EP4630810A1 - Method and system for detection of biomarkers using radiation - Google Patents

Abstract

A method is disclosed for classifying a target comprising tissue or an extract of tissue as having one or more biomarkers present. A Laser Feedback Interferometer (LFI) is operated with each of an array of THz Quantum Cascade Lasers (QCLs) activated in turn to apply laser radiation at their corresponding predetermined wavelength to the target. Optical responses at each of the predetermined wavelengths are detected by the LFI. Background effect subtraction and scaling is then performed by a processing assembly which classifies the target as having a biomarker present based on the values detected by the LFI at each of the predetermined wavelengths. In the case of the biomarker being DNA extracts that are associated with melanoma then the biomarker is deemed present if reflection amplitude at λ1 is less than reflection amplitude at λ2; and reflection amplitude at λ2 is less than reflection amplitude at λ3. The originating tissue from which the target under test was made can then be appropriately labelled as "biomarker present" or "biomarker absent".

Description

METHOD AND SYSTEM FOR DETECTION OF BIOMARKERS USING RADIATION

RELATED APPLICATIONS

Priority is claimed from Australian provisional patent application No. 2022903756 filed 8 December 2022 and from Australian provisional patent application No. 2022903774 filed 9 December 2022, the contents of which are each hereby incorporated herein by reference.

TECHNICAL FIELD

The present disclosure is concerned with a method and system for detection of one or more biomarkers, such as biomarkers of malignant cell conditions.

BACKGROUND

There are many situations where it would be desirable to be able to rapidly and conveniently detect the presence of biomarkers associated with particular conditions, such as a malignancy of interest, in tissue, including that of plants and also of animals, such as humans.

For example, skin cancer is a malignancy that is the most common cancer diagnosed in Australia and the US, with estimates in 2021 of 215,000 new melanoma cases leading to 8,500 fatalities in both countries. If metastasis occurs to lymph nodes or to distal organs (e.g., brain), the chances of survival are ~10% without further treatment. Current late-stage therapies have progressed in recent years with clinical trial data showing up to 50% overall 5-year survival rates. While early diagnosis remains the principal determinant of improving survivability (97-99% 5-year survival), the negative consequences of the overdiagnosis of melanoma are less obvious, and have yet to be fully addressed.

Existing optical skin imaging techniques, such as dermoscopy and confocal reflection microscopy, are based on assessing lesion morphology and color or morphology on the cellular level. A common weakness of these methods is the reliance on cellular architecture or macroscopic image for identification of the disease, rather than directly interrogating molecular and genetic changes associated with malignancy. This results in a large portion of skin cancers being found when they are visible and at invasive stages that have the potential to metastasize to distal organs and perhaps leading to death. In addition, the sensitivity of expert histopathology examination is approximately 80%. Diagnostic errors contribute to approximately 10% of patient deaths and are the top cause of medical malpractice payouts. A recent investigation performed in the USA shows that diagnostics spanning from moderately dysplastic naevus to early stage invasive melanoma were neither reproducible nor accurate. The introduction of standardized classification tools that employ unambiguous diagnostic technology, have a great potential to reduce communication and management errors. The detection and the triage of a skin cancer encompasses a broad spectrum of signs even in one sample between healthy skin and a well-developed carcinoma. However, one of the key current issues in melanoma screening is the incorrect diagnosis of a so called dysplastic naevus and its cascading effect to unnecessary excision of what is perhaps a benign naevus or a non-diagnosed early melanoma. The unmet need is to objectively determine where a given lesion on the cancer development spectrum, so a proper clinical decision can be made.

It is an obj ect of the present disclosure to provide a method and/or a system for detecting one or more biomarkers that are associated with a malignancy of tissue.

SUMMARY

According to a first aspect of the present disclosure there is provided a method for classifying a target comprising tissue or an extract thereof as having one or more biomarkers present, the method comprising: for each one of a plurality of wavelengths, the wavelengths being predetermined to elicit an optical response from the target that is characteristic of the one or more biomarkers, applying radiation from one or more radiation sources at the wavelength to the target; and detecting an optical response of the target at the wavelength, to thereby detect optical responses of the target at each of the wavelengths; and classifying the target as having the biomarker present based on the detected optical responses of the target. In an embodiment the one or more radiation sources comprises one or more Quantum Cascade Lasers (QCL) and the radiation comprises laser radiation.

In an embodiment the detecting of the optical response of the target at each of the plurality of wavelengths includes self-mixing of the laser radiation applied to the target with laser radiation returned from the target.

In an embodiment each of the one or more radiation sources is configured to operate in a frequency range between 2 THz and 6 THz.

In an embodiment the one or more radiation sources comprise a plurality of radiation sources, each configured to emit radiation at a corresponding one of the plurality of wavelengths.

In an embodiment the target includes a medium for containing the extract of the tissue.

In an embodiment the detected optical response is processed to account for a background optical response including an optical response of the medium.

In an embodiment wherein the one or more biomarkers comprise DNA extracts associated with cancerous cells.

In an embodiment the one or more biomarkers comprise DNA extracts associated with cancerous cells being melanoma.

In an embodiment the method includes: applying radiation to the target at first (Xi), second (X2), and third (X3) wavelengths of the plurality of wavelengths, wherein Xi > X2 > X3 and the detected optical responses comprise, reflectance and/or reflection phase shift of the target; or absorbance and/or transmission phase shift, at each of Xi, X2, and X3. In an embodiment detected optical responses comprise absorbance and the classifying of the target as having a biomarker present that is associated with melanoma is made if: absorbance at Xi is less than absorbance at A. and absorbance at Z2 is less than absorbance at X3.

In an embodiment Z2 corresponds to a wavelength at or adjacent to which a scaled absorbance curve of DNA extracts of healthy tissue crosses a scaled absorbance curve of DNA extracts of melanoma tissue.

In embodiment the method includes determining the pre-determined wavelengths that elicit the optical response from the target that is characteristic of the one or more biomarkers by: applying laser radiation over a range of wavelengths to one or more targets known to contain the one or biomarkers and known to not contain the one or more biomarkers to thereby obtain one or more biomarker optical response curves characteristic of presence or of absence of the one or more biomarkers.

