 phenotype. A second case unique case was Hürthle cell carcinoma with metastases to the meninges. The discovery of both a meningioma and a Hürthle cell carcinoma in this patient emphasizes the importance of clinicians being vigilant for the possibility of multiple neoplasms. This case highlights that there may be a relation between role of MEN1 with atypical neoplasms not commonly seen. Presentation: 6/2/2024

Detection of Alzheimer's Disease Using Hybrid Meta-ROI of MRI Structural Images.

The averaged cortical thickness of meta-ROI is currently being used for the diagnosis and prognosis of Alzheimer's disease (AD) using structural MRI brain images. The purpose of this work is to present a hybrid meta-ROI for the detection of AD. The AD detectability of selected cortical and volumetric regions of the brain was examined using signal detection theory. The top performing cortical and volumetric ROIs were taken as input nodes to the artificial neural network (ANN) for AD classification. An AD diagnostic accuracy of 91.9% was achieved by using a hybrid meta-ROI consisting of thicknesses of entorhinal and middle temporal cortices, and the volumes of the hippocampus and inferior lateral ventricles. Pairing inferior lateral ventricle dilation with hippocampal volume reduction improves AD detectability by 5.1%. Hybrid meta-ROI, including the dilation of inferior lateral ventricles, outperformed both cortical thickness- and volumetric-based meta-ROIs in the detection of Alzheimer's disease.

Denoising diffusion model with adversarial learning for unsupervised anomaly detection on brain MRI images

This paper proposes the Adversarial Denoising Diffusion Model (ADDM). Diffusion models excel at generating high-quality samples, outperforming other generative models. These models also achieve outstanding medical image anomaly detection (AD) results due to their strong sampling ability. However, the performance of the diffusion model-based methods is highly varied depending on the sampling frequency, and the time cost to generate good-quality samples is significantly higher than that of other generative models. We propose the ADDM, a diffusion model-based AD method trained with adversarial learning that can maintain high-quality sample generation ability and significantly reduce the number of sampling steps. The proposed adversarial learning is achieved by classifying model-based denoised samples and samples to which random Gaussian noise is added to a specific sampling step. Compared with the loss function of diffusion models, defined under the noise space to minimise the predicted noise and scheduled noise, the diffusion model can explicitly learn semantic information about the sample space since adversarial learning is defined based on the sample space. Our experiment demonstrated that adversarial learning helps achieve a data sampling performance similar to the DDPM with much fewer sampling steps. Experimental results show that the proposed ADDM outperformed existing unsupervised AD methods on Brain MRI images. In particular, in the comparison using 22 T1-weighted MRI scans provided by the Centre for Clinical Brain Sciences from the University of Edinburgh, the ADDM achieves similar performance with 50% fewer sampling steps than other DDPM-based AD methods, and it shows 6.2% better performance about the Dice metric with the same number of sampling steps.

252 - Motion scout trigger technique improves the quality of 3.0T MRI images of the fetal brain

252 - Motion scout trigger technique improves the quality of 3.0T MRI images of the fetal brain

Tumor Detection in MRI Data using Deep Learning Techniques for Image Classification and Semantic Segmentation

This study aims to apply image classification and semantic segmentation methods using deep learning for accurate and efficient tumor detection in brain MRI data. It looks at ways to get over issues including interpretability of the model and a lack of annotated data. A pre-trained VGG19 model and a basic convolutional neural network (CNN) were trained using a public dataset of 273 brain MRI images for the purpose of image classification. Techniques for regularization, transfer learning, and data augmentation were used. A U-Net architecture was trained using manually created masks from the same dataset for semantic segmentation, with several training setups investigated. With a mean test accuracy of 77.19%, the pre-trained VGG19 model surpassed the basic CNN in picture categorization. The U-Net model performed better for semantic segmentation after training on the entire dataset for two epochs as opposed to training on a subset for ten epochs. The work acknowledges issues including data scarcity and the requirement for model interpretability and scalability, which call for additional research, but also underlines the potential of deep learning for brain tumor identification.

