Tissue-Specific Analysis Of Complex Tumors Based On DCE-MRI

In living organism, dynamic contrast-enhancement image provides a noninvasive method to analyze various disease’s functional changes associated with its initial and development stage and treatment. So it’s an effective auxiliary tool for tumor prevention. DCE-MRI can improve the display of breast lesions with high imaging resolution, and identify subtle performance of differential lesion morphology, including chest wall invasion, tumor vessel morphology, lymph node status and signal strength of suspected site. What’s more, it can provide the blood supply situation of breast lesions that cannot reflected by regional traditional MRI and different strengthening mode in different parts of the lesion after injection of the tracer, because of these characteristics, coupled with the diagnostic characteristics of lesion morphology, it can become an important basis for the clinical diagnosis of doctors. This paper mainly studies the variational level set segmentation algorithm based on breast DCE-MRI lesions extraction applications, and clustering methods in combination compartment model applied research on the pharmacokinetic parameters measured, specific work follows, and the application of clustering methods in combination compartment model pharmacokinetic parameters measured, the specific work is as follows:(1) Breast DCE-MRI segmentation based on variational level set segmentation model of medical images in research, according to medical image intensity non-uniformity, imaging DCE-MR breast cancer and the presence of multiple lesions and other characteristics, proposes an improved variational level set segmentation model to extract the DCE-MRI interested in breast lesions. Improved level set segmentation model based on the C-V model, introduced new image model and local clustering criterion to solve the medical image’s uneven problem, and in the weighted evolution curve length added a scale change function to improve the segmentation speed and accuracy. Experiments show that the improved segmentation algorithm has more accurate segmentation results, faster breast DCE-MRI segmentation compared with the former. So the proposed algorithm has a certain practicality.(2) Breast DCE-MRI complex tumor analysis and clinical application, analyze and compare the advantage between open source software convex analysis of mixtures-compartmental modeling (CAM-CM) and clinical diagnosis, as well as theoretically learn pharmacokinetic parameter estimation of compartment model, integrated application of time series clustering and convex analysis methods and the uniqueness verification of results. According to the actual situation of domestic clinical data, we pretreat interpolation of clinical DCE-MR image, achieve the effective analysis on domestic clinical medical image of DCE-MR using the open source software CAM-CM, and further verify the validity of pharmacokinetic parameters in solving the clinical partial volume effect problem. There obtains a certain value on clinical application.(3) Breast DCE-MRI pretreatment system, under MATLAB development environment, a DCE-MRI complex tumor analysis and processing system is built by using its GUI. The system, integrated clinical experimental data pre-processing, interpolation and interested lesion area extraction operation, combined with analyzing and processing parameters of open source CAM-CM, and diagnosis processing functions of clinical outcome, is used by doctors in clinical diagnosis.The breast DCE-MRI data, used in this thesis, is provided by XX Provincial Tumor Hospital, XX No.6 People’s Hospital, XX Tumor Hospital and XX Ruijin Hospital.