In an embodiment the method includes processing the one or more biomarker optical response curves to determine the pre-determined wavelengths of the one or more radiation sources that elicit the optical response that is characteristic of the one or more biomarkers.

In an embodiment the processing of the one or more biomarker optical response curves determines the pre -determined wavelengths that create optical responses for optimal discrimination between targets in which the one or more biomarkers are present and targets in which the one or more biomarkers are absent.

In an embodiment the processing of the one or more biomarker optical response curves includes analyzing the optical response curves using Partial Least Squares Discriminant Analysis (PLSDA) to determine the pre-determined wavelengths that facilitate optimal discrimination.

In an embodiment the optical response curves are processed to account for a background optical response including an optical response of a medium of the target. In an embodiment the optical response curves are scaled to facilitate comparison of optical response of targets containing the one or more biomarkers with optical response of targets not containing the one or more biomarkers.

According to a further aspect of the present disclosure there is provided a system for classifying a target comprising tissue or an extract thereof as having one or more biomarkers present, the system comprising: one or more radiation sources, each being configured to emit radiation at each of a plurality of wavelengths, the wavelengths being predetermined to elicit an optical response from the target that is characteristic of the one or more biomarkers; a detector assembly for detecting the optical response from the target; a processing assembly responsive to the detector assembly and configured to: classify the target as having the one or more biomarkers present based on the detected optical responses.

In an embodiment the plurality of radiation sources comprises a plurality of radiation sources.

In an embodiment the one or more radiation sources comprise one or more Quantum Cascade Lasers (QCL).

In an embodiment the one or more QCLs comprise one or more THz QCLs, wherein the detector assembly is responsive to self-mixing electrical signals of the one or more THz QCLs.

In an embodiment the QCLs comprise at least three QCLs.

In an embodiment each of the plurality of radiation sources is configured to operate at a frequency between 2 THz and 6 THz.

In an embodiment the processing assembly is configured to subtract a background optical response effect of a medium of the target from the optical responses of the target to the radiation at each of the wavelengths. In an embodiment the one or more biomarkers comprise DNA extracts of melanoma and wherein the one or more radiation sources operate at wavelengths 1 2 3 respectively and the detected optical responses comprise, absorbance of the target, at each of Xi, X2, and X3; wherein the processing assembly is configured to classify the target as having a biomarker present that is associated with melanoma if: absorbance at Xi is less than absorbance at X2; and absorbance at X2 is less than absorbance at X3.

According to another aspect of the present disclosure there is provided a method for classifying a test target as containing one or more biomarkers, the method comprising: making a number of targets in which the one or more biomarkers are absent (“non-biomarker targets”); making a number of targets in which the one or more biomarkers are present (“biomarker targets”); recording optical responses of each of the non-biomarker targets (“nonbiomarker responses”) to radiation over a range of wavelengths; recording optical responses of each of the biomarker targets (“biomarker responses”) to radiation over a range of wavelengths; and where the non-biomarker responses differ from the biomarker responses then processing the bio-marker responses and the non-biomarker responses to determine a number of different wavelengths (“pre-determined wavelengths”) of radiation for subsequently eliciting an optical response from a test target to discriminate between the test target being either a biomarker target or a non-biomarker target.

In an embodiment the processing of the non-biomarker responses and the biomarker responses includes Partial Least Squares Discriminant Analysis (PLSDA) to determine the pre-determined wavelengths that facilitate optimal discrimination.

In an embodiment the recording of the non-biomarker responses and the biomarker responses are made using a wideband laser source such as a synchrotron. In an embodiment the pre-determined wavelengths are applied to the test target using one or more narrow band lasers such as Quantum Cascade Lasers.

BRIEF DESCRIPTION OF THE DRAWINGS

Preferred features, embodiments and variations may be discerned from the following Detailed Description which provides sufficient information for those skilled in the art to perform the various aspects and embodiments discussed herein. The Detailed Description is not to be regarded as limiting the scope of the preceding Summary in any way. The Detailed Description will make reference to a number of drawings as follows:

Figures 1A to ICd comprise a number of graphs of absorption spectra of DNA extracts (ranging from 135.13 ng/pl-151.51 ng/pl) from skin pathologies with synchrotron radiation: Figure 1A. normalized absorbance by TE-buffer of all samples; Figure IB. z-scored absorbance normalized by TE buffer; Figures ICato ICd, zoomed in absorbance between 2.5 THz to 4.5 THz for benign naevus, dysplastic naevus, melanoma in situ, and thin melanoma respectively.

Figure 2 is a graph of the absorption spectra of DNA extracts from skin pathologies with synchrotron radiation for prepared concentrations (1.88 ng/pl-393.94 ng/pl) from Healthy Skin, Thin Melanoma, Melanoma in Situ, Dysplastic Naevus and Benign Naevus tissue.

Figure 3 is a graph of normalized absorbance by TE buffer of all samples graphed in Figure 2 with z-scored absorbance normalized by TE buffer.

Figure 4 is a graph of mean absorbance spectra of two melanoma and two naevus groups with shaded regions corresponding to the 95% confidence interval. The inset panel corresponds to the partial least squares discriminant analysis (PLSDA) scores of individual samples, which was fit to discriminate between melanoma and naevus samples using the frequency range 2-4.5 THz. The bar plot at the bottom visualizes the (absolute) value of the PLSDA coefficients with the most important frequencies highlighted.)

Figure 5 is a detail of a first one of the inset panels of Figure 4. Figure 6 is a detail of a second one of the inset panels of Figure 4.

Figures 7A to 7D comprise four graphs of synchrotron radiation median absorption spectra of DNA extracts from skin.

Figures 8Aa to 8Cb comprise graphs showing PLSDA separation for transmission measurements.

Figures 9Aa to 9Cb comprise graphs showing PLSDA separation for ATR measurements.

Figures 10A to 10B are kernel density plots of DNA extracts with the graph of Figure 10A being of transmission measurements and the graph of Figure 10B being of ATR measurements.

Figures 11A to 11C depict overlapped spectra of five types of DNA extracts from skin pathologies.