Utilizing customized CNN for brain tumor prediction with explainable AI

Timely diagnosis of brain tumors using MRI and its potential impact on patient survival are critical issues addressed in this study. Traditional DL models often lack transparency, leading to skepticism among medical experts owing to their "black box" nature. This study addresses this gap by presenting an innovative approach for brain tumor detection. It utilizes a customized Convolutional Neural Network (CNN) model empowered by three advanced explainable artificial intelligence (XAI) techniques: Shapley Additive Explana-tions (SHAP), Local Interpretable Model-agnostic Explanations (LIME), and Gradient-weighted Class Activation Mapping (Grad-CAM). The study utilized the BR35H dataset, which includes 3060 brain MRI images encompassing both tumorous and non-tumorous cases. The proposed model achieved a remarkable training accuracy of 100 % and validation accuracy of 98.67 %. Precision, recall, and F1 score metrics demonstrated exceptional performance at 98.50 %, confirming the accuracy of the model in tumor detection. Detailed result analysis, including a confusion matrix, comparison with existing models, and generalizability tests on other datasets, establishes the superiority of the proposed approach and sets a new benchmark for accuracy. By integrating a customized CNN model with XAI techniques, this research enhances trust in AI-driven medical diagnostics and offers a promising pathway for early tumor detection and potentially life-saving interventions.

Selective segmentation of brain abnormalities in colour MRI images using variational model

Early detection of brain abnormalities is vital for enhancing patient outcomes and survival rates. However, accurately identifying and segmenting these abnormalities from MRI images remains a persistent challenge. This study assesses the efficacy of the Selective Local Image Fitting (SLIF) model in segmenting brain abnormalities from colour MRI images and compares its performance with converted greyscale counterparts. The rationale behind this comparison stems from standard practice in image segmentation, where colour images are often converted to greyscale before the segmentation task. Converting the image might degrade data by diminishing its dimensions, potentially affecting segmentation computations. This study intends to evaluate the influence of colour information on segmentation accuracy and efficiency by directly assessing the SLIF model on both colour and converted greyscale images. Segmentation accuracy was evaluated using metrics such as the Dice Similarity Coefficient (DSC), Matthews Correlation Coefficient (MCC), and Intersection-over-Union (IoU). Efficiency was determined by measuring the average elapsed processing time. Experimental results demonstrate that colour MRI brain images outperform their converted greyscale counterparts in segmentation accuracy, as colour providing essential supplementary information for precise abnormality delineation. Despite a slight increase in average elapsed processing time for colour images, the enhanced accuracy justifies this trade-off. These findings emphasize the importance of colour MRI in enhancing diagnostic accuracy, especially in detecting brain abnormalities. This study can be extended in future work to evaluate the segmentation accuracy and efficiency of brain abnormalities in 3D colour and greyscale MRI images.

MRI-Based Brain Tumor Classification using ResNet-50 and Optimized Softmax Regression

Accurate classification of brain tumors is crucial for effective treatment planning and patient management. This study presents a new hybrid deep learning classification method based on transfer learning by feature extraction to automate the categorization of MRI brain image datasets into four classes: meningioma, glioma, pituitary tumor, and no tumor. The proposed method combines a finely-tuned ResNet-50 model, a state-of-the-art convolutional neural network architecture, with optimized Softmax Regression (SR) for classification. The study explores the use of data augmentation techniques and evaluates the model's performance on both augmented and unaugmented images. The results demonstrate that the proposed method achieves an impressive accuracy of 98.4%, outperforming existing methods for automatic brain tumor detection. Furthermore, a detailed comparative analysis is presented to evaluate the proposed model's accuracy and efficiency against previous state-of-the-art hybrid models for brain tumor classification. The study suggests that the proposed methodology could be employed as a diagnostic tool to aid radiologists in identifying questionable brain regions, potentially improving the accuracy and efficiency of brain tumor diagnosis.

Hypothalamic Gliosis is Associated With Multiple Cardiovascular Disease Risk Factors.

