Duple-MONDNet: duple deep learning-based mobile net for motor neuron disease identification

Area of Science:

• Neuroscience
• Medical Imaging
• Artificial Intelligence

Background:

• Motor neuron disease (MND) is a rare neurological disorder progressively impairing voluntary muscle control.
• Early diagnosis of MND is challenging and time-consuming with current manual methods.
• Identifying MND early is crucial for potential intervention and management.

Purpose of the Study:

• To develop a novel deep learning framework for the early detection of motor neuron disease (MND).
• To enhance the accuracy and efficiency of MND diagnosis using neuroimaging data.

Main Methods:

• Diffusion tensor imaging (DTI) images were analyzed using dual feature extraction, combining color and textural information.
• Local binary pattern (LBP) methods extracted textural features, while color features were incorporated during classification.
• A deep learning model, MONDNet, was developed to classify normal and abnormal MND cases based on extracted features.

Main Results:

• The proposed MONDNet achieved a high detection rate of 99.66% for early-stage MND.
• The model demonstrated effectiveness in identifying MND based on combined color and texture features from DTI scans.

Conclusions:

• The deep learning-based MONDNet framework shows significant promise for the early and accurate detection of motor neuron disease.
• This approach offers a more efficient alternative to manual diagnostic methods for MND.