Systems and methods for intraoperative lifetime imaging - Patents.com

Abstract

A method and system for evaluating tissue to determine the presence or absence of cancer cells, the method including obtaining fluorescence lifetime (FLT) data from the tissue and processing the FLT data to identify an FLT signal at each of a plurality of sites in the tissue, the method further including identifying FLT data at any of the plurality of sites that is above a threshold indicative of the presence of cancer cells, and generating a report indicating that any of the plurality of sites is above the threshold indicative of the presence of cancer cells.

Description

[Technical field]

[0001] This application is based on and claims priority to U.S. Provisional Application Nos. 63 / 272,847 (filed October 28, 2021) and 63 / 366,483 (filed June 16, 2022), the entire contents of which are incorporated herein by reference.

[0002] <Statement Regarding Federally Sponsored Research> none. [Background technology]

[0003] The present disclosure relates generally to systems and methods for intraoperative assessment of tissue, and more particularly, to systems and methods for intraoperative inspection of tissue and identification of target cells in vivo for further guidance of surgical procedures.

[0004] Despite great advances in surgical oncology, the positive margin rate for cancer resection remains high. The presence of tumor-positive margins during resection is associated with poor outcomes. Surgeons use palpation, visual inspection, or standard scales to identify the tumor-normal boundary during surgery. These methods cannot guarantee complete resection, leading to local recurrence and poor overall survival. Post-resection margin assessment is performed by visual inspection of macroscopic specimens or plain radiographs, or by histopathology, but these methods capture the margin status of only a portion of the specimen and may miss viable cancer. There is a need for imaging techniques with sufficient sensitivity to detect any cancer remaining in the surgical bed and to postoperatively assess the margin status of the entire specimen, thereby ensuring complete tumor removal, reducing cancer recurrence, and improving overall survival and quality of life for patients.

[0005] With this in mind, various efforts have been made to assess the resection site in vivo by using imaging techniques. In light of this, fluorescence imaging techniques have shown promise for margin guidance using fluorescent probes, which are attractive due to their low cost, high sensitivity, relatively simple equipment, and the use of non-ionizing radiation. Thus, in the past decades, many fluorescent contrast agents and fluorescence imaging systems have been developed for cancer diagnosis, image-guided surgery, and drug development.

[0006] Most of these imaging applications use dyes and take advantage of the knowledge that they have a higher uptake rate in cancer cells than in healthy cells. However, background fluorescence from normal tissues and nonspecific probe uptake reduce tumor contrast, resulting in a lack of sensitivity and specificity. Conventional imaging systems detect total fluorescence intensity, which cannot easily distinguish the fluorescence signal resulting from tumor-bound probes from nonspecific fluorescence. Furthermore, the fluorescence intensity depends on the probe uptake, which in turn depends on the size of the tumor, so the smaller the tumor, the more difficult it is to detect against the nonspecific background.

[0007] Thus, some of the above contrast agents are tumor-targeting agents. For example, using therapeutic antibodies against receptors overexpressed in cancers is a powerful approach to target tumors, provided that the antibodies are more likely to be retained by cancer cells, are relatively uncomplicated, and have low manufacturing costs. One of the above contrast agents that shows great promise is panitumumab-IRDye800CW, a conjugate of panitumumab, an FDA-approved therapeutic antibody against epidermal growth factor receptor (EGFR), and IRDye800CW, an NIR dye that has been tested in multiple human clinical trials. EGFR is overexpressed in multiple cancers, including head and neck, lung, glioma, and metastatic colorectal cancer (mCRC), and is therefore a promising target for fluorescence imaging. Recent clinical trials have shown that panitumumab-IRDye800CW is safe for use in humans, can improve tumor contrast during fluorescence-guided resection surgery, and can distinguish benign lymph nodes from lymph node metastases in patients with head and neck squamous cell carcinoma (HNSCC).

[0008] Thus, despite many advances in identifying novel cancer-specific molecular targets and imaging probes, no contrast agent has been widely adopted for clinical use. This is mainly due to poor pharmacokinetics and relatively low tumor uptake. Conventional fluorescence imaging systems detect the total fluorescence intensity re-emitted from the sample. Because the fluorescence intensity depends on the product of the probe concentration and the fluorescence lifetime, tumor-specific fluorescence cannot be easily distinguished from nonspecific dye accumulation of the probe in healthy tissue. In addition, the fluorescence intensity is affected by the tumor size and the probe uptake, making it difficult to detect small lesions at the surgical site with a sufficient signal-to-background ratio. In addition, the fluorescence intensity is strongly affected by tissue attenuation and system-specific measurement parameters, including the power of the illumination light, the sensitivity and response characteristics of the detector or camera, and spurious leakage of ambient light. As a result, it is not easy to compare the fluorescence intensity measurements on an absolute scale between multiple specimens, subjects, and imaging systems, which hinders standardization and adoptability.

[0009] Despite great efforts in the development of novel imaging agents and many promising clinical trials over more than 30 years, specific labeling of cancer cells using exogenous agents has not yet been demonstrated in humans. The main problem remains the accumulation of nonspecific probes in normal or benign tissues, which significantly reduces the relative brightness of tumors to the background, leading to poor signal-to-noise ratios and low specificity (false positives) and sensitivity (false negatives).

[0010] Attempts to overcome these problems include the combination of specially designed hardware with specific imaging agents / dyes (with targeted or non-targeted tracers). Special imaging techniques have also been investigated to overcome the lack of target specificity of molecular imaging agents, including co-loading dose administration to improve probe uptake, drug pair imaging techniques in which targeted and non-targeted probes are combined, or reference schemes. However, the above techniques have not shown sufficient robustness for broad clinical use or require the administration of multiple drugs, making their clinical translation difficult. Thus, the clinical implementation of the above approaches faces significant financial and regulatory barriers, and good and consistent results still need to be established.

[0011] It would therefore be desirable to provide a system and method for in vivo tissue analysis capable of analyzing very large amounts of tissue with sufficient accuracy to provide the clinical certainty of a histopathological diagnosis. Summary of the Invention

[0012] The present disclosure overcomes the above deficiencies by providing a system and method for intraoperative tissue assessment that does not require novel dyes or special tracers, or the use of special hardware and combinations of special dyes or tracers. The present disclosure provides a system and method for intraoperative tissue assessment, such as resection sites and margins, using fluorescence lifetime (FLT) imaging. That is, the present disclosure recognizes that FLT is longer in cancer cells compared to non-specific dyes in normal tissue. Thus, the present disclosure provides a system and method that can dramatically differentiate the contrast between cancer and normal tissue with higher sensitivity and specificity than conventional dyes and imaging systems that rely on tumor uptake. The present disclosure provides a system and method that can assess FLT in absolute units (nanoseconds) that are system independent and not affected by light-tissue interactions such as scattering and absorption. Thus, the present disclosure provides a system and method that can facilitate robust standardization in intraoperative in vivo tissue assessment.

[0013] In one aspect of the disclosure, a method is provided for evaluating tissue to determine the presence or absence of cancer cells, the method including obtaining fluorescence lifetime (FLT) data from tissue and processing the FLT data to identify an FLT signal at each of a plurality of sites in the tissue, the method further including identifying FLT data at any of the plurality of sites that exceeds a threshold indicative of the presence of cancer cells, and generating a report indicating that any of the plurality of sites exceeds the threshold indicative of the presence of cancer cells.

[0014] Another aspect of the present disclosure provides a medical imaging system, comprising: a light source configured to transmit light to tissue; and a detector configured to receive light fluorescently emitted by the tissue and generate fluorescence lifetime (FLT) data. A processor is configured to analyze the FLT data to determine the presence or absence of cancer in the tissue and generate a report indicating the spatial location of any cancer determined to be present in the tissue. The system further comprises a display configured to display the report to guide a surgical procedure to remove the cancer.

