| Περίληψη: | Mammography is currently the gold standard technique for early detection of breast cancer before it becomes clinically palpable. Breast compression is essential part of mammography which provides better contrast and smaller radiation dose. It optimizes the image quality and the visualization of small lesions by reducing the breast thickness. The overall goal of this thesis is to simulate the complete mammographic imaging chain, starting from the creation of breast phantoms, compression, and mammogram simulation. A specific aim is to carry out a study that compares the quality of breast lesions on simulated mammographic images obtained from both uncompressed and compressed breast models.
For the creation of the breast models, we use a software breast phantom which incorporates a 3D voxel array with geometric primitives, a ductal network, and a model for lesions. Further on, a Simulation Algorithm for Soft Tissue Compression (SASTC) is used for the compression of the breast phantoms. For this purpose, the algorithm was implemented into C++ language and incorporated into the available Breast Simulator software. The SASTC algorithm is based on the spring-mass model according to which the volume of the breast phantom is divided into a number of elements, each one consisting of 27 nodes and the nodes are connected with springs. The compression algorithm allows stretching and compression of the volume object. Due to these deformations the voxels of the breast’s volume are distorted. Several algorithms are developed to correct for these distortions. These algorithms are based on interpolation techniques and on the Jordan Curve Theorem.
We designed 11 models of small breast phantoms that included microcalcifications of different sizes. These breast phantoms as well as the compressed model of each phantom are subjected to mammography simulation to create mammograms. The x-ray image acquisition model, used for the acquirement of the mammographic images, is based on a combination of ray tracing techniques and the x-ray attenuation law. A comparison between the acquired mammograms of both uncompressed and compressed breast phantom was performed. A preliminary visual evaluation demonstrates that the clarity of the structures in the breast model is improved in the images of the compressed breast. Quantitative evaluation involving contrast calculations supports the former results.
|
|---|
