Image Data Research Articles

RADT-15. THE ROLE OF STEREOTACTIC RADIOSURGERY IN WHO GRADE II MENINGIOMAS: RESULTS FROM A LARGE EUROPEAN MULTICENTER STUDY

Abstract BACKGROUND This study evaluates the outcomes of WHO grade II meningiomas treated with stereotactic radiosurgery (SRS) using a large, multi-institutional European database. We aim to investigate the long-term tumor control rate and morbidity associated with SRS. METHODS Data from 355 consecutive patients treated between 1992 and 2022 at 16 European radiosurgery centers were retrospectively analyzed. After excluding 40 patients due to loss to follow-up, results from 315 patients were included. Clinical and imaging data were independently collected by each center and uniformly entered into a multicenter database. RESULTS The median age of patients was 56 years, with 54% being female. The median follow-up duration was 46.3 ± 36.9 months. Tumors had prior surgical treatment and biopsy in 99% and 1% of cases, respectively. Tumors received previous radiotherapy or SRS in 27.4% of cases. SRS was performed due to local progression in 89.6% of cases, residual tumor in 6.4%, and distant progression in 4.0%. SRS characteristics included a median target volume of 5.1 cm³ and a median margin dose of 16 Gy (53.5% isodose). The 3-year local control rate was 72.0%, with a progression-free survival rate of 66.6% and an overall survival rate of 89.7%. Transient and permanent morbidity rates were 6.4% and 3.5%, respectively. Multivariate analysis identified previous radiotherapy and age as significant factors influencing these outcomes. CONCLUSIONS This study is the largest to date on SRS in WHO grade II meningiomas. Our data shows that SRS is a viable and safe treatment option for residual and recurrent WHO grade II meningiomas

A Radio Frequency Interference Screening Framework—From Quick-Look Detection Using Statistics-Assisted Network to Raw Echo Tracing

Synthetic aperture radar (SAR) is often affected by other high-power electromagnetic devices during ground observation, which causes unintentional radio frequency interference (RFI) with the acquired echo, bringing adverse effects into data processing and image interpretation. When faced with the task of screening massive SAR data, there is an urgent need for the global perception and detection of interference. The existing RFI detection method usually only uses a single type of data for detection, ignoring the information association between the data at all levels of the real SAR product, resulting in some computational redundancy. Meanwhile, current deep learning-based algorithms are often unable to locate the range of RFI coverage in the azimuth direction. Therefore, a novel RFI processing framework from quick-looks to single-look complex (SLC) data and then to raw echo is proposed. We take the data of Sentinel-1 terrain observation with progressive scan (TOPS) mode as an example. By combining the statistics-assisted network with the sliding-window algorithm and the error-tolerant training strategy, it is possible to accurately detect and locate RFI in the quick looks of an SLC product. Then, through the analysis of the TOPSAR imaging principle, the position of the RFI in the SLC image is preliminarily confirmed. The possible distribution of the RFI in the corresponding raw echo is further inferred, which is one of the first attempts to use spaceborne SAR data to ‌elucidate the RFI location mapping relationship between image data and raw echo. Compared with directly detecting all of the SLC data, the time for the proposed framework to determine the RFI distribution in the SLC data can be shortened by 53.526%. All the research in this paper is conducted on Sentinel-1 real data, which verify the feasibility and effectiveness of the proposed framework for radio frequency signals monitoring in advanced spaceborne SAR systems.