Hypothalamic gliosis is mechanistically linked to obesity and insulin resistance in rodent models. We tested cross-sectional associations between radiologic measures of hypothalamic gliosis in humans and clinically relevant cardiovascular disease risk factors, as well as prevalent coronary heart disease. Using brain MRI images from Framingham Heart Study participants (N=867; mean age, 55 years; 55% females), T2 signal intensities were extracted bilaterally from the region of interest in the mediobasal hypothalamus (MBH) and reference regions in the amygdala (AMY) and putamen (PUT). T2 signal ratios were created in which greater relative T2 signal intensity suggests gliosis. The primary measure compared MBH to AMY (MBH/AMY); a positive control ratio (MBH/PUT) also assessed MBH whereas a negative control (PUT/AMY) did not. Outcomes were BMI, HDL-C, LDL-C, fasting triglycerides, and the presence of hypertension (n=449), diabetes mellitus (n=66), metabolic syndrome (n=254), or coronary heart disease (n=25). Dietary risk factors for gliosis were assessed in a prospective analysis. Statistical testing was performed using linear or logistic regression. Greater MBH/AMY T2 signal ratios were associated with higher BMI (β = 21.5 [95% CI, 15.4-27.6]; P<0.001), higher fasting triglycerides (β = 1.1 [95% CI, 0.6-1.7]; P<0.001), lower HDL-C (β = -20.8 [95% CI, -40.0 to -1.6]; P=0.034), and presence of hypertension (odds ratio, 1.2 [95% CI, 1.1-1.4]; P=0.0088), and the latter two were independent of BMI. Findings for diabetes mellitus were mixed and attenuated by adjusting for BMI. Metabolic syndrome was associated with MBH/AMY T2 signal ratios (odds ratio, 1.3 [95% CI, 1.1-1.6]; P<0.001). Model results were almost uniformly confirmed by the positive control ratios, whereas negative control ratios that did not test the MBH were unrelated to any outcomes (all P≥0.05). T2 signal ratios were not associated with prevalent coronary heart disease (all P>0.05), but confidence intervals were wide. Self-reported percentages of macronutrient intake were not consistently related to future T2 signal ratios. Using a well-established study of cardiovascular disease development, we found evidence linking hypothalamic gliosis to multiple cardiovascular disease risk factors, even independent of adiposity. Our results highlight the need to consider neurologic mechanisms to understand and improve cardiometabolic health.

Diagnostic utility of brain MRI volumetry in comparing traumatic brain injury, Alzheimer disease and behavioral variant frontotemporal dementia

BackgroundBrain MRI with volumetric quantification, MRI volumetry, can improve diagnostic delineation of patients with neurocognitive disorders by identifying brain atrophy that may not be evident on visual assessments.ObjectiveTo investigate diagnostic utility of MRI volumetry in traumatic brain injury (TBI), early-onset Alzheimer disease (EOAD), late-onset Alzheimer disease, and behavioral variant frontotemporal dementia (bvFTD).MethodWe utilized 137 participants of TBI (n = 40), EOAD (n = 45), LOAD (n = 32), and bvFTD (n = 20). Participants had 3D T1 brain MRI imaging amendable to MRI volumetry. Scan volumes were analyzed with Neuroreader. One-way ANOVA compared brain volumes across diagnostic groups. Discriminant analysis was done with leave-one-out cross validation on Neuroreader metrics to determine diagnostic delineation across groups.ResultLOAD was the oldest compared to other groups (F = 27.5, p < .001). There were no statistically significant differences in sex (p = .58) with women comprising 54.7% of the entire cohort. EOAD and LOAD had the lowest Mini-Mental State Exam (MMSE) scores compared to TBI (p = .04 for EOAD and p = .01 for LOAD). LOAD had lowest hippocampal volumes (Left Hippocampus F = 13.1, Right Hippocampus F = 7.3, p < .001), low white matter volume in TBI (F = 5.9, p < .001), lower left parietal lobe volume in EOAD (F = 9.4, p < .001), and lower total gray matter volume in bvFTD (F = 32.8, p < .001) and caudate atrophy (F = 1737.5, p < .001). Areas under the curve ranged from 92.3 to 100%, sensitivity between 82.2 and 100%, specificity of 78.1-100%. TBI was the most accurately delineated diagnosis. Predictive features included caudate, frontal, parietal, temporal lobar and total white matter volumes.ConclusionWe identified the diagnostic utility of regional volumetric differences across multiple neurocognitive disorders. Brain MRI volumetry is widely available and can be applied in distinguishing these disorders.

Longitudinal Changes of Clinical, Imaging, and Fluid Biomarkers in Preataxic and Early Ataxic Spinocerebellar Ataxia Type 2 and 7 Carriers.

