Research On Monitoring Tissue Lesion Based On Ultrasound Imaging In HIFU Treatment

High Intensity Focused Ultrasound therapy is an emerging technology of curing tumour for clinical treatment,its advantages are no side effects and Non-invasive.The main theory of the HIFU treatment is that focusing the ultrasound on tumour areas,thus making the temperature of the focused area above 65℃ and the tumour tissue of focused area develops degeneration or coagulative necrosis,and the purpose of curing the tumour will be achieved.In this paper,The experimental samples are fresh isolated pork tissue,real-time ultrasound images are obtained by B-ultrasound imaging device before and after HIFU irradiation.We also study on the relationship between tissue lesion and images feature in focused area,and the method of detecting and identifying tissue lesion.Finally,we use the level set methods to segment HIFU tissue injury regions.The main tasks are as follows:(1)Analysis the texture features of ultrasound images at HIFU focused area.There are differences in texture feature between the ultrasound image after tissue irradiated and before tissue irradiated.Therefore,we extract the gray-scale entropy and mixed entropy of the gray-gradient co-occurrence matrix of the focused region in ultrasound images,and use FCM clustering algorithm to perform clustering.The experimental results show that: compare with the gray mean and image entropy of the traditionally used parameters,gray entropy and mixed entropy can identify the tissue lesion caused by HIFU more accurately.(2)We propose a method to detect and recognize the tissue lesion areas based on ultrasound images.Aimed at the tissue lesion areas are difficult to locate,we use the gray threshold value to locate the tissue lesion areas Firstly.Secondly,according to the relationship of distance between the tissue edge of experimental sample and initially located pixels,we filter the detected areas.Finally,gray entropy and mixed entropy are used for SVM classification and recognition.The experimental results show that: we use K-means and morphology method can calculate the edge of tissue and use the gray entropy can recognize the tissue lesion area accurately,so that locate the tissue area automatically and it is helpful for monitoring the tissue lesion of HIFU treatment.(3)Segment the tissue lesion ares of HIFU treatment.We take the localized tissue lesion areas as the initialization position,and propose a level set method to segment the tissue damage area.The experimental results show that: the LGDF model combines the image gray level and variance information for curve evolution,and has higher segmentation accuracy than the LBF model.Therefore,this method provides more detail and further visualization in tissue lesion areas.The methods of this paper studies will be helpful to adjust the plans in HIFU treatment,and its provides new ideas for monitoring the process of HIFU treatment.