Elevating Facial Expression Detection: Empowered by VGG-19 and Weight-Normalized Gradient Boost Algorithm

Facial expression recognition termed as FER, is one of the vital tasks which is able in mimicking the human ability and wellness. Face expression and other gestures delivered via face are some of the important aspects of conveying non-verbal communications which plays a vital role in interpersonal relations. This domain has a higher scope of research due to enhancing computer vision and human-computer interaction. With respect and consideration to these domain interest, DL has gained a reasonable approach in performing FER. Some of the state-of-art approaches encompassing the ML approaches have faced several cases of laybacks such as overfitting, computational complexity, less adaptable to higher and big datasets. Thus, considering these laybacks and in achieving accurate FER in aspects of bringing proximal levels of FER, the current approach deployed DL methods for both feature extraction from the input image and in classifying them. Deep multi-level feature extraction is performed using the VGG-19 model and Weight Normalized gradient boosting algorithm is adapted for classifying these expressions using the FER13 dataset. This dataset constitutes the input images, which are in ranges of 28,709 sample image data and about 3,589 image data for test. These input images are initially pre-processed for obtaining better accuracy rates when performing feature extraction and classifications. The complete model in effective FER is evaluated using the performance metrics comprising Accuracy (98%), Recall (99%), F1-score (98%) and the precision rates (97%). This analysis of the performance will aid in affirming the overall efficacy of the proposed system.

NIMG-29. EARLY REDUCTION IN MRI DIFFUSION APPARENT DIFFUSION COEFFICIENT (ADC) STRONGLY PREDICTS LONG TERM RESPONSE TO ONC201 THERAPY IN PATIENTS WITH H3K27M-DMG

Abstract ONC201 is the first monotherapy to improve outcomes in H3K27M- diffuse midline glioma (DMG) beyond radiation. Despite impressive early efficacy in H3K27M-DMG, individual response to ONC201 is variable and no radiographic biomarkers predict long term response. Previous studies demonstrated baseline MRI diffusion apparent diffusion coefficient (ADC) is lower in DMG with H3K27M compared to wildtype but change after radiation therapy was not correlated with improved OS or PFS. We hypothesize ADC will enable stratification of patients more likely to respond to ONC201. Imaging and clinical data were abstracted from chart review of patients treated with ONC201 at University of Michigan. Two neuroradiologists performed centralized review for RAPNO measurements and mean ADC was assessed using ROI measured in consistent anatomic locations, with areas of cystic degeneration or necrosis excluded. Sixty-one patients were identified for review. Preliminary analysis was performed on twenty-three patients, with upfront ONC201 therapy, median age 14.8 years old (5-25yo), primary site includes pontine (n=15), thalamic (n=7). Baseline characteristics of age, ADC and size at diagnosis or follow-up RAPNO scored size were not statistically significant. Patients with longer than median survival (334 days) demonstrated increased ADC (+58.25x10-6mm2/s) from baseline to cycle 2 whereas patients with shorter than median survival demonstrated decreased ADC (-174.8x10-6mm2/s) (p = 0.0076). Further, patients with decreased ADC demonstrated mean OS of 302.8 days and patients with increased ADC had mean OS of 588.0 (p = 0.0054). In H3K27M-DMG patients treated with ONC201, increased ADC after two cycles of ONC201 was strongly predictive of longer overall survival. Our recent work demonstrated that ONC201 disrupts metabolic and epigenetic pathways to restore pathognomonic H3K27me3 reduction, which may underline greater change in ADC. Ongoing work will validate these findings in an independent external cohort and integrate histogram analysis, parametric assessments, MRI perfusion and liquid biopsy biomarkers (cf-tDNA and metabolomics).

Radiomics-Based Diagnosis in Dentomaxillofacial Radiology: A Systematic Review.

