Foot fractures diagnosis using a deep convolutional neural network optimized by extreme learning machine and enhanced...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Orthopedics

Background:

• Foot fractures pose diagnostic challenges.
• Accurate and efficient diagnosis is crucial for effective treatment.
• Existing methods may have limitations in accuracy or speed.

Purpose of the Study:

• To propose a novel hybrid methodology for the diagnosis of foot fractures.
• To develop an efficient deep learning and metaheuristic model for foot fracture detection.
• To enhance the accuracy and efficiency of foot fracture diagnosis.

Main Methods:

• A hybrid approach combining deep learning (pre-trained ZFNet) and metaheuristics (enhanced snow ablation optimizer - ESAO).
• Integration of Extreme Learning Machine (ELM) in the final layers of the ZFNet.
• Optimization of the ELM component using the developed ESAO metaheuristic.
• Application and validation on a standard Institutional Review Board (IRB) benchmark dataset.

Main Results:

• The proposed ZFNet/ELM/ESAO model demonstrated high efficiency in diagnosing foot fractures.
• Comparative analysis showed competitive or superior performance against methods like DT/KNN, LDA, FRCNN, TL-ECNN, and DCNN/LSTM.
• The hybrid approach effectively extracts features and classifies foot fractures.

Conclusions:

• The ZFNet/ELM/ESAO-based model is an effective tool for foot fracture diagnosis.
• The hybrid methodology offers a promising advancement in medical imaging for orthopedic conditions.
• This approach can significantly aid clinicians in the accurate and timely diagnosis of foot fractures.