Publications by authors named "Alexandra J Golby"
Article Synopsis
- The retinogeniculate visual pathway (RGVP) transmits visual data from the retina to the brain, making its identification crucial for understanding vision-related conditions.
- DeepRGVP, a new deep learning framework, allows efficient and precise mapping of the RGVP from diffusion MRI data, overcoming the challenges of manual tract selection.
- Experiments demonstrate DeepRGVP's superior accuracy in identifying the RGVP compared to existing methods, even in patients with lesions, highlighting the promise of deep learning in improving visual pathway analysis.
Article Synopsis
- The study aimed to analyze the global neurosurgery training workforce, focusing on the types of training programs, trainee support, diversity of experiences, and accreditation processes.
- Data were collected from 187 countries and 25 territories, revealing an estimated 1,261 training programs and over 10,500 trainees, with significant variations in density based on country income levels and WHO regions.
- High-income countries had a much higher density of trainees (0.48 per 100,000 people) compared to lower-income countries, which faced challenges like limited subspecialty training and resources.
Article Synopsis
- The study aimed to assess the global distribution and growth of consultant neurosurgeons, especially in low- and middle-income countries, to understand the impact of efforts to expand access to neurosurgery worldwide.
- A comprehensive electronic survey was conducted across 192 countries and 25 territories, revealing an estimated 72,967 neurosurgeons, with a global density of 0.93 per 100,000 people, and significant variation by income level.
- Findings indicated that neurosurgeon density increases with national income level, highlighting the shortage in low-income areas, particularly in WHO African and Southeast Asia regions, while the Western Pacific region showed the highest density of neurosurgeons.
Article Synopsis
- The study explored the effect of perampanel, an AMPA receptor antagonist, on reducing hyperexcitability around gliomas, which could promote tumor growth.
- An open-label trial compared perampanel with standard care in patients with high-grade glioma undergoing surgery, measuring outcomes like high-frequency oscillation rates and seizure occurrence.
- Results showed no significant difference in hyperexcitability outcomes between perampanel and standard care, and early termination of the trial indicated similar seizure rates and overall survival for both treatments.
Article Synopsis
- The study introduces Fiber Microstructure Quantile (FMQ) Regression, a new statistical method designed to analyze the relationship between brain white matter fiber bundles and factors like cognitive scores.
- FMQ Regression uses quantile regression to examine different regions of white matter based on fractional anisotropy (FA), considering sex differences in brain-behavior associations.
- The analysis, carried out on data from the Human Connectome Project, shows that FMQ Regression is more effective than traditional methods in identifying significant links between brain structure and cognitive performance, highlighting unique effects for males and females.
Article Synopsis
- Presurgical planning for brain tumor removal is essential to avoid post-operative neurological deficits, with the use of advanced brain mapping techniques gaining prominence.
- Functional MRI (fMRI) helps identify important brain regions for motor, language, and visual functions, but some patients may struggle to perform tasks during imaging due to existing impairments.
- Connectome fingerprinting (CF) uses machine learning to predict task-related brain activations based on resting-state fMRI data, demonstrating high accuracy in predicting motor functions, which could be beneficial when task data is difficult to obtain.
Article Synopsis
- The study investigates methods to disrupt the blood-brain barrier (BBB) in glioblastoma patients to enhance the effectiveness of temozolomide chemotherapy through the use of focused ultrasound and microbubbles.
- Data was collected from 38 treatment sessions involving nine patients, assessing the impact of controlled ultrasound exposure on MRI imaging, including measures of BBB disruption and the associated vascular damage.
- Results indicated that modifying treatment parameters improved BBB disruption quality while minimizing side effects, with established correlations between cavitation levels and treatment outcomes, although variability among patients remained significant.
Article Synopsis
- The study explores using holographic methods with a 1024-element phased array transducer to control ultrasound fields, targeting enhanced drug delivery through blood-brain barrier (BBB) disruption.
- The method involves real-time acoustic simulations for each transducer element, allowing precise modulation to create desired ultrasound patterns, tested both in tissue phantoms and live subjects.
- Results indicate this approach could improve the efficiency and effectiveness of BBB disruption by enlarging focal volumes, reducing targets needed, and potentially minimizing unwanted reflections in clinical treatments.
Article Synopsis
- The study investigated the effects of the AMPA receptor antagonist perampanel on hyperexcitability and clinical outcomes in patients undergoing surgery for high-grade glioma.
- The trial compared perampanel to standard care using levetiracetam, measuring intraoperative hyperexcitability through high-frequency oscillation (HFO) rates and tracking seizure-free outcomes and overall survival.
- Results indicated no significant difference in hyperexcitability or survival outcomes between the two treatments, leading to the early termination of the trial due to futility, while perampanel was found to be safe and well-tolerated.
Article Synopsis
- Shape is crucial in computer graphics and brain imaging, helping to interpret the structure and function of the human brain's morphology.
- Researchers analyze 3D white matter connections using diffusion magnetic resonance imaging (dMRI) and extract shape descriptors alongside traditional features.
- The Shape-fused Fiber Cluster Transformer (SFFormer) model shows promise in predicting language performance based on brain shape, connectivity, and microstructure, indicating a strong link between the brain's structural connections and cognitive functions like language.
Article Synopsis
- TractGeoNet is a new system that uses advanced techniques from geometry and deep learning to analyze brain images and predict how well people do on language tests.
- Unlike older methods, it looks at all details of brain fibers at once instead of averaging them, which helps make better predictions.
- The system also highlights important areas of the brain that affect language skills, showing that certain brain regions, like the left arcuate fasciculus, are key to understanding language performance.