Radiomics is a quantitative tool for digital image analysis. This systematic review aims to investigate the scientific articles to evaluate the potential implications of Radiomics analysis in Dentomaxillofacial Radiology (DMFR). Studies regarding Radiomics applications in DMFR and human samples, in vivo study, a case reports/series if ≧5 samples were included, while case reports/series if < 5 samples, articles other than in English, abstracts without full text, and studies published before 2015 were excluded. Fifty-one articles were selected from 3789 literatures. The QUADAS-2 tool was used for risk of bias assessment. The accuracy of predicting dentomaxillofacial pathologies was considered as the primary outcome, and the modeling type of Radiomics was considered as the secondary outcome. A meta-analysis could not be performed due to the lack of information and standardization among the reported accuracies. The reported accuracies were found between 0.66 and 99.65%. Logistic regression (n = 6) was found to be the most common Radiomics modeling type, followed by Support Vector Machine and Decision Tree (n = 5). Second-order statistics (n = 38) was the most common type of Radiomics application, followed by first-order (n = 26), higher-order (n = 20), and shape-based (n = 15) statistics. Further work is needed to increase standardization in the Radiomics workflow. Quantitative image analysis is an alternative tool for conventional visual radiographic evaluation. Radiomics systems depend on elements such as imaging modality, feature type, data mining, or statistical method. Radiomics applications do not justify digital transformation on their own, but the potential of its integration into the digital workflow is considerable.

Multimodal neuroimaging-based prediction of Parkinson's disease with mild cognitive impairment using machine learning technique.

This study aimed to identify potential markers that can predict Parkinson's disease with mild cognitive impairment (PDMCI). We retrospectively collected general demographic data, clinically relevant scales, plasma samples, and neuroimaging data (T1-weighted magnetic resonance imaging (MRI) data as well as resting-state functional MRI [Rs-fMRI] data) from 173 individuals. Subsequently, based on the aforementioned multimodal indices, a support vector machine was employed to investigate the machine learning (ML) classification of PD patients with normal cognition (PDNC) and PDMCI. The performance of 29 classifiers was assessed based on various combinations of indicators. Results demonstrated that the optimal classifier in the validation set was composed by clinical + Rs-fMRI+ neurofilament light chain, exhibiting a mean Accuracy of 0.762, a mean area under curve of 0.840, a mean sensitivity of 0.745, along with a mean specificity of 0.783. The ML algorithm based on multimodal data demonstrated enhanced discriminative ability between PDNC and PDMCI patients.

INNV-33. ADVANCING PRECISION TREATMENT FOR DIFFUSE MIDLINE GLIOMA: THE DMG LEARNING HUB INITIATIVE

Abstract Diffuse Midline Glioma (DMG) poses a significant challenge in neuro-oncology due to its heterogeneous presentation and limited therapeutic options. Clinical trials often serve as the primary avenue for treatment, yet families encounter numerous barriers in navigating and selecting appropriate trials. Health disparities, including disparities in medical literacy and access to specialized medical centers, further complicate decision-making processes. Existing registries lack comprehensive data integration, focusing on isolated perspectives of either research or clinical data, neglecting the holistic patient profile. Evaluation of treatment endpoints beyond Progression-Free Survival (PFS) and Overall Survival (OS) remain inconsistent, despite being a priority for families. The DMG Learning Hub emerges as a transformative platform designed to empower families by offering informed options and guidance while fostering a collaborative environment where patient families actively contribute to shaping the landscape of clinical trials. Leveraging AI-assisted matching, the platform assists families in navigating the complex decision-making process surrounding clinical trial participation. The inclusion of a patient-reported outcome journal feature enables families to track vital information such as symptom changes and supplement usage, with autonomy over the timing of shared data. Using real-time data sharing facilitates invaluable insights for researchers, bridging gaps that would otherwise be overlooked and enriching the collective understanding of DMG. Harmonizing molecular, imaging, and clinical data within the platform equips care teams with comprehensive, near-real-time information crucial for personalized treatment strategies. This integrated approach not only supports the inclusion of n-of-1 arms in current trials but also extends therapeutic opportunities to patients previously ineligible for studies due to specific disease markers or progression patterns. The DMG Learning Hub represents a significant stride towards precision oncology by synergizing patient-centric priorities with cutting-edge technology, integrating harmonized data, and fostering collaborative efforts, the initiative heralds a promising era of personalized treatment for DMG and beyond the neuro-oncology community.

