Thesis Committee:

Keith D. Paulsen (chair)

Brian W. Pogue

Venkataramanan Krishnaswamy

Steven P. Poplack

Andrew Karellas (external)

Annual screening mammograms are recommended for all women above 40 years of age for early detection of breast cancer, and are known to improve cancer mortality rates. However, mammography is hampered by low sensitivity and a high rate of false positive results that can lead to invasive and possibly unnecessary biopsies. Digital breast tomosynthesis (DBT) is a recently approved screening method that takes multiple X-ray images at different positions relative to the breast, rendering excellent three dimensional anatomic detail of the breast. Near-infrared spectral tomography (NIRST) was added into a DBT system, to provide co-localized information about the metabolic status of the tissue, specifically quantification of blood content, oxygen saturation, water and lipid levels.  When used alone, NIRST is hampered by low spatial resolution.  The hypothesis in this work was that noninvasive differentiation between malignant and benign lesions can be more accurately obtained by combining spatial information from DBT and functional information from NIRST in a single exam.

To this end, the NIRST hardware has been integrated into an existing DBT unit, allowing completely co-registered X-ray and optical data.  Early simulation work demonstrated the potential of this synergistic combination, optimized for short scanning time, low cost and complexity.  The system was built and tested to obtain both clinical quality DBT scans and high fidelity broadband NIRST data.  Breast-mimicking phantoms assessed size, contrast and depth resolution as well as chromophore recovery.  A total of 32 normal subjects were examined and the recovered tissue metabolic markers were analyzed with respect to breast characteristics.  Finally,  13 women with abnormal mammographic findings were imaged prior to biopsy and the results demonstrate correlation between hemoglobin content and histo-pathological diagnosis. The addition of NIRST imaging to DBT could aid in clinical decision making and decrease the number of biopsies which are performed with negative results, improving patient experience.