Figure 12 is a diagram of an absorption spectrum measurement system with synchrotron THz beamline radiation including portions a-d that respectively show: a) A Michelson interferometer for spectrum measurements for the DNA samples; b) a cross-section of the liquid cell with two diamond windows sandwiched the DNA samples (1.5 pL). The thickness of the liquid sample is 15 pm; c) a target comprising a suspension of DNA in the TE buffer, DNA strands are held together by hydration bonds between bases on adjacent strands the TE buffer; d) Original measured spectra and the spectra of DNA by removing the effects of background optical responses, including optical responses of the liquid cell and of the TE buffer solution. Figure 13 is a graph comparing the absorption spectrum of tyrosine at various concentrations with reported data (lowest curve).

Figure 14 comprises images of slides of H&E stained images of skin pathologies from where DNA was extracted: a. benign naevus; b. dysplastic naevus; c. Melanoma in situ; d. thin melanoma.

Figure 15 is a schematic diagram of a number of radiation sources, each at a different wavelength, in the form of an array of three THz QCLs (Terahertz Quantum Cascade Lasers) mounted on a cold finger of Stirling cooler for detecting an optical response of a target at each wavelength for detecting the presence of one or more biomarkers.

Figure 16 is a flowchart of a procedure for determining wavelengths for eliciting a discriminating optical response from a test target for one or more biomarkers of interest.

Figure 17 is a graph made using the synchrotron system of Figure 12, showing optical response curves, in the form of scaled absorbance values, targets made from malignant tissue and targets made from healthy / non-malignant tissue and with three wavelengths superimposed therein for discriminating between targets made from healthy / non- malignant and malignant tissue.

Figure 18 is a flowchart of a procedure for classifying targets.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

Terahertz (THz) waves (frequency range: 0. 1 to 10 THz), since they match the intra and inter-molecular vibration domains, are sensitive to the conformation and structure of molecules. In particular, they are ideal probes to study the dynamics of hydrogen bonding, van der Waals, interactions and molecular level interactions with their surroundings. The unique properties of THz radiation make it very promising for medical and biological imaging and spectroscopy applications, such as detection and imaging of cancers. The medical applications of THz technology stem from several key characteristics. Firstly, THz radiation is highly sensitive to water content and blood flow in tumours, through the ubiquitous presence of hydrogen bonds of water. Secondly, many amino acids, protein molecules and cancer DNA exhibit unique characteristic spectral responses at THz frequencies that can be used as biomarker proxies for cancer detection. Thirdly, being a sub-millimetre wavelength wave, THz radiation is non-ionizing and therefore intrinsically biologically safe.

The Inventors have investigated the THz absorption spectra of human skin DNAs extracted from benign naevus, dysplastic naevus, melanoma in situ (non-invasive), and thin melanoma (<1 mm thick- ness), respectively (see Figure 1A). The results strongly indicate that the difference between healthy skin, benign naevi and melanoma from the THz absorption spectrum can be observed. As will be discussed, the Inventors have identified spectral windows that they have found can be used for discriminating between the presence and absence of biomarkers of interest in targets made of tissues or extracts of tissues such as DNA extracts. For example, the Inventors have found that it is possible to separate melanoma and naevus samples. Surprisingly, the Inventors have found that this delineation can be achieved by using a small number of discrete spectral lines (easily accessible by using a radiation source such as a semiconductor laser) without the need for a broadband source like a synchrotron. This provides for an early-stage detection system for biomarkers of interest, such as biomarkers of melanoma, to assist in accurate and timely diagnosis.

Whilst a preferred embodiment will be described with reference to melanoma detection, other embodiments and aspects are more generally applicable to detection of biomarkers and disease signatures in tissue and tissue extracts, including those of plants.

Results - Initial Investigations

Initially the Inventors investigated a number of DNA specimens using a synchrotron. The specimens that were investigated are summarized in Table 1. A description of the experimental setup built around the Synchrotron THz beamline is provided later in this specification in Figure 12 and under the sub-heading Materials and Methods. Specimen ID Sex Age at diagnosis Category

BN1(13O19-18BR 1A) Female 30 Benign naevus

BN2(16445-18BR 1A) Male 35 Benign naevus

DN 1(100452- 18BR 2A) Female 32 Dysplastic naevus

DN2(98104-18BR 1A) Male 52 Dysplastic naevus

MIS 1 (29245 - 18BR 1C) Female 71 Melanoma in situ

MIS2(52206-18BR ID) Male 69 Melanoma in situ

TM1(71116-18BR IB) Female 49 Thin melanoma

TM2( 107030- 18BR 1 A) Male 70 Thin melanoma

HS 1 & HS2 Male N/A Healthy skin

Table 1: DNA Specimens Investigated

DNA suspensions (i.e. “targets”) exhibited substantial broadband (1-8 THz) spectral differences between the four categories (benign naevus (BN), dysplastic Naevus (DN), Melanoma in situ (MIS), thin melanoma (TM) and healthy skin (HS)), which were further enhanced by z-score normalization. (For clarity, Figures 1A to ICd show the data for the samples from 0.5 pg aliquot with concentrations ranging from 135.13 ng/pL to 151.51 ng/pL, all data can be seen in Figure 2 and Figure 3). Closer inspection of the baseline-corrected spectra between 2.5-4.5 THz revealed several category-specific spectral signatures, superimposed on the larger broadband fluctuation. There is a striking difference between the spectra of the HS, BN, and both MIS and TM samples (i.e. “targets”).

The melanoma samples (MIS & TM) all show the same broadband trend and fine spectral features and are significantly different from the non-melanoma samples. In other words, based solely on their THz fingerprints one can clearly distinguish between the DNA extracts of melanoma, naevus, and healthy skin.

Cross-decomposition analysis of all DNA suspensions [57.14 - 393.94 ng/pl (See Figure 2 and Figure 3 for all data)] with a partial least squares regression discriminant analysis utilizing the first-order derivative of the absorption spectra in the 2 - 4.5 THz range achieved relatively good separation of melanoma and naevus samples. Figure 2 and Figure 3 are graphs of the absorption spectra of DNA extracts from skin pathologies with synchrotron radiation for all prepared concentrations (1.88 ng/pl- 393.94 ng/pl). Figure 2 shows the normalized absorbance by TE buffer of all samples and Figure 3 shows the z-scored absorbance normalized by TE buffer. Figure 2 and Figure 3 illustrate the complete data set of all the DNA samples and concentrations after removal of the TE-buffer background and were used to create the averaged traces and confidence bands show in Figure 4. This data set comprises of 130 measurements from 10 samples at several DNA concentrations from 9 individuals (4 Females, 5 Males, age at diagnosis 30-71 years).

