The project

Current methods for assessing tumor margins in breast-conserving surgery are slow, subjective, and often result in incomplete tumor removal, leading to high reoperation rates and increased healthcare costs. Postoperative histopathological analysis can take days, while intraoperative frozen-section analysis is time-consuming, prone to sampling errors, and not widely accessible.

Spectra-BREAST will introduce an advanced multimodal solution that combines a custom-designed Hyperspectral Imaging (HSI) system, fiber-optic Raman Spectroscopy (RS), and hybrid AI algorithms. First (1), a next-generation HSI camera will rapidly scan the entire resected specimen in 2 minutes, detecting suspicious or close tumor margins. Then (2), AI models will guide a robotic arm equipped with a fiber-optic Raman probe to precisely target these regions and perform in-depth spectral analysis (3) to confirm tumor presence up to 2 mm beneath the tissue surface. This integrated workflow will deliver real-time, highly accurate feedback (sensitivity and specificity >95%) in under 5 minutes, enabling surgeons to achieve complete tumor removal in a single procedure and reducing reoperations by up to 90%.