Detection and construction of chest wall on breast magnetic resonance images

The purpose of this study is to propose a method for detection and construction of chest wall for breast magnetic resonance images.

A volume of breast MR slices are firstly acquired and utilized to detect initial points of chest wall. To calibrate the chest wall curve, the points along the curve is set with reference to its neighboring points. Through the calibration method, a curve of chest wall can be detected from a volume of breast magnetic resonance (MR) slices. Such a curve can be applied for segmentation of breast region in a volume of MR images.

The experimental results reveal that the minimal RMSE was measured from the setting two polynomial functions and the points from the vertical position ≤320. If all edge points are used to simulate the curve, two circle functions can reach the minimal RMSE.

The experimental results verify that chest wall for breast density estimation can be better simulated by two circle functions, which simulate right and left chest walls respectively. Furthermore, such a simulation curve is suggested to utilize partial edge points under the given vertical position.