[0015] These and other advantages of the present invention will become apparent from the following detailed description, in which reference is made to the accompanying drawings, in which:

[0016] Embodiments of the present invention will become more apparent from the following detailed description when taken in conjunction with the drawings, in which like elements are numbered like, and in which: [Brief description of the drawings]

[0017] [Figure 1] FIG. 1 illustrates a non-limiting example system of the present disclosure. [Diagram 2] 2 is a flow chart illustrating some non-limiting example steps of a process of the present disclosure that may use a system such as that shown in FIG. 1. [Figure 3A-D] Figure 3A is a graph showing the derivation of the nonlinear fitting of the present disclosure. Figure 3B is a graph showing the quasi-time domain (QTD) of the two same lifetimes. Figure 3C is a graph of SNR for a range of lifetimes and noise levels using standard time domain (TD) data. Figure 3D is a graph of SNR at varying lifetimes and noise levels using QTD data, showing a significant improvement in SNR compared to standard TD data over the entire range of simulated lifetimes and noise levels. [Figure 4A-C]Figure 4A is a fluorescence lifetime microscopy (FLIM) image of a thin tissue slice from a skin cancer patient who received ICG 24 hours prior to surgery. Figure 4B is a standard histological (hematoxylin-eosin) stained image of the sample in Figure 4A. Figure 4C is a box plot showing the lifetimes of various tissues within the tissue section in Figure 4B and the cutoff lifetime level (dashed line) above which the tissue is deemed to have cancer. [Figure 5A-B] Figure 5A is a box plot showing the fluorescence lifetimes of hepatocellular carcinoma (HCC) tissues and normal tissues excised from patients administered ICG, and the cutoff lifetime levels (dashed lines) at which tissues are judged to have cancer. Figure 5B is a box plot showing the fluorescence lifetimes of oral SCC tissues and normal tissues excised from patients administered ICG, and the cutoff lifetime levels (dashed lines) at which tissues are judged to have cancer. [Figure 6A-D] Figure 6A is a color photograph of tissue including cancerous tissue and normal tissue excised from a patient with skin SCC. Figure 6B is an image showing the excised tissue of Figure 6A with a lifetime data overlay. Figure 6C is an image showing the excised tissue of Figure 6A with a mask. Figure 6D is an image showing the excised tissue of Figure 6A with a fluorescence intensity data overlay. [Figure 7A-D] Figure 7A is an image of a head and neck specimen resected from an oral SCC patient. Figure 7B is a microscopic FLT image of a small section within a positive and negative lymph node (LN), showing significantly longer lifetimes in the positive LNs. Figure 7C is a widefield fluorescence intensity image of the specimens of Figures 7A and 7B. Figure 7D is a widefield fluorescence lifetime image suggesting that the FLT values ​​in the positive LNs of Figure 7A are significantly and uniformly longer than those of the negative LNs, in strong contrast to what the intensity image of Figure 7C shows. [Figure 8A-D]FIG. 8A is a graph of the time domain (TD) fluorescence signal measured from a liver HCC specimen immediately after resection from a patient who received systemic ICG 24-48 hours prior to surgery. FIG. 8B is a graph of the time domain (TD) fluorescence signal measured from an oral SCC cancer surgical specimen immediately after resection from a patient who received systemic ICG 24-48 hours prior to surgery. FIG. 8C is a histogram showing the fluorescence intensity data in tumor tissue (red) and normal tissue (green). FIG. 8D is a histogram showing the FLT data in tumor tissue (red) and normal tissue (green). [Figure 9] Box plot showing the distribution of FLT in multiple tumor types (HCC, mCRC, and OSCC) compared with the FLT of ICG in various normal tissues. [Figure 10A-B] Figure 10A is a graph showing representative TD fluorescence attenuation curves of panitumumab-IRDye800CW (gray solid line), IgG-IRDye800CW (black solid line), and PBS (gray dashed line) in cancer cells. Figure 10B is a graph showing representative TD fluorescence attenuation curves of panitumumab-IRDye800CW (gray solid line) and IgG-IRDye800CW (black solid line) in culture medium, and a representative TD fluorescence attenuation curve of a stock solution of panitumumab-IRDye800CW in PBS (gray dotted line) (gray dotted line). [Figure 10C-D] Figure 10C is a set of confocal microscopy images of fluorescence intensity and FLT of cancer cells after incubation with panitumumab-IRDye800CW (100 μg), IgG-IRDye800CW (100 μg) or PBS for 2 h at 370° C. Figure 10D is a set of widefield FLT maps of media collected after incubation of imaging probes with cancer cells and panitumumab-IRDye800CW in PBS. [Figure 11A-C]Figure 11A is a set of high-resolution images including FLIM, IHC, and H&E stained images, showing increased FLT in tumor regions with high EGFR expression. Figure 11B is a set of images including FLIM, IHC, and H&E stained images, showing a magnification of the region shown in Figure 11A. Figure 11C is a set of images including FLIM, IHC, and H&E stained images, showing a magnification of the region shown in Figure 11B. [Figure 12A-C] Figure 12A is a set of representative confocal fluorescence intensity and FLIM images taken at low magnification from a clinical specimen, along with corresponding H&E and EGFR IHC images. Figure 12B is a set of images from Figure 12A at higher magnification. Figure 12C is another set of images from Figure 12A at higher magnification. [Figure 13A-C] Figure 13A shows IHC images of EGFR expression (% area positive for EGFR) in muscle, salivary gland, and tumor ROIs in ascending order of expression. Figure 13B shows a set of confocal fluorescence intensity images and intensity histograms corresponding to Figure 13A. Figure 13C shows a set of confocal FIM images and FLT histograms of the same ROIs as Figure 13A, showing that FLT values ​​tend to increase with increasing EGFR expression. [Fig. 13D-G] Figure 13D is a scatter plot of mean fluorescence intensity versus percent area positive for EGFR by IHC in all ROIs imaged. Figure 13E is a graph showing mean fluorescence intensity in EGFR negative and positive pixels from co-registered IHC and FLIM images. Figure 13F is a scatter plot of mean FLT versus percent area positive for EGFR in all ROIs. Figure 13G is a graph of mean FLT in EGFR negative and positive pixels in the same ROIs. Bars are plotted as mean with standard deviation. [Figure 14A-D] Figure 14A is a photograph of a tissue sample. Figure 14B is an image of the H&E staining of the tissue sample of Figure 14A. Figure 14C is an intensity image of the tissue sample of Figure 14A. Figure 14D is an image showing the Panitumumab-IRDye800CW amplitude in the tissue sample of Figure 14A. [Fig. 14E-G] Figure 14E is an image showing tissue autofluorescence amplitude in the tissue sample of Figure 14A. Figure 14F is a widefield FLT image of the specimen showing the tumor border (dotted line) of the co-registered H&E image shown in Figure 14B. Figure 14G is a FLIM image of the rectangular region shown in Figures 14C-14F. [Fig. 14H-I] Figure 14H is a graph of the distribution of fluorescence intensity data obtained from the sample.Figure 14I is a graph of the distribution of spectral separation data obtained from the sample. [Fig. 14J-K] Figure 14J is a graph of the distribution of fluorescence lifetime data obtained from the samples, and Figure 14K is a set of ROC curves for tumor-normal tissue classification using FLT (black solid line), fluorescence intensity (grey dashed line), and spectral separation (grey solid line) based on H&E ground truth. [Figure 15A-D] Figure 15A is a graph of the distribution of fluorescence intensity of sarcoma and normal tissues. Figure 15B is a graph of the distribution of FLT of sarcoma and various normal tissues. Figure 15C is a graph of the average fluorescence intensity of sarcoma and various normal tissues from seven different patients. Figure 15D is a graph of the average FLT of sarcoma and normal tissues from seven different patients. [Figure 16A-D] Figure 16A is a graph showing the distribution of fluorescence intensity in oral cancer and normal oral tissues of eight patients. Figure 16B is a graph showing the distribution of FLT data in oral cancer and normal oral tissues of eight patients. Figure 16C is a graph of the average tumor and normal intensity of six HN cancer patients. Figure 16D is a graph of the average tumor and normal FLT data of six HN cancer patients. [Figure 17A-B] Figure 17A shows sensitivity of FLT and intensity to false positive rate based on tumor and normal classification of 10 sarcoma patients, and Figure 17B shows sensitivity of FLT and intensity to false positive rate based on tumor and normal classification of 8 head and neck cancer patients. [Figure 18A-H]FIG. 18A shows an example of a fluorescence intensity image of a microorganism expressing five iRFP variants (iRFP670, 682, 702, 713, and 720). FIG. 18B shows an example of a fluorescence lifetime image of a microorganism expressing five iRFP variants (iRFP670, 682, 702, 713, and 720). FIG. 18C shows a graph of normalized excitation spectra of three iRFP variants. FIG. 18D shows a graph of emission spectra of iRFP670, 702, and 720. FIG. 18E shows a time domain (TD) fluorescence signal graph of iRFP670, 702, and 720 in a microorganism. FIG. 18F shows a lifetime histogram derived from FIG. 18B. FIG. 18G shows an example of a fluorescence intensity image of three MTLn3 tumors expressing iRFP670, 702, and 720 in the mammary fat pad of a female nude mouse. FIG. 18H shows example fluorescence lifetime images of three MTLn3 tumors expressing iRFP670, 702, and 720 in the mammary fat pad of the female nude mouse described above. [Figure 19A-D] Figure 19A is a fluorescence intensity reflectance image of a mouse administered a fluorescent dye targeted to bone and kidney. Figure 19B shows the attenuation amplitude obtained from bi-exponential fitting of time domain data with fixed lifetimes of 0.5 ns (green) and 0.65 ns (red). Figure 19C shows that 3D reconstruction using the asymptotic time domain (ATD) method clearly separates the skeleton and identifies the kidney. Figure 19D shows an in situ amplitude map (without skin). DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018] The present disclosure recognizes that the fluorescence lifetime (FLT) of a fluorescent dye in tumor cells is significantly longer than the FLT of the same dye in healthy tissue. This disclosure builds upon this by providing systems and methods for intraoperative tissue analysis that can distinguish tumors from healthy tissue with a consistency and specificity not achieved by previous attempts. The present disclosure thus recognizes that the FLT, measurable in absolute units (typically nanoseconds), is a parameter that is robust to measurement conditions and can be used to mitigate many of the shortcomings of previous efforts in creating robust intraoperative tissue assessment tools.