Figure 4 is a graph of mean absorbance spectra of two melanoma and two naevus groups with shaded regions corresponding to the 95% confidence interval. The upper inset panel of Figure 4, shown in detail in Figure 5, corresponds to the partial least squares discriminant analysis (PLSDA) scores of individual samples, which was fit to discriminate between melanoma and naevus samples using the frequency range 2-4.5 THz. The bar plot at the bottom of Figure 4, shown in detail in Figure 6, visualizes the (absolute) value of the PLSDA coefficients with the most important frequencies (THz) highlighted.

Furthermore, having performed the feature importance analysis of the model, the Inventors found it relies primarily on contributions from a small number (three to six in this case) of individual frequency bands.

Subsequently the Inventors performed further investigations using the experimental setup of Figure 12. DNA suspensions of five categories as set out in Table 1: benign naevus (BN), dysplastic naevus, mild-moderate (DN), dysplastic naevus severe/borderline (SDN), melanoma in situ (MIS), thin melanoma, were measured in both transmission and attenuated total reflection (ATR) configurations. Each same type was comprised of 20 individual DNA samples extracted from a 0.5 pg aliquot producing DNA concentrations ranging from ~119 ng/pL to ~142 ng/pL).

The samples exhibited substantial broadband 2-6 THz) spectral differences between each of the types as illustrated in Figures 7A to 7D. Figures 7A to 7D comprise four graphs of synchrotron radiation median absorption spectra of DNA extracts from skin pathologies (shaded areas are standard deviations (SD) of the complete data set) Figure 7A shows the transmission normalized insertion loss by TE-buffer of all samples; Figure 7B shows the transmission z-scored scaled insertion loss normalized by TE buffer; Figure 7C shows the ATR normalized reflectance by TE-buffer of all samples; Figure 7D shows the ATR z-scored scaled and baseline removed reluctance normalized by TE buffer.

A striking difference is observable between the spectra of the HS, BN, and both MIS and TM samples. The melanoma samples (MIS & TM) all show the same broadband trend and spectral features and are significantly different from the non-melanoma samples. In other words, based solely on their THz fingerprints it is possible to clearly distinguish between the DNA extracts of melanoma, naevus, and healthy skin.

Cross-decomposition analysis of all DNA suspensions for all data with a partial least squares regression discriminant analysis utilizing for both transmission and ATR spectra revealed frequencies where the separation of melanoma and naevus samples was good as shown in the graphs of Figures 8Aa to 8Cb and Figures 9Aa to 9Cb. Figures 8Aa to 8Cb comprise graphs showing PLSDA separation for transmission measurements and Figures 9Aa to 9Cb is comprised of graphs showing PLSDA separation for ATR measurements.

Using two of the highest contributing frequencies form each of the results (transmission and ATR) gives the kernel density plots shown in Figure 10A and Figure 10B. Figures 10A and 1 OB are kernel density plots of DNA extracts with Figure 10A being of transmission measurements and Figure 10B being of ATR measurements. Both plots show excellent separation between melanoma and neavus samples with SDN sitting somewhere in between and show great potential for separating diseased from benign tissue using only a handful of THz frequencies. Measuring the absorption of skin DNA at these frequencies could be utilized in rapid triage diagnostics of suspected melanoma. Therefore, it may be possible to differentiate between different pathologies based on absorption measurements at only a few discrete frequencies. The Inventors realized that results from measuring the absorption of skin DNA at these frequencies could be utilized in rapid triage diagnostics of suspected melanoma and that it would be possible to differentiate between different pathologies based on absorption measurements at only a few discrete frequencies.

Materials and Methods - Initial Investigations

Figure 12 is a diagram depicting the system setup for the absorption spectrum measurement, that has been previously discussed, with synchrotron THz beamline radiation.

In Figure 12 the drawing portions referred to by letters a-d are, a. the Michelson interferometer for spectrum measurements for the DNA samples; b. the cross-section of the liquid cell with two diamond windows sandwiched the

DNA samples (1.5 pL). The thickness of the liquid sample is 15 pm; c. Suspension of DNA in the TE buffer, DNA strands are held together by hydration bonds between bases on adjacent strands the TE buffer; d. The original measured spectra and the spectra of DNA by removing the effects of the background, including the liquid cell and the TE buffer solution.

Terahertz spectroscopy measurements were conducted at the Australian Synchrotron (Clayton, Victoria) utilizing the high-resolution THz/Far-IR Beamline. The beamline was coupled to a Bruker IFS 125HR Fourier Transform spectrometer (Bruker Optics, Ettlingen, Germany) for the absorption spectrum measurements. The IFS 125HR Bruker spectrometer utilizes a Michelson interferometer (Figure 12) with an optical path length of 942 cm for single-sided data acquisition (16.8 GHz resolution). The THz beam from the synchrotron source was divided into two paths by 6 pm mylar multilayer beam splitter, part of the beam is reflected to a fixed Mirror 1 while some is transmitted through the beam splitter to a movable Mirror 2. Both beams recombine after introducing the path difference and create variations in the output beam intensity as the difference in the path length changes. The recombined beam passes through the sample before incident on a liquid-helium-cooled Si bolometer detector with aperture setting of 4 mm. The variation in the intensity of the beams seen by the detector is a function of the path difference and ultimately provides the desired spectral information in a Fourier Transform Spectrometer. The DNA sample (1.5 pL) was loaded on the base of a liquid cell and sandwiched by two diamond windows of the cell for spectrum measurements (Figure 12, b). The cell was then mounted on the sample holder and inserted in a Janis ST-100-FTIR cryostat (Lake Shore Cryotronics, Inc. Wilmington, MA) under vacuum where the spectra were recorded in the transmission mode. Each sample was measured five times independently and each measured spectrum was an average of 100 scans recorded with the maximum frequency limit of 500 cm-1 and at a resolution of 1 cm-1. The system was tested and calibrated by measuring several common amino acids, an example of which can be seen in Figure 13, which is a graph comparing the absorption spectrum of tyrosine and the reported data. The graph of Figure 13 shows the absorption spectra after baseline removal for tyrosine at 3 different concentrations (5, 10 and 20 % by weight), compared with that measured in the literature (Matei, A., Drichko, N., Gompf, B., & Dressel, M.: Far-infrared spectra of amino acids. Chemical Physics, 316(1-3), 61- 71 (2005)). The spectra not only scale well with concentration but match perfectly with the literature.

Preparation of human skin and DNA samples

Institutional approval of experiments involving human tissues was obtained from the relevant authorities. All naevi and melanoma tissues were derived from a specialist medical pathologist and diagnosed by expert dermatopathologists to provide a consensus diagnosis. Tissue samples from four categories (BN, DN, MIS, and TM), two samples in each category (one male, one female) were selected from a range of formalin- fixed paraffin embedded (FFPE) human tissue blocks. Histopathology assessment and diagnosis was carried out on prepared 5pm-thick H&E stained slides (See Figure 14). For comparison, DNA samples extracted from healthy human skin tissue were also included. For consistency, the healthy skin was also fixed and paraffin embedded prior to sectioning for DNA extraction. The details of the samples are summarized in Table 1.

Figure 14 comprises the H&E stained images of the skin pathologies where the DNA was extracted as follows: a. benign naevus; b. dysplastic naevus; c. Melanoma in situ; d. thin melanoma. The DNA samples were extracted from the FFPE human tissue blocks. 6-8 scrolls of 10 pm sections were collected from each pathology block under DNase/RNase free conditions, followed by DNA/RNA co-extraction using the Allprep DNA/RNA FFPE kit (Qiagen, cat. 80234). All the samples were eluted in TE buffer. All DNA samples were quantified using the Qubit dsDNA high sensitivity kit (Thermo Fisher, cat. Q32851).

Three sample aliquots containing different amounts of DNA were prepared from each sample: 200 ng, 500 ng, and 1300 ng. All samples were concentrated using a Savant SPD1010 SpeedVac concentrator system. The final volume was measured by pipette and the concentration was calculated accordingly based on the measurement. Each of the DNA samples were vortexed and centrifuged before the spectral measurement.

Spectral data processing

All spectral measurements were performed in transmission mode with the extracted DNA suspended in TE buffer. Absorbance of the DNA samples was calculated by subtracting the contribution of TE buffer according to Equation 1: where 7DNA is the transmission intensity of the DNA sample and /TE" the transmission intensity of pure TE buffer (see Figure 12 graphs in portion d of that Figure). To compare the absorption spectra between healthy and pathologic skin DNA, all spectra were scaled by their respective z-score, according to Equation 2: where A is the original absorbance spectrum, . is the mean of A, and a the standard deviation of A. Since z-score scaling failed to fully extract absorbance peaks from the spectra, a baseline correction was also performed according to Equation 3: using asymmetric least squares (ALS) (49). ALS baseline correction estimates the baseline by minimizing the cost function where y is the measured signal, /baseline, /. roughness factor, and A difference operator. Weights v are assigned asymmetrically: Vi = p, whcn y, >f and Vi = 1-p otherwise. Values for X and p were set to 10 and 0.01, respectively. The raw results of these processes are shown in the graphs of Figures 11 A to 11C. Figures 11A to 11C depicts overlapped spectra of all five types of DNA extracts from skin pathologies in each of these figures as follows:

Top graph/row of graphs: Transmission normalized insertion loss by TE-buffer of all samples;

2^nd from top graph/row of graphs: Transmission z-scored scaled insertion loss normalized by TE buffer;

3^rd from top graph/row of graphs: ATR normalized reflectance by TE-buffer of all samples;

4^th from top (bottom) graph/row of graphs: ATR z-scored scaled and baseline removed reluctance normalized by TE buffer.

A partial least squares discriminant analysis (PLSDA) classifier with two latent variables was trained to differentiate between naevus and melanoma by pooling the resulting absorbance spectrums of all available naevus and melanoma samples (n = 24). Details of PLSDA can be found in e.g. Barker, M., Rayens, W.: Partial least squares for discrimination. Journal of Chemometrics: A Journal of the Chemometrics Society 17(3), 166-173 (2003).

Prior to training the PLSDA classifier, the smoothed first derivative of each spectrum was computed using a 3rd order polynomial Savitzky-Golay filter with a 15-point (433 GHz) window length.

Classifying Samples with a THz QCL Array

As previously alluded to, current approaches in biomedical imaging are based on the belief that complete spectral fingerprint detection, and broadband THz platforms are required for identification of malignancy markers. However, existing THz platforms are either based on low-power TDS or narrow bandwidth THz QCLs. One of the main advantages of THz -TDS systems is their broadband nature which enables sensing over a range of frequencies. The main drawback of THz-TDS systems is the difficulty of generating sufficient power at frequencies beyond ~2 THz (thereby limiting both their effective signal-to-noise ratio and their accessible frequency range) and slow data acquisition.

Laser feedback interferometry (LFI) is one of the simplest coherent techniques, for which the emission source can also play the role of a highly sensitive detector. The combination of QCLs and LFI, while not the only option for implementing methods and systems according to the present disclosure, is particularly attractive for sensing applications, notably in the THz band where there is a lack of high-speed high- sensitivity external detectors. Indeed, LFI with THz QCLs — which enjoy high output power, low phase-noise, and the stability under optical feedback — has been successfully employed for sensing and benefits from the intrinsic advantages inherent to homodyning detectors, in particular suppression of unwanted background radiation. The use of the QCL itself as the detector means that the detection speed is theoretically limited only by the device itself.

Recently, the Inventors have conducted THz imaging for a set of sectioned skin tissues including nevi, dysplastic nevi and melanoma. A 50 pm microtomed section was imaged with a) conventional optical microscope, b) THz signal amplitude, and c) THz signal phase. At each pixel of the imaged skin, a time-domain signal is obtained, which, when translated into frequency domain, provides the spectrum of the interferometric signal, including its amplitude and phase. Even this simplest possible reduction of the time domain signal was found to contain information representative of the condition of the skin at the location being interrogated. When analysed and plotted in the amplitudephase space it showed clustering of pixels corresponding to healthy skin and the pathology into separate, distinct areas of the amplitude-phase space.

LFI with THz QCLs to date has only been investigated and utilized at single frequencies with narrow bandwidth. High power broadband emission from THz QCLs can be realised through a QCL array, or via multiple mode operation, modelocking or frequency combs. These generation mechanisms used in conjunction with the LFI scheme permit the sensing and imaging of a sample within a range of frequencies associated with a spectral fingerprint, without compromising on power levels or sensitivity. One or more of the present Inventors are also inventors of the QCL THz platform described in international patent application PCT/AU2014/000828 (now granted in a number of countries) and also in Sw ept-frequency feedback interferometry using terahertz frequency QCLs: a method for imaging and material analysis Aleksandar D. Rakic et al., Optics Express Vol. 21, Issue 19, pp 22194-22205 (2013). the contents of which are each hereby incorporated by reference. A THZ QCL array that builds on the advances described in those documents will now be discussed.

Figure 15 depicts an imaging apparatus that is based on THz QCL array that operates in a cryogen-free Stirling cooler in a pulsed biasing scheme. The THz QCL array consists of three QCLs. The emission from the beam-combined THz QCL array will be used in the LFI configuration for THz spectroscopic scanning.

Radiation from the QCL array is collimated and subsequently focused on the target using a pair of Tsurupica plastic lenses. The reflected beam from the target follows back along the forward optical path and return into each laser cavity, where interference with the original electrical field creates a self-mixing signal that will be detected from the terminal voltage of each QCL with driving current sweeping. The reflection coefficient and the phase shift of the target at these THz frequencies is then extracted from the set of LFI signal collected from the lasers’ terminal voltage and used to extract the absorbance at these frequencies. Each THz QCL of the THZ QCL array is operated in specific frequency bands with the required tuning range within each band as dictated by the THz windows that have been previously discussed.

The time-domain LFI signals collected from the terminal voltage of multiple QCLs at different emission frequencies are used to construct the absorbance. This includes preprocessing of signals on a control FPGA and additional processing on a PC.

After focusing (by lens LI) the combined beams are examined to ensure the collinear propagation. The corresponding beam-profiles are measured at successive distances following LI to ensure proper collimation conditions are met for all three lasers. A liquid sample holder with a quartz window is used to hold the reference liquid (R) and the liquid samples under test (T) (Figure 15). Reference sample holder (R) is filled with Tris-EDTA(TE) buffer as the reference material, with the other containing the liquids being assayed. Measurements of both the reference and the liquid being assayed are taken in rapid succession thus removing the environmental effects. When carried out at all design frequencies, this provides the required information contained in the THz spectrum of the DNA sample.

The LFI imaging system is then calibrated for the purpose of extracting the complex refractive index of the target accurately at multiple THz frequencies. Subsequently, a set of test targets representative of biological samples may be used to evaluate the practical performance of the system, including milli-Q purified water, TE buffer, and typical amino acids that have strong absorption at THz band (such as tyrosine and leucine) in solution. The liquid samples may be prepared at different sample concentrations to enable investigation into the dependence of the self-mixing fringes on the concentration of the solutions. The THz spectrum of the reference samples is compared with the results obtained by other techniques, such as Synchrotron radiation, Raman spectroscopy, Atomic Force Microscope (AFM), Scattering-Nearfield Optical Microscope (SNOM) and nano-Fourier-transform infrared spectroscopy (FTIR). Practical effects of environmental changes (humidity, air temperature gradients), loss of coherence effects on maximum range, and capacity to operate in the presence of the partially obstructing interfaces between the THz QCL and the target can also been taken into account during set up of the LFI imaging system.

Analytical validation of assay performance (measuring sensitivity, specificity, and reproducibility) using reference melanoma cell lines with known somatic mutations is preferably performed. For example, at least 5 cell lines per mutation group (e.g. BRAF, NRAS, NF1, or triple wild type) may be selected to determine the spectral range produced. The chosen cell lines represent a wide range of melanoma and harbour different examples of mutational burden and copy number alterations. Furthermore, these cancer cell lines can be mixed to simulate low allele frequencies, and thereby used to evaluate the sensitivity of the proposed diagnostic assay. Normal skin cells, e.g. melanocytes derived from non-sun-exposed and sun-exposed body sites, naevus cells, as well as keratinocytes, and fibroblasts, may also be similarly assessed.

The assay is then validated using consenting retrospective patient samples from a range of different skin cancer subtypes obtained from laboratories. The performance of the assay can then be evaluated using a range of difference sample types, including fresh, frozen and FFPE-fixed samples. Whole genome sequencing and targeted panel sequencing is conducted to identify the gene mutations contained in the patient samples.

For the fresh samples, a less invasive skin microbiopsy device may be used to collect skin sample and extract DNA, for THz spectroscopic scanning. The microbiopsy is relatively painless and leaves a tiny puncture site in the skin that heals in days. It also allows clinicians to analyse the molecular profde of a potential skin cancer. The less invasive microbiopsy device would also allow dermatologists to better monitor the progression of suspected skin cancers and other skin conditions overtime, without the need for more invasive conventional biopsies.

The THz absorption measured at multiple THz frequencies for a range of skin pathologies including healthy skin, benign nevi, dysplastic nevi, melanoma in-situ, and thin melanoma will be normalized by the spectrum of the pure TE buffer. The normalized spectrum among different stages of skin cancer samples will then be compared in terms of absorption fingerprint frequencies, spectrum slope, and two- component partial least squares discriminant analysis for separation of skin pathergies from different groups. The multi-dimensional dataset acquired in the imaging process also lends itself naturally to more advanced analysis and information extraction, including in particular machine learning and artificial intelligence method, provided sufficient number of training samples were available. The balance between male and female clinical participants will also be considered for sample selection.

Figure 16 is a flowchart of a method for pre-determining wavelengths to be used, for example by one or more sources of radiation such as semiconductor lasers, to elicit an optical response from a target that is characteristic of a particular biomarker of interest. The optical response may be a response such as absorbance or reflectance of the target and in either case it may be a complex response that preserves phase information or a scalar value such as magnitude or the real component of the complex response. At box 100 a number of training targets are prepared, for example in the manner that has previously been described where some targets comprise one or more biomarkers such as DNA extracts of cancerous tissue, for example melanoma and where other targets contain one or more biomarkers such as DNA extracts of healthy or benign tissue. At box 102 a first target is selected as a “current target” and then, at box 104, an optical response of the target to laser radiation over a range of wavelengths, e.g. from 2THz to 6THz is made. For example, the optical response may be measured with a synchrotron as discussed with reference to Figure 12. The optical response of further targets, labelled as either bearing the biomarker or not bearing the biomarker are then made via boxes 106, 105 and 104 until all of the targets have been processed and their optical responses measured. The optical responses that are measured may be absorbance by the target of the radiation as a function of the wavelength at which the radiation is applied and/or phase shift imparted by the target as a function of the wavelength.

