Background and Objectives: Due to the inherent characteristics of low contrast and significant noise in ultrasound images, this study explores the feasibility of an image stitching and fusion method based on contour corner detection combined with an enhanced dung beetle algorithm. Methods: The proposed method utilizes ultrasound data acquired from abdominal scans of postoperative tumor patients to design a novel ultrasound image stitching algorithm. Initially, contour corners are employed to capture the ultrasound features that are interfered by noise. Subsequently, the dung beetle algorithm is applied to identify the optimal seam line path. To prevent the algorithm from falling into local optima, Lévy flight and Cauchy mutation are incorporated. Finally, the optimal seam line mask is passed to the Laplacian pyramid for fusion. The stitching results obtained through the proposed algorithm are evaluated using both subjective and objective methods. Results: The proposed method demonstrated superior performance over the traditional SIFT-based approach across multiple metrics, as follows: a 0.9<InlineEquation ID="IEq1"> <EquationSource Format="TEX">$\%$</EquationSource> <EquationSource Format="MATHML"><math> <mo>%</mo> </math></EquationSource> </InlineEquation> increase in Structural Similarity Index Measure (SSIM), a 1.7<InlineEquation ID="IEq2"> <EquationSource Format="TEX">$\%$</EquationSource> <EquationSource Format="MATHML"><math> <mo>%</mo> </math></EquationSource> </InlineEquation> gain in Peak Signal-to-Noise Ratio (PSNR), a 2.6<InlineEquation ID="IEq3"> <EquationSource Format="TEX">$\%$</EquationSource> <EquationSource Format="MATHML"><math> <mo>%</mo> </math></EquationSource> </InlineEquation> improvement in Signal-to-Noise Ratio (SNR), a 12.1<InlineEquation ID="IEq4"> <EquationSource Format="TEX">$\%$</EquationSource> <EquationSource Format="MATHML"><math> <mo>%</mo> </math></EquationSource> </InlineEquation> enhancement in Quality Mean Squared Error (QMSE), and a 5.9<InlineEquation ID="IEq5"> <EquationSource Format="TEX">$\%$</EquationSource> <EquationSource Format="MATHML"><math> <mo>%</mo> </math></EquationSource> </InlineEquation> reduction in Root Mean Square Error (RMSE). Conclusion: The ultrasound image stitching algorithm based on contour corner detection and the enhanced dung beetle optimization algorithm produces high-quality stitched images and effectively addresses the issue of limited field of view in ultrasound scanners.

Ultrasound image stitching fusion based on contour corner points and improved dung beetle algorithm optimization

期刊论文
发表日期：

2025年12月5日

查看全文

Yibo Jiang,
Jiayu Yin,
Chengjie Cai,
Xu Lv,
Hui Bi

摘要

Background and Objectives:

Due to the inherent characteristics of low contrast and significant noise in ultrasound images, this study explores the feasibility of an image stitching and fusion method based on contour corner detection combined with an enhanced dung beetle algorithm.

Methods:

The proposed method utilizes ultrasound data acquired from abdominal scans of postoperative tumor patients to design a novel ultrasound image stitching algorithm. Initially, contour corners are employed to capture the ultrasound features that are interfered by noise. Subsequently, the dung beetle algorithm is applied to identify the optimal seam line path. To prevent the algorithm from falling into local optima, Lévy flight and Cauchy mutation are incorporated. Finally, the optimal seam line mask is passed to the Laplacian pyramid for fusion. The stitching results obtained through the proposed algorithm are evaluated using both subjective and objective methods.

Results:

The proposed method demonstrated superior performance over the traditional SIFT-based approach across multiple metrics, as follows: a 0.9 $\%$ $%$ increase in Structural Similarity Index Measure (SSIM), a 1.7 $\%$ $%$ gain in Peak Signal-to-Noise Ratio (PSNR), a 2.6 $\%$ $%$ improvement in Signal-to-Noise Ratio (SNR), a 12.1 $\%$ $%$ enhancement in Quality Mean Squared Error (QMSE), and a 5.9 $\%$ $%$ reduction in Root Mean Square Error (RMSE).

Conclusion:

The ultrasound image stitching algorithm based on contour corner detection and the enhanced dung beetle optimization algorithm produces high-quality stitched images and effectively addresses the issue of limited field of view in ultrasound scanners.