[0019] The systems and methods provided herein are flexible. They do not require a specific combination of dyes or targeting agents and hardware. In one non-limiting example, fluorescently labeled EGFR antibodies can be used, although other dyes or fluorescent labeling mechanisms can be used. In one non-limiting aspect of the present disclosure, FLT imaging can use the near infrared (NIR) spectrum, although other wavelengths can be used. The systems and methods provided herein can provide a dramatic increase in specificity and sensitivity compared to standard fluorescence intensity-based methods for distinguishing tumors from normal tissues in situ and in vivo.

[0020] 1, a non-limiting example system 100 that may be used in the present disclosure is shown. In one non-limiting example, the system 100 may be a time-domain (TD) imaging platform configured for both in vivo and ex vivo or in vitro imaging. In particular, the non-limiting system 100 includes an in vivo imaging subsystem 102 and an in vitro or ex vivo imaging subsystem 104.

[0021] In the illustrated configuration, the in-vivo imaging subsystem 102 is configured to transmit light from the operating table 106 to a fiber optic bundle 108, which is collected by a relay lens (RL) and split by a dichroic mirror (D1) to one or more cameras (RGB). In one non-limiting example, the image can be collected directly by a camera, which can be configured for wavelengths below a threshold, e.g., 650 nm. In this case, an intensifier camera (CCD / intensifier) ​​can be included that is configured for wavelengths above a threshold, e.g., 650 nm. A second non-time-gated camera can also be included that collects intensity images in parallel in real time. Alternatively, intensity data can be obtained by cumulatively adding time-gated data using only one camera.

[0022] The intensifier / CCD can be fitted with a mirror housing (M) that can be remotely switched between receiving light from the fiber bundle 108 and from the specimen stage. Fluorescence or NIR excitation light can be collected, for example using a filter wheel (F) attached to the ICCD. A fiber can send light (e.g., 780 nm light) through a dichroic (D2) to a digital light projector (DLP) (e.g., 800 nm) for specimen illumination, as well as light (e.g., 780 nm light) through a port in the objective lens (B) to the operating table 106. These wavelengths are merely examples, and other wavelengths can be used. In particular, the near infrared (NIR) spectrum can be used, as described below. In the non-limiting example of NIR light, this light can penetrate up to 5-10 cm into tissue, which can be advantageous for evaluating tumors that are several centimeters thick under tissue layers.

[0023] As described below, system 100 can be configured to accommodate a wide field of view (FOV) while providing micron resolution. The fiber bundle is mounted on a flexible, articulated arm (A) that is attached to a portable stand (C). This arm A can be positioned to correspond to a desired viewing plane of a surgical table 106. In this manner, an in-vivo probe is provided that can be manually operated or handheld for maneuvering around a surgical site.

[0024] As shown, the system 100 can be integrated with a cart or rack 110, which can include an in vitro or ex vivo imaging subsystem 104. The in vitro or ex vivo imaging subsystem 104 can be controlled by a stepper motor driver, which can control the positioning of the sample chamber 112. The system 100 can also include a laser diode driver, a PS delay unit, and an HRI controller configured to coordinate the delivery of the laser illumination, as shown. The system 100 can also include a computer system or processor configured to perform the data acquisition, data processing, and report generation functions of the present disclosure.

[0025] Referring to FIG. 2, the disclosed process 200, which may be used for in vivo or ex vivo analysis using the system 100 as described with reference to FIG. 1, may begin at process block 202 with acquisition of either "sequential" or "accumulated" data. As described below, the data may be acquired as a time domain (TD) accumulation or expressed in other manner to constitute quasi-time domain data (QTD). Additionally, conventional intensity data may be acquired. As described below, this intensity data may be used in conjunction with the lifetime data and / or the intensity data may be reported in addition to the lifetime data or the intensity data and lifetime data may be combined. In this manner, lifetime data may be acquired, and optionally intensity data may be acquired.

[0026] In sequential TD acquisition, time-resolved (or time series) fluorescence data is collected for multiple time points with a fixed "gate width" (or acquisition window / exposure) and fitted to the following exponential decay:

number

[0027] where y TD are the data as a function of time "t", a0 is the attenuation amplitude related to the fluorophore concentration, quantum yield and other empirical scaling constants, and "τ" is the fluorescence lifetime.

[0028] However, in one aspect of the present disclosure, TD fluorescence data is acquired cumulatively from a selected time origin to an end time point or at multiple time points, thereby forming a variable gate width / time window / exposure, which can be mathematically represented by the following equation:

number

[0029] Here, QTD stands for "quasi-time domain" acquisition. Using equations (1) and (2), we can obtain:

number

[0030] In this way, the QTD data -t / τ Instead, τ(1-e -t / τ) which is fitted to the τ1 = 0.5 ns and τ2 = 1 ns lifetimes. While the two techniques are mathematically equivalent, an important empirical difference is that the QTD technique can provide a significantly improved signal-to-noise ratio (SNR) compared to the standard TD technique. That is, the small count signals at later time points are cumulatively combined with all the higher count signals earlier, resulting in a higher overall signal count and a higher SNR at all time points using the QTD technique in a short noise limited system. This principle is illustrated in Figures 3A-3D for two lifetimes, τ1 = 0.5 ns and τ2 = 1 ns. In particular, Figure 3A shows the TD fluorescence attenuation of two different fluorophores with lifetimes τ1 = 0.5 ns (circles) and τ2 = 1 ns (diamonds), where the function e -t / τ The corresponding nonlinear fits to the data using τ(1-e) are also shown (solid and dashed lines). Figure 3B shows the quasi-temporal domain (QTD) of the same two lifetimes as in Figure 3A, with the function τ(1-e -t / τ ) are also shown. The improvement in SNR with QTD is especially clear for the short 0.5 ns lifetime case. Figure 3C shows the SNR for a range of lifetimes and noise levels with standard TD. Finally, Figure 3D shows the SNR for varying lifetimes and noise levels with QTD, demonstrating a significant improvement in SNR compared to standard TD over the entire range of simulated lifetimes and noise levels. The time response of each parameter, indicated by circles and diamonds in Figures 3C and 3D, is that shown in the respective figures in Figures 3A and 3B. The improvement in SNR with QTD allows the use of fewer temporal data points, which improves imaging speed.

[0031] In one aspect of the QTD data obtained, a lifetime map can be calculated directly using the above data and the fluorescence intensity. From equation (3),

number

[0032] Where:

number

number

[0033] SNR is y QTD (∞)≧y QTD (T1)>y TD Therefore, a single gate width and CW data can improve the lifetime map estimation, whereas conventional time domain methods require multiple time delays due to the inherent noise in the data. This single time gate approach can dramatically reduce both the acquisition time and the computation time, thus aiming for real-time lifetime maps.

