Conditional Generative Adversarial Network for Predicting the Aesthetic Outcomes of Breast Cancer Treatment

Area of Science:

• Medical imaging
• Artificial intelligence
• Computer vision

Background:

• Locoregional breast cancer treatment can cause significant changes in breast appearance, impacting patient quality of life.
• Breast asymmetries are a common and notable side effect, affecting patient self-esteem and satisfaction.
• Informed decision-making requires patients to understand potential aesthetic outcomes of different treatment options.

Purpose of the Study:

• To develop and evaluate a conditional generative adversarial network (cGAN) for simulating breast shape alterations.
• To realistically reconstruct torso images while manipulating breast shape.
• To provide a tool for visualizing treatment-induced aesthetic changes in breast cancer patients.

Main Methods:

• A conditional generative adversarial network (cGAN) was proposed for image-based breast shape manipulation.
• The model was trained and tested on a private dataset of breast images.
• Performance was evaluated against state-of-the-art methods in image reconstruction and shape manipulation.

Main Results:

• The proposed cGAN model demonstrated superior performance in realistic torso reconstruction.
• The model effectively simulated alterations in breast shape due to potential surgical interventions.
• Experimental results indicate the model's capability to realistically manipulate breast appearance in images.

Conclusions:

• The developed AI model can accurately visualize breast shape changes resulting from locoregional breast cancer treatment.
• This visualization tool can assist patients in choosing treatment plans by setting realistic expectations.
• The technology holds clinical relevance for improving patient counseling and treatment selection in breast cancer care.