Deep learning enabled brain shunt valve identification using mobile phones

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Radiology

Background:

• Accurate identification of implanted medical devices is critical for patient safety during Magnetic Resonance Imaging (MRI).
• Current methods for identifying these devices are challenging and time-consuming.
• Deep learning offers a potential solution for faster and more accurate device detection.

Purpose of the Study:

• To develop and evaluate a deep learning pipeline for identifying implanted programmable cerebrospinal fluid shunt valves using X-ray images.
• To improve the speed and accuracy of medical device detection prior to MRI examinations.
• To compare the performance of the proposed deep learning method against existing techniques.

Main Methods:

• A convolutional neural network (CNN) was developed to identify shunt valves from X-ray images.
• X-ray images were captured using mobile phone cameras at various angles and lighting conditions.
• The proposed CNN was compared with methods using digitally transferred images and transfer learning on mobile phone images.

Main Results:

• The proposed deep learning approach achieved high accuracy (95%) in identifying shunt valves from mobile phone X-ray images.
• The system significantly outperformed existing methods, including those using transfer learning (94% accuracy).
• Performance metrics such as Average Precision, Recall, and F1-score were substantially higher for the proposed method.

Conclusions:

• An automated shunt valve identification system using deep learning is a valuable safety tool for radiologists.
• This mobile-deployable system enhances the efficiency of coordinating patient care with implanted devices.
• The proposed image-based system offers greater ease of integration into clinical workflows compared to traditional methods.