[0034] In another aspect of QTD data, it can be theoretically proven that the uncertainty in QTD data is lower than that in TD data. Based on equation (4), the relationship between QTD and TD can be written as follows:

number

[0035] The uncertainty in equation (6) can be stated as follows:

number

[0036] 3A and 3B, TD fluorescence data obtained by sequential or cumulative acquisition can be linearized based on a logarithmic or series expansion or decomposition scheme, and the linearized data can be fitted using any linear least-squares based method to determine the lifetime.

[0037] 2, at processing block 204, the lifetime data and any optional intensity data are processed for analysis. In one non-limiting example, the acquired data can be analyzed in a variety of different ways. In one non-limiting example, the analysis can use an exponential function or attenuation amplitude map, where the data is decomposed into an exponential function using the known lifetimes of the tumor and normal tissue based on prior characterization, as described above:

number

[0038] In another non-limiting example, the tumor life span τ tumor and normal lifespan τ normal If is known, the attenuation amplitude a can be calculated from the QTD data using two time gates and CW intensity data. tumor and a normal These calculations are performed as follows:

[0039] Based on equation (3), the QTD data for a particular time gate "T" can be written in bi-exponential form as follows:

number

[0040] term a tumor τ tumor +anormal τ normal is the CW intensity data, i.e., y QTD (∞). Substituting this into equation (8), we get:

number

[0041] The above formula can be expressed as the attenuation amplitude a using a linear matrix expression with two gate widths T1 and T2 as shown below. tumor and a normal can be solved for:

number

[0042] Next, linear fitting is performed on the raw data to obtain the amplitude coefficient a tumor and a normal The above amplitude map a, which corresponds to the tumor life span, is calculated. tumor can be displayed on the sample surface as described in detail below. The advantage of such a linear fit is that the acquisition speed increases dramatically.

[0043] Regardless of the specific analytical process, global processing of the data can be useful to enhance tumor contrast without the need for special chemicals designed to target cancer. Designing chemical probes with cancer specificity has been difficult and unsuccessful to date, since cancer-specific markers are also expressed in normal tissues. On the other hand, the present disclosure can take advantage of the fluorescence lifetime difference between the dye taken up by cancer cells and its lifetime in healthy tissue, allowing for dramatic sensitivity and specificity improvements compared to traditional fluorescence intensity-based detection. In this regard, thresholds that distinguish cancer tissue can be used. For example, an image can be reconstructed that can be analyzed by comparison to one or more thresholds in processing block 206.

[0044] In one non-limiting example, the threshold can be selected to delineate tumors from normal tissues. The threshold can be selected to perform a discrimination process that applies to multiple patients for at least a particular type of cancer. In other words, different FLT thresholds can be used to distinguish between different types of primary and metastatic cancers (e.g., oral cancer, brain tumors, skin cancer, breast cancer, liver cancer, melanoma, and sarcoma, etc.), and FLT thresholds can be used to identify a particular type of cancer for multiple patients, regardless of the measurement system or other variables.

[0045] For example, referring to Figures 4A-4C, FLT threshold selection is shown in relation to skin cancer. Specifically, as shown in Figure 4A, a table of FLTs for multiple tissue types is first generated from fluorescence lifetime microscopy (FLIM) measurements of tissue sections, which are then co-registered with histological images from patients administered a fluorescent dye, as shown in Figure 4B. In this example, the fluorescent dye is ICG. From the microscopic lifetime values, the tumor-specific FLT and the FLTs of various normal tissues can be known, and the microscopic lifetime values ​​can be tabulated in box plots, as shown in Figure 4C, for example. For tumor vs. normal tissue classification, a receiver operating characteristic (ROC, a plot of sensitivity vs. specificity vs. varying threshold) can be calculated. The FLT threshold for tumor vs. normal tissue classification can then be determined as the FLT value that provides the desired accuracy, which can be calculated, for example, as the area under the ROC curve. As another example, Figures 5A and 5B show similar box plots for liver (hepatocellular carcinoma) and (oral squamous cell) cancer, with threshold lifetimes for these cancers. Thus, the present disclosure illustrates the extension of the systems and methods provided herein to a variety of anatomical structures and tumor types.

[0046] Our preliminary data indicates a specific range of threshold life span values, but the above threshold life span values ​​can be updated with further clinical studies and multiple patients administered a particular fluorescent dye. The threshold life span for each cancer type can be determined across a variety of conditions, including but not limited to administration time and dosage, and pretreatment conditions (e.g., radiation therapy, chemotherapy, etc.).

[0047] Referring again to FIG. 2, the results of the analysis performed in process block 206 are used to generate a report in process block 208. In a surgical setting, a tumor / normal boundary may be defined using a cancer specific FLT threshold and included in the report in process block 208. Additionally, the tumor / normal boundary may be superimposed on an anatomical image or projected onto the operating table to guide the surgeon as to where to begin resection and how much tissue to cut. If intensity data is acquired / included, it may be included in the report. In this way, the intensity data may be provided for the clinician's reference or for detailed validation of the report with the life span data. Additionally, the intensity data may be reported to guide the tumor site in real-time or near real-time and / or to guide the collection of life span data and images. Thus, the report generated in process block 208 may take a variety of forms.

[0048] 6A-6D, a non-limiting example of a report including an overlay is shown. Specifically, FIG. 6A shows an autopsy photograph of a specimen from a patient with cutaneous squamous cell carcinoma, who was administered ICG 24 hours prior to surgery. And FIG. 6B shows a color-coded FLT overlay that can be easily understood by a clinician or converted into a mask region that indicates the tumor area with an accuracy of over 98% as shown in FIG. 6C. This mask can be easily created in the present disclosure, for example, by selecting a desired threshold as described above. Such a configuration is in strong contrast to the fluorescence intensity overlay of FIG. 6D, which may cause the clinician to resect too little tissue, requiring multiple consecutive operations or a large procedure.

[0049] The above FLT threshold (or cutoff) concept can also be applied to delineate positive from negative lymph nodes during surgery or neck dissection. Clinical data illustrating this application are shown in Figures 7A-7D. Specifically, Figure 7A shows a color photograph of a specimen resected from an SCC patient undergoing radical neck dissection. Two lymph nodes (LNs) were identified in the specimen (white dashed lines), only one of which was later confirmed to be positive for tumor by histology. Figure 7B shows microscopic FLT images of small sections within positive and negative lymph nodes (LNs), showing significantly longer life spans in the positive LNs. Figures 7C and 7D show wide-field fluorescence intensity and FLT maps of the above specimen, showing that FLT values ​​in the positive LNs are significantly and uniformly longer than those of the negative LNs, while the intensity shows significant heterogeneity. Furthermore, because intensity is a relative quantity and is expressed in arbitrary units, the distinction between tumor and normal tissue becomes less clear, especially across patients and imaging systems, whereas FLT is an absolute quantity and can be used to define absolutes.

[0050] In a diagnostic setting, such as a breast cancer screening test, the threshold lifetime can be directly applied to TD fluorescence data collected in a disease-free breast, provided that the subject is administered ICG (or other dyes, such as EGFR-targeted dyes) prior to imaging, and this threshold lifetime can be used to non-invasively diagnose the presence of malignant versus benign tumors.

[0051] Thus, the report generated can take a wide variety of different forms. For example, the data acquisition in processing block 202 can simultaneously acquire both conventional fluorescence intensity (also referred to as "continuous wave" (CW) fluorescence) and FLT data for display during surgery or screening examination. The system can then simultaneously display both the CW fluorescence and FLT images displayed by a standard camera in a report that can be generated in real time. This is useful in various clinical scenarios where ICG fluorescence accumulation can be directly detected by intensity images that can be collected in real time, thereby quickly guiding the surgeon to the area of ​​interest before the tumor is accurately identified using a fluorescence lifetime map. Thus, with reference to Figures 6A-6D, the above report can provide the clinician in real time with an image such as that shown in Figure 6D, which is then aligned with or replaced by a lifetime overlay (Figure 6B) or a mask overlay (Figure 6C) as the FLT data is acquired and processed.

