Diploma Thesis Success: Deep Learning for Breast Cancer Diagnosis Using Mammography and Tomosynthesis
The SIPPRE Research Group (Signal, Image Processing and Pattern Recognition Group) is pleased to announce the successful completion and public defense of the Diploma Thesis of Eleni Papadimitriou, entitled:
“Development and Evaluation of Deep Neural Networks for Breast Cancer Diagnosis Using Digital Mammography and Tomosynthesis.”
The thesis was publicly presented and examined on January 23, 2026, at the Department of Electrical and Computer Engineering, University of the Peloponnese.
Research Focus & Contribution
This diploma thesis addresses a major challenge in contemporary medical imaging: the reliable and efficient diagnosis of breast cancer using Artificial Intelligence applied to large-scale imaging data.
The work focuses on the design and evaluation of deep learning–based diagnostic pipelines using:
- Digital Mammography (DM)
- Digital Breast Tomosynthesis (DBT)
Using data from the EMBED database, the study investigates both 2D and 3D representations, exploring different network architectures and fusion strategies. A two-stage deep learning framework is proposed:
- Detection stage for distinguishing normal from suspicious cases
- Classification stage for differentiating benign and malignant findings
Extensive experimental evaluation demonstrates:
- Very high performance in the initial detection stage (accuracy > 96%, AUC > 0.97)
- Robust classification results for benign vs malignant lesions
- Insightful analysis of the impact of parameters such as slice selection and batch size on model stability and generalization
The results highlight the potential of deep neural networks as decision-support tools that can assist radiologists by improving diagnostic accuracy and reducing clinical workload.
Alignment with SIPPRE Research
This diploma thesis is fully aligned with SIPPRE’s core research directions in:
- Medical image analysis
- Deep learning for biomedical applications
- Computer-aided diagnosis (CAD) systems
- AI-assisted breast imaging and tomosynthesis
- Translational AI for clinically relevant decision support
The work contributes to SIPPRE’s ongoing research efforts in AI-driven medical imaging, with an emphasis on performance, reliability, and real-world clinical applicability.
Congratulations to Eleni Papadimitriou on this excellent academic achievement, and best wishes for her future academic and professional endeavors.