Real time intraoperative detection of residual breast cancer using active thermal imaging

Margin-negative (R0) resection is highly important for long-term survival, with reduced mortality and morbidity in breast cancer patients. Palpation and visual inspection, as well as the current intraoperative imaging techniques are insufficient for discriminating between healthy and malignant tissue types, often leading to margin-positive (R1) resections. This study introduces an active thermal imaging system that enables real-time intraoperative detection of residual cancer via dynamic thermal response patterns. A prototype integrating an infrared heating system and a high-resolution thermal imaging camera coupled with a data transformation algorithm was developed. Freshly resected breast cancer specimens were analyzed to determine thermal dissipation differences between R0 and R1 resections by correlating them with histopathological reports. The transformed data were passed through a decision tree model to determine the cutoff values of relevant thermal parameters. After the operating parameters were finalized and the safety standards were met, the system was implemented real-time during breast cancer surgeries in willing patients to determine intraoperative cutoffs. Two derived thermal parameters, v(si) of R₁₂ʺ and v(si) of R₁₃ʺ, representing the thermal decay behavior across two adjacent cooling intervals were identified as significant predictors of margin-positive (R1) or margin-negative (R0) resection (p < 0.001). Cutoff values of 68.42 for $$\:\text{v}\left({\text{s}}_{\text{i}}\right)$$ of $$\:{\text{R}}_{12}^{"}$$ and 71.43 for $$\:\text{v}\left({\text{s}}_{\text{i}}\right)$$ of $$\:{\text{R}}_{13}^{"}$$ were established for intraoperative decision-making. The optimized decision tree model using these variables achieved a classification accuracy of 93.33%, with sensitivity of 92.00% and specificity of 94.44% during intraoperative breast cancer surgery. The proposed active thermal imaging system provides a novel, real-time, contrast-free solution for intraoperative cancer detection. Its demonstrated reliability, automation, and ease of integration into surgical workflows may position it as a promising tool for enhancing surgical precision and improving oncological outcomes.