Investigation of Formalin-Fixed Tissue Optical Characteristics in the Range of 200–500 GHz Using Pulsed Terahertz Reflection Spectroscopy to Differentiate Oral Malignant, Benign, and Cyst

The application of Terahertz electromagnetic waves to diagnose oral cancer was investigated. A single case of formalin-ﬁxed oral squamous cell carcinoma (malignant), ameloblastoma (benign), and odontogenic keratocyst was examined using terahertz pulsed spectroscopy in the frequency span of 0.1–2 THz. The measured absorption coeﬃcient, refractive index, and the extinction coeﬃcient were reported to be high for malignant samples than benign and cyst. The THz results are validated with hematoxylin and eosin-stained microscopic images. The results demonstrate that the THz signal was shown to be consistently higher for the malignant sample compared to benign and the cyst. These results indicate that THz signals responded to the cell density by eliminating the eﬀect of water.


Introduction
Oral cancer is a term that refers to cancers of the tongue, mucosa, gingiva, salivary glands, and the floor of the mouth and palate. It is treatable with a good prognosis if detected early. However, delayed diagnosis is a major social and health care problem [1][2][3]. Owing to the nature of the disease and resulting morbidity, oral cancer is a major public health concern. Globally, WHO reports that 3.5 billion people are affected with oral disease [4]. India reported the highest number of oral cancer cases, with an annual diagnosis of 77000 new cases and 52000 deaths [5]. Comparatively, oral cancer is a greater issue in India than it is in the west, with 70 percent of cases being reported at an advanced stage (Stage III-IV). Only 20 percent of patients survive five years because of the late diagnosis of cancer [6].
Globally, 84-97% of oral cancer contributes to oral squamous cell carcinoma (OSCC). Oral cancer is related to tobacco use, particularly smokeless tobacco (SLT), chewing pan, poor oral hygiene, excessive alcohol use, and a nutrientdeficient diet, such as the human papillomavirus (HPV).
ere is significant variation in worldwide incidence owing to a variety of reasons such as a lack of knowledge, exposure to dangerous meteorological conditions, and psychological threat [7,8].
Numerous conventional imaging modalities are used to assess, diagnose, and plan treatment for oral cancer. is includes physical and histopathological examination, staining, biopsy conventional radiography, traditional CT, magnetic resonance imaging (MRI) [9], and functional imaging using positron emission tomography (PET in combination with conventional CT [10]. Noninvasive diagnostic techniques such as visual autofluorescence play a vital role in diagnosing oral cancer. VELscope ® is a nonmagnifying hand-held instrument approved by the FDA in 2006 for direct visualization of oral mucosa autofluorescence [11]. Lane  e results showed that the method can distinguish OSCC from normal oral mucosa with good sensitivity and specificity [12]. Future researchers intended to improve the device's specificity for wider clinical usage. Because imaging scans are difficult to interpret in diagnosing SCC, a sample of the lesion is required for histopathological testing and subtyping. Malignant tissues are often identified by biopsy, which necessitates invasive tissue sampling. e intraoperative frozen section biopsy is commonly employed to differentiate the cancerous tissue from the normal. Manually inspecting slides stained with H&E and reviewed by pathologists for disease diagnosis are a laborintensive and time-consuming process. But most crucially, this procedure involves subjective judgments that might be flawed and is vulnerable to huge inter-and intraobserver variations. Based on H&E-stained pictures, even highly trained pathologists have difficulty reaching a consensus on the diagnosis of cancers because of their variety and complexity. Such procedures are exact, but invasive and only performed during surgery, which results in delayed diagnosis and poor prognosis. e below Table 1 shows the various diagnosing techniques using oral tissue for the detection of oral cancer. Table 1 summarizes the current noninvasive diagnosing techniques of the oral tissue specimen.
ere is a need for a noninvasive method that can detect the presence of malignant tissues. Based on the malignant tissue property that contains more water [24], we propose the use of THz to differentiate between benign and diseased tissues. Since THz wavelengths display better penetration into fats than microwaves, they have a greater imaging resolution. Another benefit of THz imaging over other approaches is its ability to extract the spectral data from the time domain terahertz waveform through Fourier transformation. is method allows one to determine the sample's absorption coefficient and the refractive index [25,26].
Terahertz is positioned between the microwave and infrared areas, which include frequencies span of 0.1 to 10 THz (1 THz � 10 12 Hz), which range from 3 mm to 30 μm analogous to wavelengths. Radiations are not ionizing, harmless for medical imagery at these frequencies [27]. However, due to the absence of effective "sources and detectors," the THz region has remained undiscovered for a long time. It was colloquially known as "THz Gap." Later, the same year, David Auston of AT & T Bell Laboratories created a photoconductive emitter gated by an optical pulse, laying the foundations for bridging the THz gap [28]. Coherent detection of THz radiation has enabled THz-TDS and its imaging techniques to explore in various fields such as pharmaceuticals, physics, astronomy, security screening, and medical imaging [29]. Indeed, biomedical research has accelerated in recent years. For instance, terahertz imaging has been used to diagnose skin cancer [30][31][32], breast cancer [33,34], and colon cancer [35] as well as dental cancer [36]. Terahertz spectroscopy was also utilized for the research of protein spectroscopy [37] and the analysis of the characteristics of crystalline molecules [38], including water-protein interactions [39].
One of the fundamental works in this field has been conducted by TeraView, UK, who have reported performing long-term systemic studies on the use of THz in diagnosis of tumor.
eir publications report includes THz based characterization of tissue properties, imaging of affected tissue, and the mechanism of tissue mass interaction with the THz wave. e majority of tissues studied by THz spectroscopy are classified into three types paraffin embedded tissues [40,41], fresh tissues [42], and frozen tissues [43,44].
e study revealed the variations in THz frequency span of absorption coefficient and the refractive index of frozen or fresh skin, colon, and breast cancer tissues compared to normal tissue. It is mainly related with an increase in water content in a cancer tumor, which may be easily detected using THz spectroscopy [45]. However, due to the connection of numerous illnesses with excess fluid, the specificity of such criteria is not great.
To date, there have been no studies comparing absorption, refractive, and extinction coefficient of formalin-fixed oral cancer, ameloblastoma (benign), and cyst. Previously, studies reported [46] measurement of oral tissue malignant using THz reflection imaging at frozen temperature and room temperature. e results demonstrated that the image contrast obtained at frozen temperature is better than the room temperature. e investigation of the tissue optical properties, say the absorption coefficient, the extinction coefficient, and the refractive index, is presented in this paper. Single cancer tissues were imaged using the reflex module's THz timedomain spectroscopy method. e tissue optical properties were compared to the histopathological findings. We hypothesize that there is a dominating component impacting the THz reflection signal other than water content to demonstrate the distinctions between normal and cancerous tissues. e obtained results show the effectiveness of using terahertz time-domain spectroscopy and imaging technologies to diagnose oral cancer and, hence, are clinically significant.

