Multimodal spectroscopy : real-time diagnosis of breast cancer during core needle biopsy

Abstract

Early detection of breast cancer is critical for improved survival. Currently, breast abnormalities are diagnosed based on a histopathological evaluation of tissue removed during core needle biopsy. Microcalcifications are used as targets to position biopsy devices, as they may indicate the presence of malignancy. Despite stereotactic guidance, needle biopsy fails to retrieve target microcalcifications in up to 15% of patients. Optical techniques may help clinicians accurately diagnose and treat patients by providing important diagnostic information in real time in a minimally invasive manner. This thesis describes the results of several studies we performed to evaluate the potential of Raman, reflectance, and intrinsic fluorescence spectroscopy to provide biochemical and morphological information for discriminating breast lesions. Each modality was evaluated individually, as well as in combination, using a technique known as multimodal spectroscopy (MMS). For the first part of this project we conducted a clinical study in which spectra were acquired from excised tissue in 99 patients and physically meaningful parameters were extracted by modeling the data. The goals of the study were as follows: 1) To prospectively validate previously developed diagnostic algorithms on the data from these patients; 2) To develop a new algorithm to evaluate additional histopathology diagnoses. Diffuse reflectance (DRS) spectra were modeled using diffusion theory and provided information about tissue absorbers and scatterers. Intrinsic fluorescence (IFS) spectra were extracted from the combined fluorescence and DRS spectra and analyzed using multivariate curve resolution. Raman spectroscopy data were fit using a linear combination of Raman active components (e.g. collagen, calcium, adipose) found in breast tissue. Prospective validation of Raman spectroscopy resulted in sensitivity and specificity and negative predictive value (NPV) of 78%, 98%, and 98%, respectively. An MMS system was developed to evaluate the benefit of combining information from all three spectroscopic modalities. We found that using new 3D Raman algorithm we could discriminate among 6 histopathology categories as compared to 4 categories previously diagnosed with Raman spectroscopy. For the second part of this project, we designed and developed a portable, miniature Raman clinical spectroscopy system to evaluate the potential of spectroscopy to guide the retrieval of microcalcifications during core needle biopsies. We focused specifically on the use of Raman spectroscopy for this application, as it is particularly sensitive to calcium-containing minerals. The system employs a side-viewing Raman probe that can be used in conjunction with commercial stereotactic needle biopsy devices. Prior to core needle excision, the Raman probe was inserted into the core needle biopsy device and spectra were acquired and analyzed in real time (<Is). The results from our work indicate that spectroscopy has the potential to accurately diagnose breast lesions and enable targeted biopsies of diseased tissue and retrieval of microcalcifications.