FisMat2017 - Submission - View

Abstract's title: Time domain diffuse optical spectroscopy of tissue composition for in vivo clinical diagnostics
Submitting author: Paola Taroni
Affiliation: Politecnico di Milano
Affiliation Address: Piazza Leonardo da Vinci 32 20133 Milano
Country: Italy
Oral presentation/Poster (Author's request): Oral presentation
Other authors and affiliations: S. Konugolu (Politecnico di Milano − Dipartimento di Fisica, Milano, Italy), A. Pifferi (Politecnico di Milano − Dipartimento di Fisica, Milano, Italy), R.Cubeddu (Politecnico di Milano − Dipartimento di Fisica, Milano, Italy), F. Ieva (Politecnico di Milano − Dipartimento di Matematica, Milano, Italy), A.M. Paganoni (Politecnico di Milano − Dipartimento di Matematica, Milano, Italy), F. Abbate (European Institute of Oncology – Breast Imaging Unit, Milano, Italy), E. Cassano (European Institute of Oncology – Breast Imaging Unit, Milano, Italy), T. Durduran (ICFO, Barcelona, Spain)

Diffuse optical spectroscopy performed in the time domain allows the simultaneous estimate of the absorption and scattering properties of turbid media, like biological tissues. Thus, it can potentially allow the development of powerful tools for non-invasive clinical diagnostics.

Breast cancer is a leading cause of death in women and a major health burden worldwide. Mammography screening has high sensitivity, but unsatisfactory specificity in many situations (e.g., young women). After the preliminary optical characterization of collagen, using a portable optical mammograph operating in the time domain, we acquired images at 7 wavelengths (635-1060 nm) from more than 200 subjects, and evaluated the average composition of breast tissue (oxy- and deoxyhemoglobin, water, lipid and collagen), and the microscopic structure of tissue, as provided by scattering parameters. Malignant lesions turned out to be characterized by significantly higher collagen content than benign lesions. Furthermore, collagen is the most important parameter when the Discrete AdaBoost procedure, a machine-learning algorithm, is applied to discriminate malignant from benign lesions (sensitivity 88%, specificity 79%).

The knowledge of risk factors and the ability to identify high-risk women is important for prevention and design of personalized screening paths. Breast density (i.e. the fraction of fibroglandolar tissue) is a strong and independent risk factor for breast cancer. Currently, its evaluation requires x-ray mammography. We estimated non-invasively by optical means the average composition and scattering parameters of breast tissue of 200 subjects. The best regression logistic model to identify high-risk (BI-RADS 4) subjects, is based only on collagen content and scattering parameters.

Collagen is involved in the onset and progressions of breast cancer, and affects its invasiveness. To investigate whether it is a risk factor independent of breast density, we estimated collagen content in 109 subjects (56 cancer patients and 53 healthy subjects). Individuals with high collagen content or high mammographic density (top 15%) show markedly increased occurrence of cancer, but the two parameters identify different population subsets, potentially indicating collagen as an independent risk factor for breast cancer.

Recently, we have also extended our work towards the diagnosis of thyroid cancer. Specifically, we have performed the optical characterization of thyroid-specific tissue constituents: Thyrosine, thyroglobuline and iodine, which are potentially of interest for in vivo diagnostics.