Brain MRI abnormalities and increases in neurofilament light chain (NfL) have mostly been observed in cross-sectional studies before ataxia onset in polyglutamine spinocerebellar ataxias. Our study aimed to identify longitudinal changes in biological, clinical, and/or imaging biomarkers in spinocerebellar ataxia (SCA) 2 and SCA7 carriers over 1 year. We studied SCA2 and SCA7 carriers and controls (expansion-negative relatives) at the Paris Brain Institute. Inclusion criteria included Scale for the Assessment and Rating of Ataxia (SARA) scores between 0 and 15. Assessments at baseline, 6 months, and 12 months comprised neurologic, quality of life, orofacial motor, neuropsychological, and ophthalmologic examinations, along with gait and oculomotor recordings, brain MRI, CSF, and blood sampling. The primary outcome was the longitudinal change in these assessments over 1 year. We included 15 SCA2 carriers, 15 SCA7 carriers, and 10 controls between May 2020 and April 2021. At baseline, the ages were similar (41 [37, 46] for SCA2, 38 [28.5, 39.8] for SCA7, and 39.5 [31, 54.5] for controls, p = 0.78), as well the sex (p = 0.61); SARA scores were low but different (4 [1.25, 6.5] in SCA2, 2 [0, 11.5] in SCA7, and 0 in controls, p < 0.01). Pons and medulla volumes were smaller in SCAs (p < 0.05) and cerebellum volume only in SCA2 (p = 0.01). Plasma NfL levels were higher in SCA participants (SCA2: 14.2 pg/mL [11.52, 15.89], SCA7: 15.53 [13.27, 23.23]) than in controls (4.88 [3.56, 6.17], p < 0.001). After 1-year follow-up, in SCA2, there was significant pons (-144 ± 60 mm3) and cerebellum (-1,508 ± 580 mm3) volume loss and a worsening of gait assessment; in SCA7, SARA score significantly increased (+1.3 ± 0.4) and outer retinal nuclear layer thickness decreased (-15.4 ± 1.6 μm); for both SCA groups, the orofacial motor assessment significantly worsened. For preataxic and early ataxic carriers, the strongest longitudinal deterioration on outcome measures was orofacial motility in SCA2 and retinal thickness in SCA7. Despite the limitation of the small sample size, we detected annual changes in preataxic and early ataxic SCA individuals across brain MRI imaging, clinical scores, gait parameters, and retinal thickness. These parameters could serve as potential end points for future therapeutic trials in the preataxic phase. ClinicalTrials.gov NCT04288128.

A Unified Pipeline for Simultaneous Brain Tumor Classification and Segmentation Using Fine-Tuned CNN and Residual UNet Architecture.

In this paper, I present a comprehensive pipeline integrating a Fine-Tuned Convolutional Neural Network (FT-CNN) and a Residual-UNet (RUNet) architecture for the automated analysis of MRI brain scans. The proposed system addresses the dual challenges of brain tumor classification and segmentation, which are crucial tasks in medical image analysis for precise diagnosis and treatment planning. Initially, the pipeline preprocesses the FigShare brain MRI image dataset, comprising 3064 images, by normalizing and resizing them to achieve uniformity and compatibility with the model. The FT-CNN model then classifies the preprocessed images into distinct tumor types: glioma, meningioma, and pituitary tumor. Following classification, the RUNet model performs pixel-level segmentation to delineate tumor regions within the MRI scans. The FT-CNN leverages the VGG19 architecture, pre-trained on large datasets and fine-tuned for specific tumor classification tasks. Features extracted from MRI images are used to train the FT-CNN, demonstrating robust performance in discriminating between tumor types. Subsequently, the RUNet model, inspired by the U-Net design and enhanced with residual blocks, effectively segments tumors by combining high-resolution spatial information from the encoding path with context-rich features from the bottleneck. My experimental results indicate that the integrated pipeline achieves high accuracy in both classification (96%) and segmentation tasks (98%), showcasing its potential for clinical applications in brain tumor diagnosis. For the classification task, the metrics involved are loss, accuracy, confusion matrix, and classification report, while for the segmentation task, the metrics used are loss, accuracy, Dice coefficient, intersection over union, and Jaccard distance. To further validate the generalizability and robustness of the integrated pipeline, I evaluated the model on two additional datasets. The first dataset consists of 7023 images for classification tasks, expanding to a four-class dataset. The second dataset contains approximately 3929 images for both classification and segmentation tasks, including a binary classification scenario. The model demonstrated robust performance, achieving 95% accuracy on the four-class task and high accuracy (96%) in the binary classification and segmentation tasks, with a Dice coefficient of 95%.

Atlas-based assessment of hypomyelination: Quantitative MRI in Pelizaeus-Merzbacher disease.

