Breast cancer (BC) is the most common cancer among women in the United States; approximately one out of every 24 women die of related causes. BC screening is a critical factor for improving patient prognosis and survival rate. Infrared (IR) thermography is an accurate, inexpensive and operator independent modality that is not affected by tissue density as it captures surface temperature variations induced by the presence of tumors. A novel patient-specific approach for IR imaging and simulation is proposed. In this work, multi-view IR images of isolated breasts are obtained in the prone position (face down), which allows access to the entire breast surface because the breasts hang freely. The challenge of accurately determining size and location of tumors within the breasts is addressed through numerical simulations of a patient-specific digital breast model. The digital breast models for individual patients are created from clinical images of the breast, such as IR imaging, digital photographs or magnetic resonance images. The numerical simulations of the digital breast model are conducted using ANSYS Fluent, where computed temperature images are generated in the same corresponding views as clinical IRI images. The computed and clinical IRI images are aligned and compared to measure their match. The determination of tumor size and location was conducted through the Levenberg-Marquardt algorithm, which iteratively minimized the mean squared error. The methodology was tested on the breasts of seven patients with biopsy-proven breast cancer with tumor diameters ranging from 8 mm to 27 mm. The method successfully predicted the equivalent tumor diameter within 2 mm and the location was predicted within 6.3 mm in all cases. The time required for the estimation is 48 minutes using a 10-core, 3.41 GHz workstation. The method presented is accurate, fast and has potential to be used as an adjunct modality to mammography in BC screening, especially for dense breasts.
Department, Program, or Center
Satish G. Kandlikar
Gonzalez Hernandez, Jose Luis, "A Patient-Specific Approach for Breast Cancer Detection and Tumor Localization Using Infrared Imaging" (2019). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus
Available for download on Saturday, September 05, 2020