TMIC-14. RAPID EARLY PROGRESSION IN GLIOBLASTOMA: A COMPREHENSIVE REVIEW OF CLINICAL, IMAGING AND HISTOPATHOLOGICAL FEATURES

Abstract BACKGROUND In glioblastoma patients, Rapid Early Progression (REP) refers to tumour growth between the time of surgery and postoperative chemoradiotherapy. Despite its high incidence relatively little is known about it. The aim of this study was to characterise REP using a combination of clinical, imaging and histopathology data. METHODS Ethical approval was obtained. Supratentorial IDH-wildtype glioblastomas undergoing gross-total resection (GTR) between 2015-2021 were included. REP was defined as a nodular increase in enhancement located within or around the surgical cavity, occurring within 6 weeks of surgery, that was not accounted for by initial areas of diffusion-restriction. REP cases were propensity-score matched to controls on baseline characteristics. Imaging was compared using mean apparent diffusion coefficient (ADC) and T1-postcontrast radiomics. Tissue analyses included DNA methylation subtyping and immunohistochemistry for markers of T-cells (CD3), microglia (P2RY12) and proliferation (PHH3). RESULTS REP occurred in 8/60 GTR cases (13%) and these patients were matched to 16 controls. REP cases had a worse preoperative performance status (median Eastern Cooperative Oncology Group [ECOG] 2 vs. 1), overall-survival (p=0.038) and progression-free survival (p = 0.029). REP was more common in patients over 65 years (27%) and significantly fewer REP patients received radical postoperative chemoradiotherapy (p=0.023). REP and control cases were well matched with respect to preoperative T1-postcontrast radiomics and mean ADC in enhancing (p=0.99) and non-enhancing regions (p=0.67). There was no relative excess of any one DNA methylation subtype in REP cases. The two groups had a similar proportion of T-cells (p=0.81), but REP cases had a trend towards a higher proportion of microglia (31% vs. 17%; p=0.22), and a lower proliferation rate (3% vs. 9%; p=0.19). CONCLUSION REP cases had a worse performance status from diagnosis and received less radical postoperative chemoradiotherapy protocols compared to controls, yielding a significantly shorter median OS/PFS. Differences in immune landscape and proliferation rate should be further investigated.

Patient centered HCC surveillance - complementary roles of ultrasound and CT/MRI.

Hepatocellular carcinoma (HCC) is a leading cause of cancer-related mortality worldwide and is the fastest growing cause of cancer death in the United States (U.S.) In the U.S., current national clinical practice guidelines from the 2023 American Association for the Study of Liver Diseases (AASLD) Practice Guidance and the recently updated Liver Imaging Reporting & Data Systems (LI-RADS) Ultrasound (US) Surveillance v2024 core recommend semi-annual serum α-fetoprotein and US screening of patients deemed to be high risk for developing HCC. In this article, we will explore the transition to a patient-centered approach to HCC surveillance, including the role of the new LI-RADS US Surveillance v2024 core and the use of visualization score for determining ultrasound quality, the known risk factors for poor US image quality, and the potential options for alternative surveillance strategies when US may not be a viable option for certain patients, including multiphasic computed tomography (CT), magnetic resonance imaging (MRI), and several abbreviated MRI protocols.

NIMG-60. SUBTYPING GLIOBLASTOMA FROM JOINT CLUSTERING OF IMAGING AND GENOMIC DATA

Abstract PURPOSE To comprehensively unravel the heterogeneity of glioblastoma, we developed a subtyping method using integrated imaging and genomic data. METHODS We assembled a retrospective cohort of 571 IDH-wildtype glioblastoma patients, obtaining pre-operative multi-parametric MRI (T1, T1-GD, T2, T2-FLAIR, DSC, DTI) scans (available for 462 patients) and targeted next-generation sequencing (NGS) data (available for 355 patients). We extracted 971 radiomic features from these MRI scans, selecting 12 significant dimensions through L21-norm minimization guided by 13 key driver genes from the five most frequently altered pathways in glioblastoma compared to healthy controls (RB1, P53, MAPK, PI3K, and RTK). Subtypes were identified using a joint clustering approach that integrated radiomics and genomics. RESULTS Our method identified three glioblastoma subtypes with distinct risk levels: high-risk (subtype 1), medium-risk (subtype 2), and low-risk (subtype 3), each characterized by their differential overall survival outcomes (Kaplan-Meier analysis, p &amp;lt; 0.05). The intensities of axial diffusivity in enhancing tumor regions and radial diffusivity in non-enhancing cores increased across three subtypes. All subtypes shared a common tendency for more frequent mutation co-occurrence pattern in [TP53,RB1] but exhibited unique molecular characteristics. For example, subtype 1 displayed a relatively high frequency of co-occurring mutations in [TP53,KDR] and [NOTCH2,MDM4]; subtype 2 had six co-occurring pairs [TP53,KDR], [PTPN11,NF1], [RB1,FUBP1], [PTEN,CIC], [PTEN,KRAS], and [NOTCH2,CDKN2A]; subtype 3 showed six co-occurring pairs [PTPN11,NF1], [PDGFRA,NF1], [NTRK1,NOTCH2], [MET,ATRX], [KRAS,KDR], and [FGFR2,DNMT3A] plus four significant patterns of mutual exclusivity in [TP53,PTPN11], [TP53,EGFR], [EGFR,RB1], and [EGFR,NF1]. CONCLUSION Our study discovered distinct glioblastoma subtypes, revealing complex patterns and structures across multiple data modalities without relying on prior clinical assumptions. This is promising to help enhance the precision of diagnosis and treatment of glioblastoma.

CNSC-35. TEMOZOLOMIDE POTENTIALLY POSTPONES THE DEVELOPMENT OF SUBSEQUENT METASTASES IN PEDIATRIC DIFFUSE INTRINSIC PONTINE GLIOMA (DIPG) PATIENTS

Abstract OBJECTIVE Surgical intervention is thought to increase the likelihood of subsequent dissemination in diffuse intrinsic pontine glioma (DIPG) patients. However, there is no study evaluating the factors influencing the timing of metastasis development. Hence, the present analysis was conducted to shed light on factors associated with subsequent dissemination timing. METHODS This analysis included 28 pediatric (age≤12) DIPG cases with metastasis treated at Sanjiu Brain Hospital between June 1, 2017 and June 30, 2023. The clinical data were retrospectively collected following the initial diagnosis. The diagnosis was established by typical imaging features of DIPG. The same three radiologists analyzed the imaging data independently. Time to subsequent metastasis was estimated using Kaplan–Meier analysis, and a log-rank test evaluated the differences between groups. RESULTS Of 28 patients, 13 (46.4%) were males and 15 were females. 23 (82.1%) underwent surgical intervention, while 5 (17.9%) did not receive surgery. The patients’ median age was 7 years (2-12 yrs.). The Kaplan-Meier method showed that the median time to metastasis was 5.8 months (95%CI 4.2-9.1). The log-rank test demonstrated that adjuvant temozolomide was the only variable that could significantly delay metastasis in the pediatric DIPG population (P=0.03). The median time to metastasis from the day of initial diagnosis for those who received adjuvant temozolomide was 6.8 months (95%CI 5-10.7) vs. 3 months for those who did not receive adjuvant temozolomide (95%CI 2.3-NA). Sex, age, surgical intervention, and concurrent irradiation and temozolomide were not associated with metastasis timing. CONCLUSION Adjuvant temozolomide administration might benefit DIPG patients by delaying metastasis, especially in those who are at higher risk for subsequent metastasis, such as those who underwent surgical intervention. However, prospective studies are warranted to confirm these results.

Evaluating the effectiveness of AI-powered UrologiQ's in accurately measuring kidney stone volume in urolithiasis patients.

Kidney stones and urolithiasis are kidney diseases that have a significant impact on health and well-being, and their incidence is increasing annually owing to factors such as age, sex, ethnicity, and geographical location. Accurate identification and volume measurement of kidney stones are critical for determining the appropriate surgical approach, as timely and precise treatment is essential to prevent complications and ensure successful outcomes. Larger stones often require more invasive procedures, and precise volume measurements are essential for effective surgical planning and patient outcomes. This study aimed to compare the ability of artificial intelligence (AI) to detect and measure kidney stone volume via CT-KUB images. CT KUB imaging data were analyzed to determine the effectiveness of AI in identifying the volume of kidney stones. The results were compared with measurements taken by radiologists. Compared with radiologists, the AI had greater accuracy, efficiency, and consistency in measuring kidney stone volume. The AI calculates the volume of kidney stones with an average difference of 80% compared with the volumes calculated by radiologists, highlighting a significant discrepancy that is critical for accurate surgical planning. The results suggest that artificial intelligence (AI) outperforms radiologists' manual calculations in measuring kidney stone volume. By integrating AI with kidney stone detection and treatment, there is potential for greater diagnostic precision and treatment effectiveness, which could ultimately improve patient outcomes.

The Brain Image Library: A Community-Contributed Microscopy Resource for Neuroscientists.

Advancements in microscopy techniques and computing technologies have enabled researchers to digitally reconstruct brains at micron scale. As a result, community efforts like the BRAIN Initiative Cell Census Network (BICCN) have generated thousands of whole-brain imaging datasets to trace neuronal circuitry and comprehensively map cell types. This data holds valuable information that extends beyond initial analyses, opening avenues for variation studies and robust classification of cell types in specific brain regions. However, the size and heterogeneity of these imaging data have historically made storage, sharing, and analysis difficult for individual investigators and impractical on a broad community scale. Here, we introduce the Brain Image Library (BIL), a public resource serving the neuroscience community that provides a persistent centralized repository for brain microscopy data. BIL currently holds thousands of brain datasets and provides an integrated analysis ecosystem, allowing for exploration, visualization, and data access without the need to download, thus encouraging scientific discovery and data reuse.

Automated Vertebral Bone Quality Determination from T1-Weighted Lumbar Spine MRI Data Using a Hybrid Convolutional Neural Network–Transformer Neural Network

Vertebral bone quality (VBQ) is a promising new method that can improve screening for osteoporosis. The drawback of the current method is that it requires manual determination of the regions of interest (ROIs) of vertebrae and cerebrospinal fluid (CSF) by a radiologist. In this work, an automatic method for determining the VBQ is proposed, in which the ROIs are obtained using a trained neural network model. A large, publicly available dataset of sagittal lumbar spine MRI images with ground truth segmentations was used to train a BRAU-Net++ hybrid CNN–transformer neural network. The performance of the trained model was evaluated using the dice similarity coefficient (DSC), accuracy, precision, recall and intersection-over-union (IoU) metrics. The trained model performed similarly to state-of-the-art lumbar spine segmentation models, with an average DSC value of 0.914 ± 0.007 for the vertebrae and 0.902 for the spinal canal. Four different methods of VBQ determination with automatic segmentation are presented and compared with one-way ANOVA. These methods use different algorithms for CSF extraction from the segmentation of the spinal canal using T1- and T2-weighted image data and applying erosion to the vertebral ROI to avoid a sharp change in SI at the edge of the vertebral body.

Noninvasive Monitoring of Steatotic Liver Disease in Western Diet-Fed Obese Mice Using Automated Ultrasound and Shear Wave Elastography.