[0052] The systems and methods described herein can be incorporated into a wide variety of implementations depending on the specific clinical application. For example, in open surgery, a time-gated wide-field intensifier camera or array detector can be used directly or in conjunction with a fiber bundle for light collection from the operating table. In some scenarios, the same camera or detector array can also be used to visualize the resected specimen to identify the presence of tumor at the surgical border and inform the pathology process.

[0053] To image organs deep inside the body, such as the lungs or liver, endoscopes or laparoscopes can be used for minimally invasive surgery. In such a scenario, pulsed light can be sent via a fiber to the light entry port of the laparoscope / endoscope, and the fluorescence emitted from the detection port of the laparoscope / endoscope can be entered into either an intensifier charge-coupled device (ICCD) camera or one or more photomultiplier tube (PMT) detectors. In the case of an ICCD camera, detection is performed by a time-gated mechanism. In the case of a PMT detector, time-resolved data can be obtained using a time-correlated single photon counting (TCSPC) method. Confocal imaging techniques can also be used in conjunction with endoscopes to achieve micron-level resolution both during surgery and in ex vivo tissues.

[0054] Illumination can be in the form of either wide-field uniform illumination or spatially patterned illumination. Patterned illumination is useful in obtaining accurate FLT maps in the presence of tissue scattering as described in U.S. Pat. No. 9,921,156, the entire contents of which are incorporated herein by reference.

[0055] <Example> Using multiple clinical and animal studies, we observed that the fluorescence lifetime of tumors is significantly longer than that of the surrounding normal tissues immediately after and up to 96 hours after intravenous injection of indocyanine green (ICG) and EGFR-targeted fluorescent probes. This difference in ICG lifetime between tumors and normal tissues allows tumors to be separated from background with an accuracy of over 98%, which is significantly better than current methods using intensity-based fluorescence imaging techniques that have an accuracy of only 50%. This 98% accuracy can be further improved by further instrumentation design targeting FLT imaging.

[0056] Studies using high resolution microscopic imaging and cell culture have also confirmed that the increased lifetime of ICG and other near infrared dyes is due to the dye localizing within the tumor cells (and not due to the dye being retained outside the tumor cells in the tumor periphery). Thus, the disclosed systems and methods can be used to detect microscopic cancers remaining in the body using appropriate equipment such as an endoscope.

[0057] We conducted an IRB-approved pilot clinical study using freshly removed specimens from 25 patients scheduled for surgery for metastatic colorectal cancer (mCRC), hepatocellular carcinoma (HCC), and oral squamous cell carcinoma (OSCC), cutaneous squamous cell carcinoma (CSCC), or bone and soft tissue sarcoma. Patients were injected with ICG 2–72 h prior to surgery, and resected specimens were imaged in the frozen section laboratory using a prototype time-domain (TD) fluorescent specimen imaging system, such as the one described above and shown in Figure 1. Figure 8A shows the time-domain (TD) fluorescent signal measured from a liver HCC. The FLT map τ(r) is obtained by fitting the attenuation portion of the TD signal (arrow in Figure 8A) to a single exponential attenuation, where r is the pixel location. Figure 8B similarly shows the TD fluorescent signal measured from an oral SCC cancer surgical specimen freshly resected from a patient who received systemic ICG 24–48 h prior to surgery. In both of these cases, FLT(t) in the tumor is a single exponential function e -t / τ (where t is the time delay), which is calculated from the attenuation of TD fluorescence data by fitting to the time delay (t), is significantly longer compared to normal tissue.

[0058] Data obtained from liver specimens with HCC tumors were also used for additional comparison and analysis. Despite the high nonspecific ICG uptake, the FLT in the tumor was significantly longer than that in the surrounding tissue, with minimal overlap. Figure 8C shows a histogram of instantaneous intensity, which indicates a large overlap between the tumor and normal tissue. Meanwhile, the histogram generated from the FLT data shows minimal overlap, as shown in Figure 8D.

[0059] Sensitivity was plotted against specificity for varying intensity and FLT thresholds to generate receiver operating characteristic (ROC) curves. Sensitivity was defined as the number of pixels in the tumor with intensity or FLT above threshold divided by the total number of pixels in the tumor. False positive rate was defined as the number of pixels outside the tumor with intensity or FLT above threshold divided by the total number of pixels outside the tumor. Accuracy was calculated as the area under the curve (AUC) and was 98% for FLT-based tumor / normal classification. This is in contrast to 40% accuracy for intensity-based tumor / normal classification.

[0060] It is noteworthy that the above was also true for patients with significant cirrhosis of the liver, suggesting that any nonspecific uptake due to cirrhosis does not affect the accuracy of tumor delineation using FLT data. Ratiometric fluorescence / reflectance analysis did not improve the accuracy of intensity-based classification. This is likely because the poor tumor contrast is due to nonspecific dye accumulation rather than variability in tissue absorption or scattering. FLT imaging microscopy (FLIM) (Stellaris 8, Leica) of lymph nodes with mCRC tumor infiltration resected from another patient showed excellent agreement between FLT-based classification results and histology, demonstrating the ability of FLT to delineate sub-millimeter-sized tumors composed of microscopic cancer cell nests with FLTs longer than the tumor stroma. For all tumor types studied, tumor FLTs were significantly longer than those of normal tissues, with classification accuracies exceeding 97%. These results demonstrate that the systems and methods disclosed herein can be used to perform margin guidance in liver cancer, where standard intensity-based fluorescence methods are not suitable.

[0061] Figure 9 shows box plots of FLTs from 13 different subjects, which shows that the mean tumor FLTs vary by approximately 5% across subjects and are significantly longer than the FLTs of tissue autofluorescence or non-specific ICG in normal / healthy tissues. Thus, this figure also shows that a "threshold" FLT can be used to classify tumors and normals, which is consistent across patients for a particular tumor type.

[0062] In one specific, non-limiting study, a system similar to that in Figure 1 was used to perform wide-field time-domain imaging, but with a supercontinuum laser providing 770 ± 30 nm excitation and a tunable filter (EXR-20, SuperK Varia, NKT Photonics, repetition rate 80 MHz, tuning range 400-850 nm), a multimode fiber (Thorlabs, Newton, NJ, USA) to deliver light to the sample, and a gated-intensifier CCD (LaVision PicoStar, gain 500 V, integration time 0.1-1 s, 150 ps step, pixels 256 × 344 (after 4 × 4 hardware binning)). A digital micromirror device was used to spread the output of the optical fiber and deliver it to the animal's body surface. The average total power over the illumination area (approximately 6 × 8 cm) was 10-20 mW. Fluorescence was collected in reflection mode using an 835 / 70 nm bandpass filter. TD fluorescence imaging was performed with gate widths of 600 ps and 150 ps steps for a total duration of approximately 6 ns for a laser duty cycle of 12.5 ns. In vivo animal imaging was performed 48 hours after intravenous injection of panitumumab-IRDye800CW (150 μl, 1 mg / ml). After imaging, animals were sacrificed and tumors were immediately cryo-mounted in optimal cutting temperature compound for fluorescence lifetime imaging microscopy (FLIM), histological staining, and immunohistochemistry (IHC) staining.

[0063] For multispectral imaging, paraffin blocks of ex vivo clinical specimens were imaged with an IVIS Spectral CT Imaging System (PerkinElmer, Waltham, MA, USA) using 710 nm excitation and 760–840 nm emission wavelengths. The camera accumulation time was automatically adjusted during image acquisition, and Living Image® software was used to extract fluorescence images normalized for accumulation time. The true fluorescence emission spectrum of panitumumab-IRDye800CW and tissue autofluorescence were used as basis functions to perform linear deconvolution of the multispectral images, and the amplitude of panitumumab-IRDye800CW and tissue autofluorescence were extracted.

[0064] We also used a STELLARIS 8 FALCON (Leica, Germany) FLIM system to perform NIR FLIM of 10 μm-thin tissue sections (murine tumors and clinical specimens). Imaging was performed using 730 nm excitation and a 750 nm notch filter, and detection was performed using a HyD R detector operating in the 770-850 nm range. A 10x, 0.4 NA objective lens was used for image collection, and digital images with a pixel count of 512 × 512 (2.275 μm / pixel) and four line repetitions and four line averages were obtained. TD data were collected using time-correlated single photon counting.

