An inverse heat transfer approach for patient-specific breast cancer detection and tumor localization using surface thermal images in the prone position

•An iterative method to detect and localize tumors from IR images is presented.•Patient-specific 3D model used to estimate the tumor parameters.•Results indicate that method is able to pinpoint size and location of tumor.•Large scale clinical trial using method can determine value as adjunctive tool. A novel technique is proposed based on the solution of the inverse heat transfer problem using the Levenberg-Marquardt algorithm. The method utilizes a patient-specific digital breast model and clinical Infrared imaging (IRI) images in the prone position to detect and localize a malignant tumor within the breast. The clinical IRI images are obtained from multiple views in the prone position, which allows access to the entire breast surface. The challenge of detecting a malignant tumor and accurately estimating its size and location within the breast is addressed through the solution of the inverse heat transfer problem built upon numerical simulations using patient-specific digital breast models. The proposed method was validated using clinical multi-view IRI images of the breasts of seven patients (one with bilateral breast cancer) with biopsy-proven breast cancer. The screening technique is found to be fast and accurate, and has the potential to be an effective adjunct to mammography especially for breasts with dense tissue.