Ultrasound imaging and shear wave elastography (SWE) can be used to noninvasively stage hepatopathologies and are widespread in clinical practice. These techniques have recently been adapted for small animal use in a novel 3D invivo imaging system capable of high-throughput automated scanning. Our goal was to evaluate the feasibility of using this imaging tool in the murine Western diet (WD) model, a highly translatable preclinical model of obesity, metabolic disease and liver fibrosis. Female C57BL/6 mice (N = 48) were placed on WD or chow diet and imaged longitudinally for a period of 48 weeks. Imaging consisted of 3D B-mode and targeted SWE captures. Liver volume, liver echogenicity and liver stiffness were quantified from invivo imaging data. A subset of mice was sacrificed at various timepoints (0, 12, 24 and 48 weeks) for histological workup. Correlation analysis was performed between invivo imaging and histological measurements to determine level of agreement. Noninvasive imaging showed statistically significant increases in liver volume and echogenicity, but non-significant increase in liver stiffness in the WD-fed cohort, suggesting development of hepatomegaly and steatosis, but negligible fibrosis. Exvivo analysis confirmed significant increases in liver weight, liver triglycerides and ALT, but limited increases in fibrosis corroborating noninvasive imaging results. Correlation analysis between imaging and histology demonstrated good agreement between liver volume/liver weight (R2 = 0.85) and echogenicity/triglycerides (R2 = 0.76). This study demonstrated that noninvasive ultrasound liver assessments are feasible in the WD mouse model and closely reflect the underlying pathological state of the animal. Automated ultrasound can serve as a high-throughput noninvasive screening method for preclinical liver disease research and drug development.

Enhancing wildfire detection: a novel algorithm for controllable generation of wildfire smoke images

Background The lack of wildfire smoke image data is one of the most important factors hindering the development of image-based wildfire detection. Smoke image generation based on image inpainting techniques is a solution worthy of study. However, it is difficult to generate smoke texture with context consistency in complex backgrounds with current image inpainting methods. Aims This work aims to provide a wildfire smoke image database for specific scenarios. Methods We designed an algorithm based on generative adversarial networks (GANs) to generate smoke images. The algorithm includes a multi-scale fusion module to ensure consistency between the generated smoke and backgrounds. Additionally, a local feature-matching mechanism in the discriminator guides the generator to capture real smoke’s feature distribution. Key results We generated 13,400 wildfire smoke images based on forest background images and early fire simulation from the Fire Dynamics Simulator (FDS). Conclusions A variety of advanced object detection algorithms were trained based on the generated data. The experimental results confirmed that the addition of the generated data to the real datasets can effectively improve model performance. Implications This study paves a way for generating object datasets to enhance the reliability of watchtower or satellite wildfire monitoring.

NIMG-06. AUTOMATED VOLUMETRY OF MENINGIOMAS IN CONTRAST-ENHANCED T1-WEIGHTED MRI USING DEEP LEARNING

Abstract BACKGROUND With the widespread use of MRI equipment and brain docks, there are more opportunities to perform follow-up observation for meningiomas. In recent years, attempts have been made to predict prognosis by using tumor volume as a highly objective imaging biomarker. On the other hand, manual lesion extraction by a reader with anatomical knowledge is indispensable for volume measurement, which requires a certain amount of labor and time. This is a major barrier to the generation of evidence and clinical application in this field. OBJECTIVE To establish the clinical significance of tumor volume as an imaging biomarker in meningiomas, and to develop a simple and highly reproducible application for automated volume measurement using artificial intelligence, and to investigate its accuracy and potential for clinical application. METHODS 178 patients with meningiomas who underwent MRI at Osaka University Hospital were included in the study. Contrast-enhanced T1-weighted images were used for manual lesion extraction by two neurosurgeons. These annotated image data were randomly divided into training (90%) and testing (10%). The deeplabV3 model was used to train the artificial intelligence, and leave-one-out cross-validation was used to train and validate the automated model. The test data were used for comparison, and the agreement rate (Dice index) on the images of the automatically extracted regions and the error rate of the calculated tumor volume (mean square error rate: RMSPE) were used as indices of measurement performance. RESULTS The average agreement rate between the automatically extracted area and the test data was 92.3±5.1%. The error rate of tumor volume was 9.5±1.2%. CONCLUSION This application using artificial intelligence was considered to have a certain validity in the accuracy of automatic lesion extraction. This method, which enables simple tumor volume measurement, is expected to create new imaging biomarkers that will contribute to the optimization and personalization of meningioma treatment.

