A multi-path 2.5 dimensional convolutional neural network system for segmenting stroke lesions in brain MRI images

Keyphrases

• Magnetic Resonance Imaging Image 100%
• Multipath 100%
• Convolutional Neural Network 100%
• Human Expert 100%
• Neural Network System 100%
• Wisconsin 100%
• Brain Magnetic Resonance Imaging 100%
• Stroke Lesion 100%
• Medical Colleges 100%
• 2.5-dimensional 100%
• U-Net 100%
• Dice Coefficient 66%
• Anatomical Tracing 66%
• Stroke Lesion Segmentation 66%
• Post-stroke 33%
• Brain-behavior Relationships 33%
• Publicly Available 33%
• Automatic Identification 33%
• Wide Applicability 33%
• Modal 33%
• Patient Treatment 33%
• Concatenated 33%
• Training Data 33%
• Model Performance 33%
• 3D Volume 33%
• Magnetic Resonance Imaging Scan 33%
• Inter-rater 33%
• Order of Accuracy 33%
• Stroke Survivors 33%
• Random Forest 33%
• Validation Data 33%
• Treatment Planning 33%
• Brain Scans 33%
• Brain Lesions 33%
• Cross-study Validation 33%
• Validation Accuracy 33%
• 3D Convolutional Neural Network (3D CNN) 33%
• Patient Diagnosis 33%
• Complete Overlap 33%
• Independent Training 33%
• Small Lesions 33%
• Rater Accuracy 33%
• Ischaemic Aetiology 33%
• Naïve Bayes 33%
• Dice Value 33%

Neuroscience

• Magnetic Resonance Imaging 100%
• Neural Network 100%
• Behavior (Neuroscience) 25%
• Brain Lesion 25%