Methodology to simulate temporal changes in breast tumors and parenchyma using Perlin noise

Simulating temporal changes in breast tissue has been challenging, despite advancement in virtual clinical trials. This study aims to enable virtual trials over time, motivated by interest in the effect of tissue involution on tumor visibility. We present a framework for creating 3D voxel-based breast phantoms and lesions using in-house Perlin Noise-based algorithm. The framework consists of three modules “Population Creator”, “Phantom Creator” and “Lesion Creator”. Population Creator generates a population based on characteristics from real clinical cases. Phantom and Lesion Creator enable selection of breast shape, size, density, and lesion shape and size. Perlin Noise parameters are selected to match the appearance of different tissue types. We used open-source software to reconstruct digital breast tomosynthesis images. Alternative software can be used for other imaging modalities. We illustrate the use of the framework to simulate temporal changes in the breast. First, assuming that the average volumetric breast density is 10.7% at 57 years of age and decreases exponentially over time, we create the phantom at the start and the end of the screening program (40 and 74 years, respectively, in Sweden). The breast density was calculated as 16.0% and 7.2%, respectively. Second, we show a simulated lesion at different time points, assuming a tumour volume doubling time of 282 days. We present a simulation framework for temporal changes in breast anatomy. We illustrate two applications which support the use of virtual clinical trials over time.