Learning Expected Appearances for Intraoperative Registration during Neurosurgery.
Med Image Comput Comput Assist Interv
October 2023
Article Synopsis
- We introduced a new technique for aligning intraoperative images with patient-specific views using preoperative imaging and expected appearances.
- Our method improves camera pose estimation by comparing intraoperative images with synthesized expected views, effectively addressing issues related to low-quality images during surgery.
- Testing on both synthetic and clinical data showed that our approach surpassed existing methods, achieving accuracies that fulfill clinical requirements in brain surgery.
Article Synopsis
- The retinogeniculate visual pathway (RGVP) transmits visual information from the retina to the brain, and understanding its anatomy is crucial for studying visual system diseases.
- Diffusion MRI (dMRI) tractography is a technique for mapping the RGVP, but current methods for identifying it are slow and subjective, relying on manual expert input.
- This paper introduces a deep learning framework called DeepRGVP, which uses advanced techniques to quickly and accurately identify the RGVP from dMRI data while maintaining effectiveness even in cases of brain lesions, demonstrating the promise of AI in medical imaging.
Article Synopsis
- The corticospinal tract (CST) is essential for voluntary body movement control and is organized by body part representation.
- This study evaluates six tractography methods, including probabilistic and deterministic approaches, to analyze the CST's structure and organization using diffusion MRI data from 100 subjects.
- The unscented Kalman filter approach (UKF2T) shows the best performance in reconstructing the CST and is notably linked to the volume of motor cortex areas for different body parts, highlighting the importance of advanced imaging techniques in understanding brain function.
Deep Learning for Detection and Localization of B-Lines in Lung Ultrasound.
IEEE J Biomed Health Inform
September 2023
Article Synopsis
- Lung ultrasound (LUS) is used by emergency physicians to assess pulmonary congestion, with B-line artifacts being crucial indicators of this condition.
- A new dataset, BEDLUS, consisting of 1,419 videos from 113 patients and 15,755 expert-annotated B-lines, has been created to evaluate various deep learning methods for automated B-line detection.
- Results show promising performance, with detection methods achieving an area under the curve ranging from 0.864 to 0.955, and the introduction of a "single-point" B-line localization method performing comparably to human observer agreement.
Article Synopsis
- The study focuses on the GCS-Pupil score (GCS-P) as a tool to predict hospital mortality in Brazilian patients with severe traumatic brain injury (TBI), comparing its effectiveness to a model that includes additional clinical and radiological factors.
- Data from 1,066 patients were analyzed, revealing an overall hospital mortality rate of 32.8%, with the GCS-P score showing an accuracy (AUROC) of 0.73, compared to 0.80 for the more comprehensive model.
- The results indicate that the GCS-P can be used to predict hospital mortality in severe TBI cases, but further research is needed to assess its effectiveness for long-term outcomes in TBI patients
Article Synopsis
- Diffusion magnetic resonance imaging white matter tractography helps plan brain surgeries by aiming to excise tumors while preserving brain function, but faces challenges due to limitations in standard imaging techniques.
- Track density imaging (TDI) offers enhanced visualization of brain connections over traditional diffusion tensor imaging (DTI), potentially improving surgical planning.
- This study compared TDI-based maps to DTI-based maps using patient data, with five neurosurgeons evaluating their utility in pre-surgical planning.
Article Synopsis
- Surgical resection remains the main approach for treating brain tumors, but traditional navigation methods struggle with brain movement during surgery.
- A new system improves surgical guidance by using the positions of instruments to create an accurate map of tumor removal in real-time, evaluated through 80 surgeries.
- While the system estimates residual tumors well under ideal conditions, it still provides valuable insights even when brain shift occurs, helping surgeons detect areas that may need further attention.
Article Synopsis
- Diffusion MRI tractography maps the brain's white matter connections and divides these into clusters for better visualization, but existing methods struggle with the complex superficial white matter (SWM).
- A new framework called Superficial White Matter Analysis (SupWMA) uses deep learning to efficiently and consistently identify 198 SWM clusters from brain imaging data.
- SupWMA outperforms existing techniques in accuracy, consistency, and speed across various age groups and health conditions, demonstrating its potential for broader clinical application.
Article Synopsis
- The study reviews global cancer surgery literature since a key report in 2015, focusing on articles from low-resource countries.
- It performs a SWOT analysis identifying themes such as local research, resource limitations, and the impact of COVID-19 on surgical care.
- The findings aim to offer strategies to improve cancer surgery in underserved areas while addressing significant challenges and opportunities in the field.
Article Synopsis
- A recent study used fMRI to explore whether gliomas show synchronized neurovascular activity with other brain regions, examining its potential relationship with patient survival.
- The research analyzed resting-state fMRI scans from 54 patients, focusing on functional connectivity (FC) between the tumor and the rest of the brain, while also evaluating how well this connectivity could predict overall survival (OS).
- Findings indicated significant BOLD signal synchronization in both newly diagnosed and recurrent glioma patients, with stronger glioma-brain connectivity correlating positively with individual survival rates, outperforming traditional clinical and genetic prognostic factors.
Article Synopsis
- - The case study presents a 74-year-old man with severe essential tremor and diffuse skull thickening (hyperostosis), who did not improve after undergoing magnetic resonance-guided focused ultrasound (MRgFUS) treatment.
- - Pre-treatment imaging indicated a low skull density ratio (SDR) and the presence of tricortical hyperostosis, yet post-treatment MRI showed no changes or lesions.
- - A literature review revealed that out of three similar cases of patients with hyperostosis, all also failed MRgFUS treatment, suggesting that tricortical hyperostosis may be a contraindication for this therapeutic approach.