Samples.
A single case of oral malignant, benign (ameloblastoma), and cyst specimens is collected from the pathologist between April 2019 and May 2020 who underwent an operation for oral melanoma. e institutional research board and the hospital ethical committee have approved our study protocol. e activities are carried out in compliance with the Helsinki Declaration's instructions. [47]. Further, the tissue is sliced into thin samples of 1.5 cm × 1.2 cm × 0.3 cm in dimension and moved in for measuring using THz reflection spectroscopy.

Terahertz Pulsed Spectroscopy.
Menlo Systems GmbH's Tera K15 THz-TDS equipment is utilized for the measurements.
e arrangement is depicted schematically in Figure 1. A femtosecond laser generates 90 fs 1550 nm pulses at an 80 MHz rate, which excites the InGaAs semiconductor in the emitter photoconductive antenna (PCA), resulting in a picosecond pulse with a frequency range of 0.1 THz-2.5 THz. To concentrate the THz beam on the sample, two pairs of gold-coated off-axis parabolic reflectors were used.
Because of the nature of the THz signal (>10 W), a lockin amplification (LIA) is used to separate it from the thermal history. ermal background noise totally overshadows the  signals due to the THz-TDS method's coherent and gated detection. e delay line is controlled by a clock, which reads the lock-in amplification values at every single point and reports the terahertz pulsed waveforms. As predicted, the THz field's phase and amplitude data are received concurrently. e difference between the THz properties of the recorded samples defines the variance between the THz pulses generated, Eref, and THz-pulses transmitted [48].