[0065] Various studies were performed with these systems. Panitumumab-IRDye800CW conjugation was performed under cGMP conditions. Briefly, panitumumab (Vectibix®, Amgen, Thousand Oaks, CA, USA) was concentrated and pH adjusted to pH 8.5 by buffer exchange against a 10 mg / mL solution of 50 mmol / L potassium phosphate. IRDye800CW (IRDye800CW-N-hydroxysuccinimide ester, LI-COR Biosciences, Lincoln, NE, USA) was conjugated with panitumumab at a molar ratio of 2.3:1 at 200 °C for 2 h in the dark. Unconjugated dye was removed by filtration using a desalting column (Pierce Biotechnology, Rockford, IL, USA) and the final protein concentration was adjusted to 2 mg / ml after buffer exchange with PBS (pH 7). The product was sterilized by filtration, packed into single-use vials and stored at 40° C. until use.

[0066] The EGFR-overexpressing cell line, MDA-MB-231, and the EGFR-negative cell line, MCF7, were purchased from ATCC and cultured in high glucose DMEM supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin (Life Technologies). Oral cancer cell lines were maintained in RPMI medium supplemented with 10% FBS and 1% penicillin-streptomycin. Cells were harvested at 80% confluency for tumor induction.

[0067] For in vitro experiments, FaDu cells were plated at 0.2 × 106 cells per well in 12-well plates with poly-D-lysine-coated glass coverslips and allowed to adhere to the coverslips for 24 h. The cells were then incubated with panitumumab-IRDye800CW (100 μg), IgG-IRDye800CW (100 μg) or PBS (pH 7.4) at 370 °C for 2 h. After probe incubation, cells were fixed in 4% paraformaldehyde (PFA) and mounted for confocal FLIM with the addition of Prolong Gold browning inhibitor (ThermoFisher Scientific, Waltham, MA, USA).

[0068] All patients provided informed consent, and the study protocol was approved by the IRB and FDA. The study was conducted in accordance with the 1975 Declaration of Helsinki and its amendments, the FDA's ICH-GCP guidelines, and US regulations. The manufacturing process of panitumumab-IRDye800CW at Frederick National Laboratory is described above. Consented patients who met the study criteria were allowed access to the infusion center for administration of panitumumab-IRDye800CW. Panitumumab-IRDye800CW was administered systemically at a dose of 0.6 mg / kg 48 hours before surgery. Ex vivo OSCC tissues from patients who received systemic panitumumab-IRDye800CW were formalin fixed, sectioned, and embedded in paraffin. Paraffin-embedded tissue blocks were then sent for TD imaging studies.

[0069] All animal studies were approved by the Animal Care and Use Committee in accordance with the Massachusetts General Hospital Animal Welfare Guidelines. Seven female nu / nu mice (4–6 weeks old) were purchased from Charles River Laboratories and housed in the animal facility at Massachusetts General Hospital (Boston, MA). The animals were quarantined for 1 week and fed a normal diet with a 12-h light / dark cycle. After 1 week, animals were anesthetized with 3% isoflurane and injected subcutaneously with 2 × 106 MDA-MB-231 (n = 5, EGFR overexpressing) or MCF7 (n = 2, EGFR negative) cells in a 1:1 PBS:Matrigel mixture. Mice receiving MCF7 cells were also implanted with sustained-release estrogen pellets to promote tumor growth. Tumors were measured every 2 days until the tumors reached a diameter of 5–10 mm in one dimension.

[0070] For histopathology and immunohistochemistry, OSCC tumors surrounded by surrounding normal tissue were fixed in 10% formalin, embedded in paraffin, sectioned (10 μm thick), stained with hematoxylin and eosin (H&E), or processed for IHC. For IHC, 10 μm thick paraffin-embedded tissue sections were dewaxed with xylene and rehydrated in decreasing concentrations of alcohol. Antigen retrieval was performed with EDTA (pH 9.0) at subboiling temperature for 15 min. Tissue sections were incubated overnight at 4°C in a 1:50 dilution of anti-EGFR antibody (Cat#4267, Cell Signaling Tech). Secondary antibodies were applied for 30 min at 37°C, slides were developed with DAKO HRP-compatible DAB (Cat#SF-4100, Vector Laboratories, Burlingame, CA, USA), and secondary staining was performed with Harris hematoxylin. Images of H&E and IHC stained tissue sections were obtained using a Keyence BZ-X810 inverted microscope (Keyence, Itasca, IL, USA). A Plan Apo 10x, 0.45NA air objective (Nikon, Tokyo, Japan) and a monochromatic CCD (colorized by an LC filter) were used to capture images. Tissue images were graded by two experienced pathologists.

[0071] For wide-field TD data analysis, TD fluorescence images were analyzed using custom software in MATLAB (Mathworks, Natick, MA, USA). As shown in Figures 10A and 10B, TD data for individual pixels were time-gated and plotted against log(counts), where FLT is the decay portion of the TD fluorescence profile as a single exponential function e -t / τ(r)where r is the pixel position and τ(r) constitutes the lifetime map. More specifically, FIG. 10A shows representative fluorescence attenuation curves of panitumumab-IRDye800CW (gray solid line), IgG-IRDye800CW (black dashed line) and PBS (gray dashed line) in cancer cells. FIG. 10A also shows representative fluorescence attenuation curves of panitumumab-IRDye800CW (gray solid line), IgG-IRDye800CW (black dashed line) in culture medium, and a representative fluorescence attenuation curve of the stock solution of panitumumab-IRDye800CW in PBS (gray dotted line).

[0072] Tissue images were co-registered with the fluorescence intensity and FLT maps. Regions of interest (ROIs) of histologically confirmed tumor and normal tissues were then mapped onto the co-registered fluorescence intensity and FLT images. The intensity and FLT of pixels enclosed by the ROI were used to calculate the probability distribution of pixels as normal or tumor. Receiver operating characteristic (ROC) curves were obtained by calculating sensitivity and specificity with varying intensity and FLT thresholds. Sensitivity is expressed as the number of pixels above the intensity or FLT threshold in the tumor ROI divided by the total number of pixels in the tumor ROI. Specificity is calculated as the number of pixels below the threshold in the normal ROI divided by the total number of pixels in the normal ROI.

[0073] For FLIM and IHC image analysis, FALCON® / FLIM software was used to collect and analyze FLIM data. Lifetime values ​​for each pixel location were calculated by using single exponential fitting of the fluorescence attenuation curves. First, large area FLIM and IHC images stitched from each tissue slice were co-registered using a custom MATLAB® code. Images were then divided into multiple regions of interest (ROIs) with pixel size of 300×300. ROIs with less than 10% of pixels represented by tissue were excluded from subsequent analysis. EGFR-positive pixels were extracted in each ROI by analyzing the IHC image ROIs by color deconvolution using the IHC toolbox in ImageJ (NIH, version 1.48u). EGFR expression levels in the ROIs were expressed as percentage of EGFR-positive pixels. The corresponding FLIM image ROIs were analyzed by averaging the FLT values ​​above approximately 0.3 nm. The EGFR expression rate and mean FLT value for each pair of IHC and FLIM ROIs were compared using scatter plots and correlation coefficients.

[0074] Statistical analysis was performed using the Mann-Whitney U test (left and right tailed) to estimate p-values ​​for the bar graphs. A p-value of less than 0.05 was considered significant. * P < 0.05, ** P<0.01. The study was not randomized and the investigators were not blinded to the allocation during the study and outcome assessments. Results are presented as mean±standard deviation.

[0075] <Result> In vitro measurements confirmed that the cell specificity of FLT was improved in head and neck cancer cell lines with high EGFR expression using the fluorescence lifetime imaging microscopy (FLIM) technique. For example, Fig. 10C shows the microscopic fluorescence intensity and FLT maps of FaDu cells cultured with panitumumab-IRDye800CW (left), IgG-IRDye800CW (middle), or PBS (right). Cells treated with panitumumab-IRDye800CW showed higher probe uptake and longer FLT values ​​(≈0.75 ns) compared to cells cultured with IgG-IRDye800CW, whereas cells cultured with IgG-IRDye800CW showed lower uptake and shorter FLT (≈0.61 ns). Cells cultured with PBS showed basal levels of autofluorescence and an average FLT of 0.55 ns under the same experimental conditions. As shown in FIG. 10A, a representative fluorescence decay profile shows that panitumumab-IRDye800CW in cells has the longest decay time, followed by IgG-IRDye800CW, and then PBS. As shown in FIG. 10D, panitumumab-IRDye800CW and IgG-IRDye800CW in the medium collected from the in vitro experiment shown had short FLTs of 0.58 ns for the former and 0.56 ns for the latter, which were equivalent to the FLT of the original solution of panitumumab-IRDye800CW dissolved in PBS (0.58 ns). This result suggests that environmental effects due to medium conditions are not responsible for the FLT improvement of panitumumab-IRDye800CW observed in cancer cells. Figure 10A shows representative fluorescence attenuation profiles of panitumumab-IRDye800CW and IgG-IRDye800CW in culture medium, which did not differ from the attenuation profile of panitumumab-IRDye800CW in PBS. These in vitro results suggest EGFR specific uptake and improved FLT of panitumumab-IRDye800CW in cancer cells.