At box 108 a programmed personal computer (PC) of Figure 12 (or other suitable processing assembly) processes the measured optical response for each target to take into account the target medium, as previously discussed with reference to Equation 1 and then scale the measurements as previously discussed with reference to Equation 2 to produce absorbance curves for DNA extracts of healthy tissue

At box 112 the processing assembly applies a discriminant analysis procedure to the measured optical responses, for example PLSDA, to determine a plurality of wavelengths, e.g. “pre-determined wavelengths” Xi, X2, and X3 as illustrated in Figure 17, that can subsequently be used by one or more radiation sources, such as the array of THz QCLs LFI of Figure 15 for classifying test targets.

Figure 18 is a flowchart of a method for classifying a test target as containing a particular biomarker of interest.

At box 122 a LFI with an array of THz QCLs, for example as discussed with reference to Figure 15, is prepared with the three lasers of the array set to operate at the predetermined wavelengths for the biomarker of interest that have been arrived at through the procedure of Figure 17. In other embodiments other radiation sources, apart from lasers may be used. Furthermore, a single radiation source that is sequentially operated at the different wavelengths may also be used rather than a plurality of laser sources. Other embodiments may also be implemented, for example a time domain spectroscopic (TDS) source / detector-based system, where an antenna is used as a source of THz radiation and is driven from an ultrashort-pulse laser. Furthermore, in other embodiments, IR lasers may be used, optical pumping of gas cells may be used to produce high power narrow band THz radiation, molecular THz lasers and gas lasers may also be used.

At box 124 the LFI is operated with each of the array of THz QCLs activated in turn to apply laser radiation at their corresponding predetermined wavelength to the target.

At box 126 the optical responses at each of the predetermined wavelengths are detected by the LFI. Background effect subtraction and scaling is then performed at boxes 128 and 130, e.g. by the personal computer of Figure 15, suitably programmed, or other suitable processing assembly.

At box 132 the processing assembly classifies the target as having a biomarker present based on the values detected by the LFI at each of the predetermined wavelengths. For example, in the case of the biomarker being DNA extracts that are associated with melanoma then the biomarker is deemed present if (with reference to Figure 17), reflection amplitude at Xi is less than reflection amplitude at A. and reflection amplitude at Z2 is less than reflection amplitude at X3.

The originating tissue from which the target under test was made can then be appropriately labelled as “biomarker present” or “biomarker absent” at boxes 136 and 138 respectively.

Although methods and systems described herein have made use of an exemplary one or more biomarkers in the form of biomarkers that indicate the presence of DNA extracts of melanoma, the methods and systems are appropriate for detecting other biomarkers, such as one or more biomarkers that indicate a mutation in a gene for example. A gene mutation that is transcribed into the RNA may be translated into a protein that folds incorrectly. Accordingly, it is useful to be able to detect one or more biomarkers for gene mutation in a target such as a tissue extract.

Furthermore, it is believed that methods and systems described herein may be used to detect the presence of other biomarkers that are commonly indicative of malignancies. Such biomarkers include alterations of DNA CpG methylation (i.e. hyper or hypo methylation) and increases and/or decreases in in DNA copy number for whole chromosome arms or regional or focal areas.”

Whilst preferred embodiments have been discussed in the context of human samples such as human skin, biomarkers in other types of tissues, such as plant material, may also be detected.

More generally, it will be realized that a method for classifying a test target as containing one or more biomarkers has been disclosed. The method involves making a number of targets in which the one or more biomarkers are absent (“non-biomarker targets”) and also making a number of targets in which the one or more biomarkers are present (“biomarker targets”). Optical responses of each of the non-biomarker targets (“non-biomarker responses”) and of each of the biomarker targets (“biomarker responses”) to radiation over a range of wavelengths can then be made, for example using the synchrotron interferometer set up that is discussed with reference to Figure 12.

The non-biomarker responses may differ from the biomarker responses, for example akin to the different responses of the targets made from normal and melanoma DNA extracts of Figure 2 and Figure 3. In that case then it will be possible to process the biomarker responses and the non-biomarker responses to determine a number of different wavelengths (“pre-determined wavelengths”) of radiation for subsequently eliciting an optical response from a test target to discriminate between the test target being either a biomarker target or a non-biomarker target.

In compliance with the statute, the present disclosure has used language more or less specific to structural or methodical features. The term “comprises” and its variations, such as “comprising” and “comprised of’ is used throughout in an inclusive sense and not to the exclusion of any additional features. It is to be understood that the claimed subject matter is not limited to specific features shown or described since the means herein described comprises preferred embodiments. The disclosed subject matter is, therefore, claimed in any of its forms or modifications within the proper scope of the appended claims appropriately interpreted by those skilled in the art. Throughout the specification and claims (if present), unless the context requires otherwise, the term "substantially" or "about" will be understood to not be limited to the value for the range qualified by the terms.

Throughout the description and the claims of this specification, the singular encompasses the plural unless the context otherwise requires. In particular, where the indefinite article is used, the specification is to be understood as contemplating plurality as well as singularity unless the context requires otherwise.

Features, integers, characteristics, compounds, chemical moieties or groups described in conjunction with a particular aspect, embodiment or example of the disclosed subject matter are to be understood to be applicable to any other aspect, embodiment or example described herein unless incompatible therewith.

Any references to methods, apparatus or documents of the prior art are not to be taken as constituting any evidence or admission that they formed, or form part of the common general knowledge.

Any embodiment of the disclosed subject matter is meant to be illustrative only and is not meant to be limiting. Therefore, it should be appreciated that various other changes and modifications can be made to any embodiment described herein.

Claims

Claims:

1. A method for classifying a target comprising tissue or an extract thereof as having one or more biomarkers present, the method comprising: for each one of a plurality of wavelengths, the wavelengths being predetermined to elicit an optical response from the target that is characteristic of the one or more biomarkers, applying radiation from one or more radiation sources at the wavelength to the target; and detecting an optical response of the target at the wavelength, to thereby detect optical responses of the target at each of the wavelengths; and classifying the target as having the biomarker present based on the detected optical responses of the target.

2. The method of claim 1, wherein the one or more radiation sources comprises one or more Quantum Cascade Lasers (QCL) and the radiation comprises laser radiation.

3. The method of claim 2, wherein the detecting of the optical response of the target at each of the plurality of wavelengths includes self-mixing of the laser radiation applied to the target with laser radiation returned from the target.

4. The method of any one of claims 1 to 3, wherein each of the one or more radiation sources is configured to operate in a frequency range between 2 THz and 6 THz.

5. The method of any one of claims 1 to 4, wherein the one or more radiation sources comprise a plurality of radiation sources, each configured to emit radiation at a corresponding one of the plurality of wavelengths.

6. The method of any one of claims 1 to 5, wherein the target includes a medium for containing the extract of the tissue.

7. The method of claim 6, wherein the detected optical response is processed to account for a background optical response including an optical response of the medium.

8. The method of any one of the preceding claims, wherein the one or more biomarkers comprise DNA extracts associated with cancerous cells.

9. The method of claim 8, wherein the one or more biomarkers comprise DNA extracts associated with cancerous cells being melanoma.

10. The method of claim 9, including: applying radiation to the target at first (Xi), second (X2), and third (X3) wavelengths of the plurality of wavelengths, wherein Xi > X2 > X3 and the detected optical responses comprise, reflectance and/or reflection phase shift of the target; or absorbance and/or transmission phase shift, at each of Xi, X2, and X3.

11. The method of claim 10, wherein the detected optical responses comprise absorbance and wherein the classifying of the target as having a biomarker present that is associated with melanoma is made if: absorbance at Xi is less than absorbance at X2; and absorbance at X2 is less than absorbance at X3.

12. The method of claim 10 or claim 11, wherein X2 corresponds to a wavelength at or adjacent to which a scaled absorbance curve of DNA extracts of healthy tissue crosses a scaled absorbance curve of DNA extracts of melanoma tissue.

13. The method of any one of claims 1 to 12, including determining the pre-determined wavelengths that elicit the optical response from the target that is characteristic of the one or more biomarkers by: applying radiation over a range of wavelengths to one or more targets known to contain the one or biomarkers and known not to contain the one or more biomarkers to thereby obtain one or more biomarker optical response curves characteristic of presence or of absence of the one or more biomarkers.

14. The method of claim 13 , including processing the one or more biomarker optical response curves to determine the pre-determined wavelengths of the one or more radiation sources that elicit the optical response that is characteristic of the one or more biomarkers.

15. The method of claim 14, wherein the processing of the one or more biomarker optical response curves determines the pre-determined wavelengths that create optical responses for optimal discrimination between targets in which the one or more biomarkers are present and targets in which the one or more biomarkers are absent.

16. The method of claim 15, wherein the processing of the one or more biomarker optical response curves includes analyzing the optical response curves using Partial Least Squares Discriminant Analysis (PLSDA) to determine the pre-determined wavelengths that facilitate optimal discrimination.

17. The method of any one of claims 13 to 16, wherein the optical response curves are processed to account for a background optical response including an optical response of a medium of the target.

18. The method of claim 17, wherein the optical response curves are scaled to facilitate comparison of optical response of targets containing the one or more biomarkers with optical response of targets not containing the one or more biomarkers.

19. A system for classifying a target comprising tissue or an extract thereof as having one or more biomarkers present, the system comprising: one or more radiation sources, configured to emit radiation at each of a plurality of wavelengths, the wavelengths being predetermined to elicit an optical response from the target that is characteristic of the one or more biomarkers; a detector assembly for detecting the optical response from the target; and a processing assembly responsive to the detector assembly and configured to: classify the target as having the one or more biomarkers present based on the detected optical responses.

20. The system of claim 19, wherein the one or more radiation sources comprise a plurality of radiation sources.

21. The system of claim 19, wherein the one or more radiation sources comprise one or more THz QCLs and wherein the detector assembly is responsive to self-mixing electrical signals of the one or more THz QCLs.

22. The system of any one of claims 19 to 21, wherein each of the one or more radiation sources is configured to operate at a frequency between 2 THz and 6 THz.

23. The system of any one of claims 19 to 22, wherein the processing assembly is configured to subtract a background optical response effect of a medium of the target from the optical responses of the target to the radiation at each of the wavelengths.

24. The system of any one of claims 19 to 23, wherein the one or more biomarkers comprise DNA extracts of melanoma and wherein the one or more radiation sources operate at wavelengths Xi > X2 > X3 respectively and the detected optical responses comprises absorbance of the target, at each of Xi, X2, andX ; wherein the processing assembly is configured to classify the target as having one or more biomarkers present that are associated with melanoma if: absorbance at Xi is less than absorbance at X2; and absorbance at X2 is less than absorbance at X3.

25. A method for classifying a test target as containing one or more biomarkers, the method comprising: making a number of targets in which the one or more biomarkers are absent (“non-biomarker targets”); making a number of targets in which the one or more biomarkers are present (“biomarker targets”); recording optical responses of each of the non-biomarker targets (“nonbiomarker responses”) to radiation over a range of wavelengths; recording optical responses of each of the biomarker targets (“biomarker responses”) to radiation over a range of wavelengths; and where the non-biomarker responses differ from the biomarker responses then processing the bio-marker responses and the non-biomarker responses to determine a number of different wavelengths (“pre-determined wavelengths”) of radiation for subsequently eliciting an optical response from a test target to discriminate between the test target being either a biomarker target or a non-biomarker target.

26. The method of claim 25, wherein the processing of the non-biomarker responses and the biomarker responses includes Partial Least Squares Discriminant Analysis (PLSDA) to determine the pre-determined wavelengths that facilitate optimal discrimination.

27. The method of claim 25 or claim 26, wherein the recording of the non-biomarker responses and the biomarker responses are made using a wideband laser source, such as a synchrotron.

28. The method of any one of claims 25 to 27, wherein the pre-determined wavelengths are applied to the test target using one or more narrow band lasers, such as Quantum Cascade Lasers.