Feasibility study of breast cancer detection using infrared imaging

Ashish Gupta, Purdue University

Abstract

The American Cancer Society estimates that this year in the United States alone, approximately 211,300 women will be diagnosed with breast cancer and approximately 39,800 will die as a result of this disease. Currently, mammography is the gold standard for breast cancer detection. Synthetic mammograms were generated for semi-realistic normal and malignant compressed female breasts using Monte Carlo simulations. Realistic initial and boundary conditions were used. The modeling of physical processes, and the model, was validated by the coincidence between Monte Carlo and theoretical data; and with literature. Statistical noise, tumor size and tumor depth studies on the energy recorded by the digital detector were conducted. A novel application of the Discrete Probability Function method to solution of equation of radiative transfer in mammography has also been successfully explored and preliminary results have been reported. This method, as compared to Monte Carlo, holds tremendous potential for producing low statistical noise results with comparable computational time. Early detection is the best defense against breast cancer, and infrared imaging holds promise as an adjunct to other diagnostic modalities, in conjugation with newly emerging analytical and numerical computational tools. A computational bioheat transfer model has been developed, which simulates the bioheat transfer equation in a three dimensional realistic patient specific heterogeneous female breast, as an adjunct to mammography, and predicts the surface temperature distributions for normal and malignant breasts. The previously reported in-vivo measurements were used to derive the blood flow rate values for normal and cancerous tissue. These derived properties were used to predict the limitations of infrared detection capability at its present developed stage to detect deeper tumors. The probable causes of past controversies related to IR imaging are also discussed. The benefit of thermal stress, vasoconstriction, dynamic imaging and image processing techniques to improve the thermal signature of deep tumors on the skin surface were explored. A novel micro scale heat transfer model of the tumor has also been developed and a parametric study was done for the tumor radial temperature distributions, as a function of its attributes.

Degree

Ph.D.

Advisors

Gore, Purdue University.

Subject Area

Biomedical research|Mechanical engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS