Efficient segmentation of active and inactive plaques in FLAIR-images using DeepLabV3Plus SE with efficientnetb0...

Area of Science:

• Medical Image Analysis
• Artificial Intelligence in Healthcare
• Neurology

Background:

• Multiple Sclerosis (MS) is characterized by active and inactive brain plaques.
• Accurate segmentation of these plaques in FLAIR MRI is crucial for diagnosis and monitoring.
• Existing segmentation methods may lack efficiency and precision.

Purpose of the Study:

• To introduce an efficient Convolutional Neural Network (CNN) model for segmenting active and inactive plaques in FLAIR images.
• To evaluate the performance of the proposed DeepLabV3Plus SE with EfficientNetB0 backbone model against other CNN architectures.
• To enhance plaque segmentation capabilities through advanced pre-processing and network components.

Main Methods:

• Dataset pre-processing: cranial bone removal, image resizing, normalization.
• Utilized DeepLabV3Plus SE with EfficientNetB0 backbone, incorporating Squeeze and Excitation (SE-Block) and atrous spatial pyramid pooling.
• Trained the model on 1500 labeled MRI slices from 100 MS patients using Dice coefficient loss and Adam optimization.

Main Results:

• The proposed DeepLabV3+SE_EfficientNetB0 model achieved high performance metrics.
• Key results include: Intersection over Union (IOU) of 69.87%, Dice Score of 76.24%, Precision of 88.89%, Recall of 73.52%, and F1-Score of 80.47%.
• Demonstrated superior segmentation accuracy compared to other evaluated CNN architectures.

Conclusions:

• The developed DeepLabV3Plus SE with EfficientNetB0 offers an efficient and accurate approach for MS plaque segmentation in FLAIR images.
• This method shows significant potential for improving medical image analysis and diagnostic accuracy in Multiple Sclerosis.
• The study highlights the effectiveness of advanced CNN architectures and pre-processing techniques in neuroimaging applications.