Optical Parameter Extraction.
As previously mentioned, the optical tissue characteristics of a specimen can be assessed by transmission or reflection. For media with a strong THz absorption coefficient, such as raw biological specimen, there is a tissue thickness limit for Terahertztime domain spectroscopy in transmission geometry. Beyond the specified limit, reflection geometry can be used ( Figure 1). As shown in Figure 2(a), the terahertz signal is computed precisely as a function of time, and the frequency spectra of both the sample material signal and the reference signal (without sample) are produced using a numerical Fourier transformation. Further measurements on the obtained spectra offer information about the sample material under examination. As measurements are performed rather than intensity in the electric field, both amplitude and phase may be monitored at the same time. e absorption coefficient and the refractive index of tissues in the spectrum domain between 0.2 and 1.5 THz are depicted in Figures 2(a) and 2(b) of the spectral lines. us, the complex reflection coefficient is simply the Fresnel reflection coefficient [49].
Here, the reference signal S R,the ref is obtained by using a metallic mirror instead of the sample, which is assumed to be a perfect reflector. e value of the refractive index n R � n R − jk R extracted from reflection measurement is directly calculated by inverting expression (1): e primary disadvantage of reflection measurements is the inaccuracy introduced by even a slight misalignment of the reference mirror Δx concerning the sample position.
is results in the following phase error: And thus, errors on the extracted value of the refractive index where α R � 2ωK R /c is the material's absorption coefficient.
With regard to (4), the relative errors on the real and imaginary components of the complex refractive index change as the material becomes more absorbent. Indeed, the error introduced on n R by the misalignment Δx grows larger as α R rises, but the error made on k R decreases until the absorption coefficient reaches 2ω ������� n 2 R − 1/c. As a result, for highly absorbing material, errors in the optical characteristics of the tested sample are relatively large and should be rectified numerically or experimentally [50]. Even though this methodology is imprecise, it enables the estimation of the refractive index of the absorbing material even in spectral bands with high absorption.
As shown in the following section, obtaining the frequency-dependent characteristics is slightly more difficult in reflection geometry than in transmission geometry. Similarly, all sample and the reference signals are captured during transmission. Further, the temporal shift is caused by the reference window content, an additional spatial shift occurs due to absorption, and the refraction data are stored within the measurement's phase detail, which is susceptible to devise artifacts. Additionally, there is another issue that arises due to the polarisation shift caused by reflection at a certain angle, which results in sample and the reference pulses being polarised differently.
is is significant because the detector's sensitivity is polarizationdependent [51].

Limitation due to Formalin Fixing.
Since the samples will be utilized in another study, formalin fixation must be conducted prior to THz measurements. is was not appropriate because, in addition to stiffening the samples, the fixation technique changes the terahertz characteristics of the tissues [52], which may induce tissue shrinkage. Our THz measurements would have a more difficult time identifying and comparing the reflections of distinct tissue layers as a result of the fixing technique. Given the positive results achieved using formalin-fixed samples despite these limitations, this investigation serves as a catalyst for further research employing fresh tissue samples. Figures 2(a)-2(e), it depicts the reflected THz pulses and their spectrum power levels in oral cancer tissue of various kinds, such as benign, malignant, and cyst.  e increase in the impulse function of oral malignant, benign, and cyst is subtle; the obvious difference is the peak altitude. e result emphasizes that there is a sufficient variation in tissue optic characteristics to be observed in geometry of reflection. Figure 2(c) shows the difference in refractive index (n) of the oral cancer, benign, and cyst. e maximum spectra obtained from 0.2 to 0.3 THz are solely contributed not only by the water component, but also by the increased cell density.

