A Robust edge detection technique for bone extraction from X-ray images based on image processing techniques

• Published in: Multidimensional systems and signal processing Vol. 34; no. 1; pp. 249 - 270
• Main Authors: Hassan, Nashaat M. Hussain, Mahmoud, Mohamed Hassan M.
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.03.2023; Springer Nature B.V
• Subjects: Algorithms; Artificial Intelligence; Circuits and Systems; Edge detection; Electrical Engineering; Engineering; Image contrast; Image processing; Medical imaging; Processing speed; Signal,Image and Speech Processing; Soft computing; Image inverse; X-ray bone image; Bounded-sum operator; Strategies of edge detections; Sharping filter; Method of gradient image
• ISSN: 0923-6082; 1573-0824
• DOI: 10.1007/s11045-022-00860-w

In spite of the fact that there are several bone edge detection techniques for x-ray images available in the literature, detecting the edges of medical images in the traditional sense without extracting areas of interest is not helpful. In this work, an accurate scheme for identifying the bone boundaries from human beings' X-ray bone images with low calculation burdens and high processing speed was planned. The planned method is based on the combination of soft computing techniques ( bounded-sum and negation operators) and a sequence of image processing techniques. The proposed approach depends in its implementation on two parts: first, adjusting the contrast of the images by reversing the image; then the bounded-sum operator is applied to make a shift of the pixels from low levels to high levels. This first part ends by sharpening the contrasted image. Detecting the edges of the outcome of the first part is realised by applying adaptive threshold value of the gradient magnitude. The efficiency of the proposed technique is evaluated using different measures to ensure the degree of efficiency, such as dice similarity, confusion matrix, effectiveness measures, RMSE, and processing speed. The obtained results confirmed the dominance of the planned method compared to the other algorithms in both accuracy of performance and computation times. The excellent results of the proposed method make it a suitable way to be employed in many physical requests.