Theoretical considerations and future directions

As a hybrid optical/acoustic imager, the iPAT system that was developed for, and employed in, this work, derived much of its inherently unique capabilities from exploitation of the molecular specificity associated with multispectral optical imaging. In contrast, examination of section 1.4.2 reveals that many groups working on PAT of the breast settled on fixed-wavelength illumination1–3. For example, the fundamental Nd-YAG laser emission at 1064 nm was often used, probably due to factors associated with convenience, reliability and expense of employed laser systems. Our group recognized early-on that these wavelengths are rarely optimal for differentiation of various tissue types, as can be appreciated in Figure 2.4 of chapter 2, as well as our earlier work4,5. Consequently, multispectral illumination capability was one of the defining features of this investigation. In this section, merits and pitfalls of the employed wavelength-selection are discussed and the theoretical basis behind the proposed wavelength optimization approach are outlined.

5.1.1 Lipids vs Hemoglobin in ex-vivo breast tumour imaging

A survey of photoacoustic imaging literature suggests that the primary contrast of tumours is provided by elevated hemoglobin concentration associated with cancer induced angiogenesis2,6–9. Consequently, it was

142

expected that iPAT scans targeting hemoglobin would exhibit increased signal intensity originating from periphery of tumour regions. However, contrary to this hypothesis, this investigation found that hemoglobin weighted images generally showed little or no correlation with hypoechoic US regions inside ex vivo specimens. This was the case for most of the specimens encountered in this study and is exemplified in Figure 2.5 of chapter 2. Therefore, the extendibility of documented in vivo results to ex vivo applications must be questioned. The extendibility hypothesis rests upon two assumptions.

First, it was assumed that a freshly excised tissue specimen would quickly reach low blood oxygenation levels due to a lack of blood supply. This motivated the 690 nm wavelength selection, where deoxyhemoglobin dominates tissue absorption as shown in Figure 1.4 of the Introduction. The validity of this assumption was corroborated by the increased 690 nm signal intensity associated with bloody regions. For example, note the hyper-intense appearance of the region indicated by the dashed oval in the 690 nm image shown in Figure 2.5(a) of chapter 2. This region corresponds well to the area of high blood content marked by a dashed oval in the photograph of Figure 2.5(h). In comparison, the corresponding 800 nm images in Figure 2.5(b) show reduced intensity. This result suggests a dominance of deoxygenated blood since the 800 nm wavelength is absorbed equally by deoxy- and oxy-hemoglobin while the 690 nm wavelength is dominantly absorbed by only deoxyhemoglobin.

The second assumption is that blood distribution inside the specimen will remain relatively unchanged after excision. However, the blood distribution depends on ability of the vascular network to maintain blood vessel volume, which in turn depends on blood pressure, provided by cardiac output. Because an excised specimen is no longer connected to the cardiovascular system, it is possible that many of its blood vessels collapsed during surgery, leading to a heterogeneous reduction in blood volume. Therefore, hemoglobin distribution inside excised tissue specimens may not be a good biomarker of malignancy.

Alternatively, low lipid concentration has been well documented by optical methods as being a significant indicator of breast abnormalities10. Furthermore, unlike the mobile nature of blood, the lipid concentration within breast tissue is unaffected by surgical excision. Consequently, it is possible that low

143

lipid concentration is a good indicator of tumour location and extent, in excised tissue. Moreover, lipid may be a more effective biomarker for in vivo tumour mapping as well, and going forward, researchers with access to high-fidelity photoacoustic imaging systems should make an effort to examine this approach. The preliminary results of the study presented here provide motivation for this direction. Finally, recently developed sophisticated lipidomic profiling methods have demonstrated outstanding diagnostic accuracy near 100%11. These findings independently indicate that, in differentiating malignant lesions, lipid content may be a promising new alternative to hemoglobin.

5.1.2 Optimizing wavelength selection for tissue discrimination

Beyond the two employed wavelengths, which targeted hemoglobin and lipids, other tissue chromophores should ideally be exploited to maximize the sensitivity and specificity of iPAT. For example, breast density, or the amount of fibro-glandular tissue in the breast, has been correlated to water concentration, and independently, to elevated breast cancer risk12,13. Specifically, increases in water concentration were associated with volumes containing increased fibrous and glandular tissue, while still higher concentrations were associated with malignant lesions. Furthermore, NIR optical methods have demonstrated excellent sensitivity to water content, indicating altered concentrations in most malignant breast tumours10. Therefore, use of an appropriate water-sensitive illumination wavelength during iPAT imaging of lumpectomy specimens may uncover new, potentially cancer-relevant, indicators of underlying tissue composition.

To that end, there are numerous wavelength selection possibilities in the deeper NIR optical spectrum where hemoglobin plays a reduced role in tissue absorption. For example, the spectral water peaks near 970 nm and 1180 nm seem logical staring points, as can be appreciated in Figure 5.1. On the other hand, the close spectral proximity of those peaks, to the 930 nm and 1210 nm lipid peaks, may cause reduced tissue-differentiation capacity. Consequently, a wavelength near 1140 nm may produce more effective results.

144

Similarly, the protein collagen and to a lesser degree, elastin, are found throughout the breast, and indeed the entire body. Furthermore, it is well known that malignant neoplasms frequently induce a desmoplastic reaction via elevated secretion of collagen, ultimately leading to an area of increased stiffness, or a palpable lump14. Importantly, more recent studies have linked the degree of collagen concentration within a malignant lesion to an increased propensity of tumour metastasis15,16. Therefore, in addition to hemoglobin, lipids, and water, collagen may also be a significant indicator of malignancy, and perhaps may even provide insight into the aggressiveness of tumours.

Inspecting Figure 5.1 reveals potential wavelengths that could be used to sway iPAT contrast towards collagen. For example, the absorption peak near 1170 nm may be a good option. However, the high absorption of water and lipids near that spectral region may overshadow collagen contrast. Instead, a wavelength of 1040 nm, where both lipids and water play reduced roles, might yield more effective collagen-biased imaging results. Ultimately, in practice, wavelength optimization will likely require consideration of both theoretical and empirical evidence.

145

Figure 5.1. Optical absorption spectra of common breast tissue constituents in the 900 nm to 1300 nm range, including collagen, lipid, and water. Adapted from 17.