[0076] We also performed retrospective imaging studies using tissue specimens excised by glossectomy from the lingual margin of an OSCC patient who received a systemic injection of panitumumab-IRDye800CW 48 hours prior to surgery. Specifically, Figures 11A-11C show representative specimen FLIM images (left), EGFR IHC images (middle), and H&E stained tissue images (right), which demonstrate that FLT is longer in OSCC tumors corresponding to the high EGFR expression in the tumor regions. More specifically, Figure 11A shows a wide-field region of interest (ROI), where longer FLT (red) is observed in two EGFR-overexpressing tumor clusters indicated by arrows. Normal tissues with low EGFR expression surrounding the tumors show shorter FLT (green / blue), which is consistent with the FLT of nonspecific panitumumab-IRDye800CW and tissue autofluorescence. Figure 11B also shows a magnified view of the region of interest (ROI, shown as a dashed box) in Figure 11A, and Figure 11C shows a magnified view of the ROI in Figure 11B. FLIM images show spatial colocalization of long FLT in high EGFR-expressing tumor cells, and tumor specificity of FLT enhancement is observable down to single-cell resolution level in individual OSCC cell clusters (Figure 11C, arrows).

[0077] Referring now to Figures 12A-12C, several examples are shown in which tumor contrast of FLT is superior to fluorescence intensity for tissue regions with strong nonspecific uptake of panitumumab-IRDye800CW. In Figure 12A, the spatial distribution of fluorescence intensity and FLT is shown within a large ROI of a specimen containing high EGFR expression tumor regions and normal tissue with low or no EGFR expression. The corresponding tissue and IHC images clearly delineate the tumor border (dashed line) from the normal salivary gland (SG), muscle (M) and connective tissue (CT) layers. Although the fluorescence intensity was comparable between tumor, salivary gland and muscle, the FLT of the tumor region was significantly longer than that of the surrounding normal tissue. In the IHC images, EGFR expression was highest in the tumor, followed by salivary gland and lowest in muscle. This trend was highly consistent with the FLIM images, in which the FLT at the tumor front was significantly longer (0.9–1 ns) compared to the FLT in salivary glands (0.7 ns) and muscles (0.5 ns). It should be noted that the lower uptake and shorter FLT in the tumor interior in Figure 12A are due to the presence of connective tissue that does not express EGFR, the high density of lymphocytes, and the heterogeneous tumor penetration of panitumumab-IRDye800CW. More homogeneous tumor penetration of panitumumab-IRDye800CW can be achieved by coadministering a loading dose of unlabeled panitumumab with panitumumab-IRDye800CW. FLT in tumor cell clusters was consistently longer than that in normal tissue, and regions of longer FLT values ​​coexisted with regions of high EGFR expression.

[0078] In Figure 12B, a representative ROI containing one cluster of a few OSCC cells far from the primary tumor mass is shown, as observed at high magnification (20x). In this ROI, both fluorescence intensity and FLT show high contrast of tumor cells due to low local nonspecific uptake of panitumumab-IRDye800CW. In addition, Figure 12C also shows ROIs with high nonspecific fluorescence due to three additional OSCC cell clusters (arrows) surrounded by muscle and lymphocytes. These EGFR-overexpressing OSCC cells are only distinguishable on the FLT image and are barely distinguishable from the surrounding normal tissue on a fluorescence intensity basis. All three cell clusters showed a mean FLT of 0.96 ns, which was significantly longer than the FLT of 0.5 ns for muscle. From the above data, it can be seen that while fluorescence intensity imaging can identify specific tumor cell clusters, many tumor cell clusters are indistinguishable from the background due to nonspecific fluorescence. On the other hand, the FLT of panitumumab-IRDye800CW was consistently longer in tumor cells and specific for EGFR expression in tumor cells, which allowed robust separation of tumor from normal tissues at the microscopic level, and confirmed that the longer tumor FLT in oral cancer specimens was not due to intrinsic tissue autofluorescence.

[0079] From the FLT images in Figures 11A-12C, it can be seen that the FLT is longer in regions with higher EGFR expression. Although increased FLT is expected at individual foci of EGFR binding in the cell membrane or cytoplasm, resolution at the level of individual molecules in non-cellular compartments cannot be achieved using confocal imaging methods. Thus, the FLT at a particular pixel of a microscopic image will be the spatial average across the various sites in the cell that encompass a range of EGFR expression levels, and the FLT at this particular pixel should correlate with the average EGFR expression in the tissue region corresponding to that pixel.

[0080] To investigate this correlation, we quantified the relationship between EGFR expression and the mean FLT in multiple tissue sections. IHC-based quantification of EGFR expression in ROIs with low, medium, and high EGFR expression revealed that FLT of panitumumab-IRDye800CW correlated strongly with EGFR expression, as generally shown in Figures 13A-13G. Specifically, Figure 13A shows IHC of three representative ROIs in ascending order of EGFR expression, including muscle, salivary gland, and tumor. Despite accumulation of panitumumab-IRDye800CW in tumor cells with overexpressing EGFR, there was significant nonspecific uptake in muscle with low EGFR expression and salivary gland with moderate EGFR expression, rendering the tumor indistinguishable from normal areas based on fluorescence intensity, as evidenced by the large overlap of intensity histograms for the three ROIs, as shown in Figure 13B. On the other hand, the FLT of panitumumab-IRDye800CW was shortest in muscle regions with low EGFR expression and longest in tumor cells with high EGFR expression, as shown in Figure 13C. The average FLTs in the representative muscle, salivary gland, and tumor ROIs were 0.4, 0.5, and 0.88 ns, respectively. Figure 13D shows a scatter plot of the mean fluorescence intensity versus the percent area positive for EGFR by IHC in all ROIs investigated. The scatter plot shows low correlation (r=-0.12), suggesting that the presence of strong nonspecific uptake makes it difficult to reliably quantify EGFR expression using the fluorescence intensity of panitumumab-IRDye800CW. Figure 13E shows the mean fluorescence intensity in IHC-confirmed EGFR negative and positive pixels, obtained by co-registration of IHC and FLIM images, which are not statistically different. The mean FLT of each ROI showed a strong positive correlation (r=0.87) with the percent area positive for EGFR in IHC, as shown in Figure 13F, suggesting that ROIs with high EGFR expression had longer mean FLTs.Furthermore, as shown in Figure 13G, EGFR-positive pixels exhibited significantly longer mean FLTs than EGFR-negative pixels. Thus, the results shown in Figures 13A-13G confirm that the FLTs of panitumumab-IRDye800CW in human tissues are highly correlated with EGFR expression and can provide a quantitative estimate of tissue EGFR expression in the presence of strong nonspecific uptake.

