| Interventional medical ultrasound imaging is currently one of the main directions of ultrasound imaging,and it has been widely used for diagnosis and treatment.The key to surgical interventions lies in whether the doctor can estimate the position between the surgical instruments and the lesions,and place the surgical instruments in the lesion of the patient accurately.In order to improve the success rate of surgical interventions,this thesis proposes two fast detection algorithms for linear surgical instruments with the purpose of detecting these instruments accurately and assisting doctors with implementing surgical interventions.The research outcomes are as follows:For the contrast of ultrasound images is low and the background in the ultrasound images is complicated,this thesis proposes a straight line detection algorithm based on fuzzy enhancement and Hough transform.This algorithm includes the image enhancement,image thinning and line detection.Firstly,improved fuzzy enhancement algorithm is used to improve the image contrast.Secondly,image thinning algorithm extracts the linear skeleton to improve the accuracy of determining the angle of the straight line.Then,Hough transform is used to detect the straight line.Finally,the endpoint detected using Hough transform is corrected based on the correlation between image patches in order to improve the detection accuracy of the endpoints of the straight line.This algorithm can be used for the ultrasound image with the complicated background,and it has high accuracy and strong robustness.To reduce the complexity of the algorithm to ensure the real-time performance of the line detection algorithm,this thesis proposes another line detection algorithm based on the principle of gray projection.For the image background is relatively simple and the line has a certain width,this algorithm has the unique advantages and it can also be used for three dimensional line detection.The algorithm establishes the gray-level projection matrix using the image grey scale information directly.The extreme values in the matrix are found to determine the angle of the straight line quickly.Then,the endpoints of the straight line are determined based on the correlation of image patches.This algorithm is simple and can reduce the processing time greatly under the condition of meeting the precision.To evaluate the performance of the proposed two algorithms,experiments on the real clinical ultrasound images have been done.The quantitative analysis of the experimental results is made in terms of four indexes such as the detection accuracy,time,angle error and endpoint location error.The detection results of the proposed methods are compared with those of other compared algorithms.Experimental results show that our proposed algorithm can detect the linear surgical instruments more accurately and it can meet the demand of clinical practices due to its high accuracy and efficiency. |
