| dc.description.abstract | Breast Cancer is the leading cause of mortality among women worldwide. The key to efficiently controlling cancer on a larger scale is effective diagnosis at an early stage by distinguishing the vital signatures of the malignant from the adjacent normal breast tissue. The gold-standard pathway for diagnosis of breast cancer typically involves triple breast assessment – (1) history and clinical examination, (2) imaging with mammography and ultrasound, and (3) tissue biopsy and its examination by microscopy-histopathology that needs preparation of a few hundred-micron thick tissue
slices which are place on glass slides and stained initially with Hematoxylin and Eosin (H&E) and, later with immunohistochemical analysis. Only after such a confirmation by histopathological examination can the treatment be given, in which typically, the surgeon removes cancer from the body by performing either a lumpectomy (breast-conserving surgery (BCS)) or a mastectomy. The key to efficiently controlling cancer on a larger scale is effective diagnosis at an early stage by distinguishing the vital signatures of the malignant from the adjacent normal breast tissue. This doctoral work explores the bulk optical properties of breast biopsy tissue, exploiting different
properties of light in different phases of the study and ultimately developing a handheld Multispectral pen using near-infrared spectroscopy (NIRS).The Multispectral Pen uses multiwavelength diffuse reflectance spectroscopy (DRS) to characterize the bulk optical properties of breast tissues, potentially aiding the clinician in rapid breast cancer diagnosis. Measurements were performed using fresh (N = 12) and formalin-fixed (FF) (N = 50) human breast biopsy tissues. The first phase of the research shows how a Time-domain (TD)-NIRS-based tool (i.e., Spectro-IRTDx) can delineate malignant and adjacent normal breast biopsy tissue ex vivo. The TD-NIRS-based optical measurements were performed on (n = 12) FF malignant and adjacent normal tissue samples obtained from N = 6 human subjects. The mean effective attenuation coefficient (μeff) for malignant tissue is 5.41 and 2.41 times higher than adjacent normal tissue with good statistical significance (p = 2.16e-03) at 850 nm and 940 nm, respectively. In the second phase, polarization-sensitive NIRS-based tools (i.e., POLS-NIRDx) were developed. The polarization-sensitive NIRS-based measurements were performed on (n = 20) FF malignant and adjacent normal tissue samples obtained from N = 10 human subjects. The degree of linear polarization (DOLP) and circular polarization (DOCP) were quantified. The average DOLP values, along with the standard error of mean (SEM) while operating at 850 nm, are found to be 0.080 ± 0.003 for malignant and 0.136 ± 0.006 for adjacent normal at 850 nm. While operating at 940 nm, the DOLP is 0.106 ± 0.005 and 0.179 ± 0.014 for the malignant and adjacent normal, respectively. It was observed that the average value of DOLP is lower in malignant compared to adjacent normal tissue at 850 and 940 nm and is highly statistically significant at 850 nm (p = 1.085e-05) and 940 nm (p = 7.577e-05), respectively. The average value of DOCP, along with SEM, was lower, 0.928 ± 0.004 at 15° for malignant, than adjacent normal 0.942 ± 0.003, respectively, while operating at 850 nm. A similar trend of lower DOCP for malignant 0.906 ± 0.005 at 15° compared to adjacent normal 0.917 ± 0.008 at 15°, respectively, while operating at 940 nm was observed. At 15°, the DOCP measurement variation shows better delineation than other angles of the rotating quarter wave plate 2 (QWP2). In the third phase of the study, the continuous wave (CW)-NIRS-based measurements were performed using a multispectral pen on two groups of tissue samples, namely Group 1 and Group 2. Group 1 comprised n = 50 FF and n = 12 fresh tissue samples obtained from N = 25 and N = 6 human subjects, respectively. For FF tissue samples, the mean reflectance values (< RCW >) with SEM for n = 25 adjacent normal breast tissue (0.72±0.01, 0.64±0.02, and 0.59±0.02) were lower compared to n = 25 malignant breast tissue samples (0.84±0.00, 0.80±0.01, and 0.74±0.02), with a high statistical significance (p < 0.0001) while operating at a wavelength of 850 nm, 940 nm, and 1050 nm respectively. A similar trend between the adjacent normal and malignant tissues was observed in the mean reduced scattering coefficient (<μ'_s>). The malignancy can be differentiated with high statistical significance (p < 0.0001) at 850 nm, 940 nm, and 1050 nm, respectively. For fresh tissue samples, the mean reflectance values (<RCW>) with SEM for n = 6 adjacent normal breast tissue (0.57±0.06, 0.51±0.08, and 0.37±0.06) were found to be lower as compared to n = 6 malignant breast tissue samples (0.82±0.02, 0.79±0.02, and 0.77±0.03), with a high statistical significance p = 4.33e-03, p = 2.60e-02, and p = 2.16e-03 while operating at a wavelength of 850 nm, 940 nm, and 1050 nm respectively. A similar trend between adjacent normal and malignant breast tissues was observed in the (<μ'_s>). The malignancy can be differentiated with high statistical significance (p = 2.16e-03 at 850 nm, 940 nm, and 1050 nm, respectively). For Group 2 (with malignant and adjacent normal regions imbibed) samples obtained from (N = 4) patients, a blind study was performed on the fresh breast tissue samples obtained from different margins (i.e., Lateral, Medial, and Superior). 2D surface maps and corresponding heat maps were obtained using the multispectral pen, which visually interprets tissue properties based on the diffuse reflection obtained from the tissue samples of different margins. Hence, different properties of light can play a vital role in effective breast cancer diagnosis. The results obtained using the multispectral pen establish the proof-of-principle for rapid breast cancer diagnosis ex vivo. In the final phase, an LED-based fiber-optic probe is developed for delineating tumors from adjacent normal ex vivo breast biopsy tissues. We envisage that breast cancer margin assessment using a fiber-optic probe on a large cohort of samples can aid clinicians in intraoperative settings. | en_US |
