[Development of temporal subtraction method for chest radiographs by using pixel matching technique].

Area of Science:

• Medical Imaging
• Radiology
• Image Processing

Background:

• Temporal subtraction images enhance interval changes in chest radiographs by removing normal structures.
• Conventional methods suffer from subtraction artifacts, hindering the detection of interval changes.

Purpose of the Study:

• To develop and evaluate a pixel matching technique for reducing artifacts in temporal subtraction images.
• To improve the effectiveness of temporal subtraction in detecting interval changes, specifically lung nodules.

Main Methods:

• A pixel matching technique was developed, replacing warped image pixels with values from a kernel closest to the current image.
• The technique involves nonlinear warping and kernel-based pixel value replacement.
• Evaluation used one hundred temporal subtraction images with simulated nodules.

Main Results:

• A 3x3 kernel size reduced misregistration artifacts by 72% compared to conventional methods.
• Contrast-to-noise ratio of simulated lung nodules increased by 5%.
• Nodule area on subtraction images decreased by 6%.

Conclusions:

• The pixel matching technique substantially reduces misregistration artifacts in temporal subtraction images.
• This method enhances the visibility of simulated lung nodules.
• The improved temporal subtraction method can assist radiologists in lung nodule detection in digital chest radiography.