An approach of modelling of irregular masses

Abstract

The goal of this study is to create and evaluate a methodology for generation of realistic 3D computational models of breast tumours with irregular shapes. Methods The methodology for the creation of breast masses consists of two steps: (a) creation of the initial diffusive tumour shape, by choosing Brownian motion or nearest neighbour random walk algorithm, (b) creation of a solid tumour shape by applying a set of 3D filters, and morphological operations. The initial models were smoothed by applying a set of image processing methods. Thereafter, projection images of these 3D lesions were generated by using an in–house developed software, capable to simulate the X-ray transport through the computational phantoms. Subsequently, a technique for embedding the simulated masses in patient mammography images was developed and applied. Results Thirty irregular masses with different sizes and shapes were generated, and projection images were simulated. The realism of the projected masses on patient images was evaluated by comparison of extracted features, such as the exponential parameter of the power spectrum, fractal dimension and other statistical parameters to these extracted from patient data. The obtained results confirmed that the methodology is capable of producing realistic 3D tumour models with user defined sizes and irregular shapes. Conclusions The methodology will be used to generate unique and realistic in shape and size computational models of breast adenoma, intraductal papilloma, cysts and duct hyperplasia. These computational models are stored in an open source database to be used by all professionals working toward the creation of new technologies for breast-screening and diagnosing. Acknowledgements This research is supported by the Bulgarian National Science Fund under grant agreement DN17/2. This project also has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 692097.