Automated Pleural Line Detection Based on Radon Transform Using Ultrasound

Area of Science:

• Medical Imaging
• Ultrasound Technology
• Pulmonology

Background:

• The pleural line in lung ultrasound is crucial for diagnosis but challenging for novices to detect visually.
• Current clinical practice relies on experienced clinicians for pleural line identification, posing limitations.

Purpose of the Study:

• To develop and validate a computer-aided technique for automated pleural line detection in ultrasound images.
• To provide an objective and potentially more accessible method for identifying the pleural line.

Main Methods:

• Utilized the Radon transform for line object detection in ultrasound images.
• Incorporated body mass index and chest wall thickness to estimate pleural thickness range.
• Considered ultrasonic properties of the pleural line for enhanced detection.

Main Results:

• The automated method successfully identified pleural lines in 76 out of 83 datasets (91.6% detection rate).
• Measurements of pleural line depth by the automated method closely agreed with manual measurements (Bland-Altman test).
• Measurement errors for pleural line depth were less than 5%.

Conclusions:

• The developed computer-aided technique can automatically detect pleural lines in ultrasound images.
• This automated method shows potential as an alternative to visual inspection, pending further validation on diverse datasets.