| Texto completo | |
| Autor(es): |
Machado, Aline Y.
;
Rodrigo, T. Massera
;
Tomal, Alessandra
Número total de Autores: 3
|
| Tipo de documento: | Artigo Científico |
| Fonte: | MEDICAL IMAGING 2025: PHYSICS OF MEDICAL IMAGING, PT 1; v. 13405, p. 6-pg., 2025-01-01. |
| Resumo | |
The generation of synthesized mammographic images is pivotal for the advancement of virtual clinical trials and the enhancement of machine learning models through data augmentation applied to breast imaging. In this study, it is presented a comprehensive framework that integrates Monte Carlo simulations, ray-tracing methods and deep learning model for scatter estimation to efficiently generate synthesized mammographic images. 311 anthropomorphic breast phantoms were utilized to produce a dataset of 6220 simulated images across different beam energies, which were used to train a deep learning model based on the ResNet architecture. The model achieved a mean absolute percentage error of 2.49 +/- 4.04% in scatter prediction compared to MC simulations, along with a structural similarity index measure of 0.98 +/- 0.02. Subsequently, a ray-tracing algorithm was employed to generate primary images from a compressed breast phantom in CC view, which were combined with deep learning scatter predictions to produce synthesized polyenergetic mammographic images. This approach significantly reduced computational time by an order of magnitude compared to traditional MC methods, enabling faster generation of synthesized images, compared to traditional methods. The proposed pipeline facilitates the creation of large and diverse datasets, supporting the optimization of image analysis and virtual clinical trials in mammography. (AU) | |
| Processo FAPESP: | 18/05982-0 - P&D&I em metrologia das radiações na área da saúde |
| Beneficiário: | Linda Viola Ehlin Caldas |
| Modalidade de apoio: | Auxílio à Pesquisa - Temático |
| Processo FAPESP: | 15/21873-8 - Desenvolvimento e implementação de metodologias para otimização de técnicas de imagens em radiologia digital |
| Beneficiário: | Alessandra Tomal |
| Modalidade de apoio: | Auxílio à Pesquisa - Regular |
| Processo FAPESP: | 21/08923-7 - Redes neurais aplicadas à geração de modelos de mama para dosimetria em modalidades de imagens por raios X |
| Beneficiário: | Alessandra Tomal |
| Modalidade de apoio: | Auxílio à Pesquisa - Regular |
| Processo FAPESP: | 23/03535-4 - Caracterização e otimização de tecnologias inovadoras de imagens funcionais da mama por raios X |
| Beneficiário: | Alessandra Tomal |
| Modalidade de apoio: | Bolsas no Exterior - Pesquisa |