Automatic Assessment of Radiological Parameters of the Distal Radius Using a Hybrid Approach Combining Deep Learning...

Area of Science:

• Radiology
• Medical Imaging
• Artificial Intelligence

Background:

• Deep learning in medical imaging is rapidly advancing.
• Automated analysis of wrist radiographs is crucial for accurate diagnosis.
• Existing methods may require significant manual input.

Purpose of the Study:

• To develop an automated, hybrid approach for detecting anatomical landmarks and measuring radiological parameters in wrist radiography.
• To combine deep learning with conventional computer-aided diagnosis for enhanced accuracy.
• To evaluate the performance of the developed method.

Main Methods:

• A hybrid method combining deep learning and computer-aided diagnosis was developed.
• 487 wrist radiographs were used for training and validation, with 100 for testing.
• Anatomical landmarks for radial inclination (RI), radial length (RL), volar tilt (VT), and ulnar variance (UV) were identified and measured.

Main Results:

• The model achieved a successful detection rate (SDR) of 97-99%.
• Mean absolute errors (MAEs) for RI, RL, VT, and UV were within acceptable ranges (e.g., 1.62° for RI, 0.43 mm for UV).
• High reliability (ICC > 0.86) was observed for RI, VT, and UV, with moderate reliability for RL (ICC = 0.75).

Conclusions:

• The automated hybrid method accurately identifies landmarks and measures distal radius parameters on wrist radiographs.
• This approach significantly saves time and reduces human labor in dataset creation and algorithm development.
• The method offers a valuable tool for improving efficiency in radiological analysis.