Assessment of Tissue Optical Parameter and THz Spectroscopy Data. As seen in
Figures 2(d) and 2(e) present the calculated absorption coefficient (α) and the extinction coefficient of benign, malignant, and cyst samples. All absorption characteristics are followed by a distinct shift in the refractive index. Comparing the n and α spectra of benign, malignant, and cyst samples, the malignant sample shows higher absorption based on 10% of the water that is present in formalin-fixed samples.
e presence of water with absorption is to be around 140 cm −1 at 0.3 THz. It is likely understood that, along with water, the functional group like proteins or even the increased cell density plays a major role for the higher absorption spectra. e decrease in amplitude emphasizes the absorption of the sample from the beam's reflection via sample and their negligible losses.
Figures 2(f )-2(h) show the THz refractive index, absorption coefficient, and extinction coefficient of oral cancer, benign, and cyst samples. e error bars are overlapped, and the total error bar reflects a 95% confidence interval. eir THz absorption coefficient trend was most likely influenced by the liquid water in their tissues. In the frequency range of 0.2-0.5 THz, there is also a significant variance in tissue absorption coefficients, with the largest difference occurring for oral cancer at 140 cm −1 at 0.37 THz and 120 cm −1 for oral benign at 0.34 THz. e THz spectra of oral cancer, benign, and cyst are distinguishable in the frequency range of 0.2-0.5 THz.
Another factor that contributes to contrast in this study is tissue density, tumor microenvironment conditions (which are quite different from those of normal tissue, and undetermined cell division leading to increased cell density, as well as the presence of certain proteins). And this is linked to variations produced by cell modifications and aberrant protein density changes. e rise in vascularisation surrounding tumors, an increase in cell division and cell density, results in the release of growth factors. e difference between the absorption coefficient and refractive index spectra of benign, cyst, and malignant samples is shown in greater detail in Figures 2(c) and 2(d).
e disparity between two spectra is most significant at 0.42 THz. e increase in the variance at 0.42 THz is more likely to be caused by water interaction with a functional group such as proteins or increased cell/protein density alone [53].

Validation of Results.
To perform this validation, formalin-fixed tissues were dehydrated by replacing the water with paraffin. Typically, tissues are preserved in 10% neutral formalin, and multiple ethanol solutions are used with increased concentration (say 70%, 80%, and 90%) to dehydrate the water from the sample. Following that, at a temperature of 60°C, the material was saturated with paraffin. Paraffinembedded tissue bits were put out in specific moulds (regular plastic material with holes is utilized to better attach paraffin) and filled with molten paraffin at 60°C. e embedded paraffin at room temperature is then taken manually sectioned with a microtome to generate paraffin slices measuring 4-5 μm thick. Sections are then dewaxed. Finally,  Journal of Spectroscopy without immersing, thin sections were impregnated in glass slides, stained with haematoxylin and eosin (H & E), and evaluated under a light microscope as shown in Figure 3.
Haematoxylin is a standard stain dye used in this process; it gives the nuclei a bluish colour, while eosin (another stain dye used in histology) gives the cytoplasm and other tissue components a pinkish-orange stain [54]. To examine the differences between the benign, cyst, and cancerous tissues at the cellular level under the microscope, the cell density of cancer tissue is substantially high. At the same time, accelerated development of malignant/tumor cells set off considerable increases in tissue nuclei and often significantly more nucleoli in oral cancer cells [55,56]. e result was tested using the image j software tool [57] by measuring the proportion of cell nucleus, cytoplasm per unit area of tissue. Nuclear recognition findings are seen in Figures 4(a)-4(c), and the proportions of each factor are calculated in Table 2. e findings indicate that the cancer tissue has almost double the number of nuclei than the healthier tissue, implying a significant rise in cell density. When the same cellular materials are used, a higher cell density often results in a higher refractive index and absorption coefficient. Table 2 shows that as the internal cell density of malignant, benign, and cyst increases, the substantial increase in the refractive index and the absorption coefficient of malignant, benign, and cyst is observed. e internal density of the cell nucleus from cyst to malignant varies from 17% to 37%, whereas there occurs a higher reflected peak from cyst to malignant, i.e., 1.6 to 2.8, and absorption coefficient from cyst to malignant is 80 cm −1 to 140 cm −1 .
However, to understand the detailed correlation between the cell density and the absorption coefficient, a large number of tissue samples for the clinical study are required. By overcoming the challenge in obtaining the multiple tissues, the future direction of this study can be formulated.

Conclusion
Terahertz spectroscopy can be utilized to distinguish oral malignant tumour from benign and cystic tissue since the estimated refractive indices and absorption coefficients were high in tumour tissues despite their comparatively low water content. is indicates that terahertz can help oral surgeons diagnose cancer tumor rapidly and correctly based on clinical distinctions observed in the lesions. THz can be utilized as an experimental generic technique for identifying oral cancer. Although this method will help new surgeons discover cancer lesions in formalin-fixed paraffin blocks, which is the gold standard for diagnosis, THz spectroscopy may be a helpful approach for noninvasive, effective intraoperative diagnosis of oral cancer, according to the current data.
Data Availability e data are available upon request to the corresponding author.
Ethical Approval e study was approved by institutional/regional/national ethics/committee/ethics board of SRMIST/1509/IEC/2018, and

Consent
Individual consent for this retrospective analysis was waived.