SFNet: Spatial and Frequency Domain Networks for Wide-Field OCT Angiography Retinal Vessel Segmentation.

Automatic segmentation of blood vessels in fundus images is important to assist ophthalmologists in diagnosis. However, automatic segmentation for Optical Coherence Tomography Angiography (OCTA) blood vessels has not been fully investigated due to various difficulties, such as vessel complexity. In addition, there are only a few publicly available OCTA image data sets for training and validating segmentation algorithms. To address these issues, we constructed a wild-field retinal OCTA segmentation data set, the Retinal Vessels Images in OCTA (REVIO) dataset. Second, we propose a new retinal vessel segmentation network based on spatial and frequency domain networks (SFNet). The proposed model are tested on three benchmark data sets including REVIO, ROSE and OCTA-500. The experimental results show superior performance on segmentation tasks compared to the representative methods.

Neutrophil-albumin ratio serves as a superior prognostic biomarker for traumatic brain injury.

Traumatic brain injury (TBI) represents a common and severe medical condition necessitating prompt risk stratification to enhance patient outcomes. Although substantial research has been conducted on the prognostic utility of various biomarkers for TBI, no single biomarker has been definitively recognized as the most precise predictor of disease outcomes. In comparison to other markers, the neutrophil-albumin ratio (NAR) has emerged as a cost-effective and reproducible inflammatory biomarker, demonstrating potential in evaluating the severity of inflammation and prognosticating outcomes in infections and cerebrovascular diseases. This study evaluated the prognostic significance of the NAR in comparison to two other readily accessible and cost-effective composite indices: the Neutrophil-Lymphocyte Ratio (NLR) and the Platelet-Lymphocyte Ratio (PLR) in individuals with TBI. We conducted a retrospective cohort analysis involving 297 hospitalized TBI patients, gathering comprehensive demographic, anthropometric, medical, clinical, laboratory, and imaging data to assess the expression changes of these biomarkers. Our findings suggest that both the NAR and the NLR possess predictive value regarding prognosis following TBI. However, receiver operating characteristic (ROC) curve analysis revealed that NAR outperformed NLR as a prognostic predictor. In conclusion, our examination of blood biochemistry composite indicators indicates that, while both NAR and NLR serve as significant prognostic markers, NAR is a more effective predictor of outcomes in patients with TBI.

Influence of exposure time on the aero-optical effects of a high-speed aircraft optical window

Under high-speed conditions (Mach number &gt; 3), the turbulent shear layer formed on the optical window causes phase changes in the light, resulting in the aero-optical effects, which significantly affect the imaging quality at different exposure times, τ. In this study, imaging data of an optical window were collected in a Mach 3.8 supersonic low-noise wind tunnel for a facula and a USAF 1951 resolution board, as well as distorted wavefront data of visible light. As τ increases from 0.2 ms to 5 ms, the centroid jitter of the facula shows an overall decreasing trend and the increase of τ mainly inhibits the jitter in the spanwise direction. While τ varies from 0.2 ms to 28 ms, structural analysis indicated that the structural similarity and stability of imaging improves and this improvement has an upper limit of τ0 = 22 ms, coined as the saturation exposure time. Normalized modulation transfer function curves derived from USAF 1951 images show a decrease in the imaging resolution as τ increases. The average wavefront distortion can be reduced by up to 71.7% as τ increases from 5 ms to 30 ms, while the standard deviation of wavefront distortion initially decreases and then stabilizes at 0.0025λ.