Pelizaeus-Merzbacher disease (PMD) is a rare childhood hypomyelinating leukodystrophy. Quantification of the pronounced myelin deficit and delineation of subtle myelination processes are of high clinical interest. Quantitative magnetic resonance imaging (qMRI) techniques can provide in vivo insights into myelination status, its spatial distribution, and dynamics during brain maturation. They may serve as potential biomarkers to assess the efficacy of myelin-modulating therapies. However, registration techniques for image quantification and statistical comparison of affected pediatric brains, especially those of low or deviant image tissue contrast, with healthy controls are not yet established. This study aimed first to develop and compare postprocessing pipelines for atlas-based quantification of qMRI data in pediatric patients with PMD and evaluate their registration accuracy. Second, to apply an optimized pipeline to investigate spatial myelin deficiency using myelin water imaging (MWI) data from patients with PMD and healthy controls. This retrospective single-center study included five patients with PMD (mean age, 6 years ± 3.8) who underwent conventional brain MRI and diffusion tensor imaging (DTI), with MWI data available for a subset of patients. Three methods of registering PMD images to a pediatric template were investigated. These were based on (a) T1-weighted (T1w) images, (b) fractional anisotropy (FA) maps, and (c) a combination of T1w, T2-weighted, and FA images in a multimodal approach. Registration accuracy was determined by visual inspection and calculated using the structural similarity index method (SSIM). SSIM values for the registration approaches were compared using a t test. Myelin water fraction (MWF) was quantified from MWI data as an assessment of relative myelination. Mean MWF was obtained from two PMDs (mean age, 3.1 years ± 0.3) within four major white matter (WM) pathways of a pediatric atlas and compared to seven healthy controls (mean age, 3 years ± 0.2) using a Mann-Whitney U test. Our results show that visual registration accuracy estimation and computed SSIM were highest for FA-based registration, followed by multimodal, and T1w-based registration (SSIMFA = 0.67 ± 0.04 vs. SSIMmultimodal = 0.60 ± 0.03 vs. SSIMT1 = 0.40 ± 0.14). Mean MWF of patients with PMD within the WM pathways was significantly lower than in healthy controls MWFPMD = 0.0267 ± 0.021 vs. MWFcontrols = 0.1299 ± 0.039. Specifically, MWF was measurable in brain structures known to be myelinated at birth (brainstem) or postnatally (projection fibers) but was scarcely detectable in other brain regions (commissural and association fibers). Taken together, our results indicate that registration accuracy was highest with an FA-based registration pipeline, providing an alternative to conventional T1w-based registration approaches in the case of hypomyelinating leukodystrophies missing normative intrinsic tissue contrasts. The applied atlas-based analysis of MWF data revealed that the extent of spatial myelin deficiency in patients with PMD was most pronounced in commissural and association and to a lesser degree in brainstem and projection pathways.

264. SIGNIFICANCE AND USEFULNESS OF PLASMA SUBSTANCE P CONCENTRATION IN PREDICTING HIGH-RISK PATIENTS FOR ASPIRATION AFTER ESOPHAGEAL CANCER SURGERY

Abstract Background The greatest challenge in postoperative complications of esophageal cancer is the prevention of pulmonary complications, especially aspiration pneumonia. Although the introduction of minimally invasive surgery and intervention in breathing and swallowing rehabilitation can improve the motor function of breathing and swallowing, we reported the possibility that a decline in the swallowing reflex may play a significant role in esophageal cancer patients, many of whom are elderly. In fact, about 80% of esophageal cancer patients in Japan are over 60 years old, and 53% of them have asymptomatic cerebral infarction, about half of which are reported to be found near the basal ganglia, which control the swallowing reflex. The swallowing reflex is controlled by the secretion of substance P, synthesized in the cervical sympathetic ganglia, into the pharynx and trachea upon stimulation of dopamine synthesized in the basal ganglia. Methods To test the hypothesis that patients with preoperative subclinical cerebral infarction are more likely to develop aspiration postoperatively due to decreased substance P secretion. To compare the association between the presence or absence of cerebral infarction near the basal ganglia and serum substance P levels and the development of postoperative aspiration or aspiration pneumonia due to videoendoscopy and video fluorography in patients with thoracic esophageal cancer scheduled for video-assisted esophagectomy by preoperative brain MRI imaging and blood sampling. Results We report the results of an interim analysis of 54 cases in which the measurement of substance P was completed. Male: 39 cases, female: 15 cases. Mean age: 69.3 (49-82) years. Tumor location: Ut/Mt/Lt=13/25/14. MRI showed infarction in or near the basal ganglia in only 2 cases. Median preoperative blood substance P concentration: 48.2 pg/ml. Eleven one patients had postoperative aspiration. Preoperative substance P levels ≥48.2 pg/ml were defined as the High group, and those below 48.2 pg/ml were defined as the Low group. Postoperative aspiration was observed in 2/27 (7.4%) patients in the High group, while 9/26 (34.6%) patients in the Low group, which was significantly higher (p=0.015). Conclusion In fact, subclinical cerebral infarction was much less common than reported and was not useful as a predictor of postoperative aspiration. On the other hand, patients with low preoperative blood substance P levels had a very high risk of developing aspiration postoperatively due to impaired swallowing reflexes, suggesting that preoperative blood substance P levels may be useful in assessing postoperative aspiration risk in patients with thoracic esophageal cancer.