[0081] Having thus established the cellular specificity of FLT enhancement of panitumumab-IRDye800CW, we next assessed the tumor-normal classification capability of widefield FLT imaging in thick macroscopic tissues, which is important for intraoperative in situ imaging and ex vivo imaging of large resected specimens. Specifically, Figures 14A-14K show widefield time-domain (TD) and spectral reflectance imaging of the same tissue block, which contained the slide used for the microscopic FLIM and histological analysis shown in Figure 12A. A color photograph of the specimen in paraffin block (Figure 14A) was co-registered with the histological diagnosis (Figure 14B), and the tumor ROI was delineated by the pathologist (black dotted line). Wide-field fluorescence imaging showed a broad and heterogeneous distribution of fluorescence intensity inside and outside the tumor boundary (white dotted line) defined by histology (Fig. 14C), suggesting high levels of nonspecific fluorescence in uninvolved muscles and salivary glands (white arrows in Fig. 14C), which was nearly indistinguishable from tumor fluorescence. We confirmed that spectral separation using predefined tumor and normal spectral basis functions could not clearly distinguish tumor and normal regions (Fig. 14D and Fig. 14E). However, the FLT in tumor regions was significantly longer than that of normal tissue (Fig. 14F), with only slight overlap. Figures 14H, 14I, and 14J show histograms of fluorescence intensity, histograms of spectral separation amplitude, and histograms of FLT inside and outside the tumor boundary, respectively. These distributions show significant overlap in fluorescence intensity and spectral amplitude, but the FLTs are different between tumor and normal ROIs, and the overlap of these FLT distributions is minimal. Receiver operating characteristic (ROC) analysis using pixel intensity and FLT distributions is shown in Figure 14K, and results in an area under the curve (AUC) of 0.98 for FLT-based tumor / normal classification, in contrast to an AUC of 0.32 for intensity-based classification (Figure 14K).Although the accuracy was improved over intensity using spectral unmixing (AUC=0.58), the performance of spectral classification is still significantly lower than that of FLT imaging. The high accuracy of FLT imaging for tumor / normal classification is confirmed by the excellent agreement between the macroscopic FLT imaging-based tumor-normal boundary and the corresponding boundary obtained from microscopic FLIM images of thin sections of the same slide (Figure 14G, reproduced from Figure 12A for ease of comparison) and tissue images (Figure 14B). Thus far, FLT images have been obtained using thin tissue slices or in thick excised tissues, but in vivo detection of FLT changes in biological tissues in the presence of strong light absorption and scattering is also feasible under a wide range of conditions. We have extensively validated the accuracy of FLT imaging for in vivo detection of EGFR-overexpressing tumors using animal tumor models.

[0082] The above examples and results have clinically demonstrated that the FLT of a cancer-targeted NIR probe injected systemically into a patient is increased when the cancer-targeted NIR probe is in the cancer cells compared to normal tissues. Furthermore, the increased tumor FLT of the probe can be utilized to achieve unprecedented precision in tumor delineation both at the microscopic level in thin tissue sections and in macroscopic thick tissue specimens, thereby enabling precise quantification of receptor expression in tissues. This demonstrates cancer specificity at the cellular level in human tissues using exogenous cancer-targeting drugs.

[0083] In addition to providing improved tumor detection accuracy, the systems and methods provided herein are also insensitive to measurement parameters such as optical illumination power, camera sensitivity, and other system-specific scaling factors, and therefore FLT can serve as an absolute parameter that can be easily compared across multiple imaging systems and studies, facilitating better standardization of image-guided surgery.

[0084] The cell specificity of the FLT in cancer has implications beyond microscopic imaging of thin tissue sections, and can be exploited to image tumors in deep tissues. FLT measurements are not altered by tissue light propagation under a wide range of conditions, and FLT can be estimated in the presence of thick tissues without the need to know the tissue optical properties, which are often difficult to estimate. Thus, the cell specificity of the FLT for cancer indicates that FLT measured through thick biological tissues is due exclusively to tumor cells and not to non-specific probes. This contrasts strongly with fluorescence intensity, which is strongly attenuated by tissue light propagation (besides the inability to distinguish cancer cell-specific from non-specific fluorescence), and therefore requires full knowledge of the tissue optical properties and tissue thickness to accurately quantify probe uptake.

[0085] The ability to measure FLT through deep tissues can be useful when tumors need to be detected that are embedded in thick macroscopic tissues, for example to assess the depth of margins in resection specimens, or when non-invasively imaging deep tumors in whole organs. Tomographic reconstruction algorithms (e.g., U.S. Pat. No. 9,927,362, “System and Methods for tomographic lifetime multiplexing,” the entire contents of which are incorporated herein by reference) can be used to detect and localize FLT in deep tissues. Such methods take advantage of the relative independence of FLT from tissue scattering and absorption, assuming that the in vivo FLT is longer than the intrinsic tissue absorption timescale (≈0.2 ns), which is sufficient for many NIR fluorophores, including IRDye800CW. Thus, the systems and methods provided herein can be extended to create diagnostic systems that quantify cancer-related biomarkers using whole-body measurements, or to provide a rapid on-off "light switch" readout based on a pre-determined FLT threshold.

[0086] Previously, FLT imaging has been applied in preclinical studies at the microscopic and whole animal levels. Visible FLIM has been evaluated for image-guided surgery that utilizes the intrinsic FLT of tissue components, but the intrinsic FLT contrast between tumor and normal tissues is necessarily low, which reduces sensitivity and specificity. Furthermore, endogenous fluorescence imaging systems use visible light, which precludes the ability to image subsurface tumors due to high tissue attenuation, limiting intraoperative applications to exposed tumors. NIR drugs can take advantage of the deeper sensitivity of NIR light for intraoperative imaging or imaging of deep tissues. Furthermore, external targeting drugs can be used to report on molecular expression markers. However, endogenous visible spectrum FLIM can clearly delineate various tissue structures that may provide valuable morphological information to complement the NIR tumor signal with external drugs.

[0087] The data presented in Figures 14A-14K suggest that spectral unmixing techniques can mitigate the lack of tumor contrast due to nonspecific uptake and may be useful when FLT imaging systems are not readily available. The tumor-normal spectral contrast is still not sufficient to provide tumor detection accuracy comparable to that of FLT. It should be noted that because spectral contrast is essentially an intensity-based measure, it has similar limitations as intensity, such as dependence on measurement parameters, tissue absorption and scattering. This makes it more difficult to quantify in thick tissues using such techniques.

[0088] The usefulness and safety of EGFR antibody-labeled NIR probes have been extensively studied for OSCC, pancreatic tumors, and brain tumors. Although these studies have shown significant improvements in sensitivity compared to visual identification and palpation, intensity is not always reliable because nonspecific uptake in tissues is heterogeneous and can vary in multiple regions of a given specimen. This is clearly illustrated by the data above, where intensity performs well in some oral tissue regions with good tumor uptake, but other areas of oral tissue, such as the salivary glands, have high nonspecific uptake and significantly reduce tumor contrast. On the other hand, the FLT of tumors is consistently and uniformly longer than that of normal oral tissues, providing a robust measure of tumor uptake.

[0089] Panitumumab-IRDye800CW has been extensively tested for safety in humans and intraoperative FLT imaging has been shown to be clinically feasible, so the results presented here have direct clinical significance for intraoperative surgical guidance in EGFR-overexpressing cancers. More than 90% of head and neck cancers overexpress EGFR. There are multiple clinical trials of anti-EGFR antibody-labeled probes for OSCC, as well as EGFR antibody-based probes in brain tumors, colorectal cancer, and pancreatic cancer. Thus, FLT imaging with panitumumab-IRDye800CW is expected to have a strong impact on surgical guidance in these cancers. In addition to targeting EGFR, FLT contrast may also be advantageous for tumor imaging with other receptor-targeted probes. Early preclinical studies have shown tumor-specific FLT changes in mice with a fluorescently labeled antibody against human epidermal growth factor receptor-2 (HER-2). Probes targeting immune-expressed markers also show longer life spans in tumor cells than in normal tissues. As new probes targeting cancer-specific molecular markers continue to be developed, the systems and methods provided herein can also be applied to these newly developed agents. Due to its powerful and unique advantages, FLT imaging using targeted molecular imaging agents may play an important role in a wide range of clinical settings, from cancer diagnosis to surgical treatment.

[0090] 15A-15D are a series of graphs showing fluorescence lifetime enhancement in sarcomas. The violin diagrams in 15A and 15B show the distribution of fluorescence intensity (FIG. 15A) and FLT (FIG. 15B) in sarcomas and several normal tissues resected from multiple patients (n=10) who received ICG 0-48 hours before surgery. Here, BV is blood vessel, CT is connective tissue, and AT is adipose tissue. The graph in FIG. 15C shows the average fluorescence intensity of tumors and normal tissues in seven patients, and the corresponding FLT of tumors and normal tissues in seven patients is shown in FIG. 15D. In both figures, the circle on the left represents the tumor value, and the normal tissue is represented by the circle on the right. The average was calculated for multiple (>20) ROIs of tumor or normal tissues of each patient identified by histology. The dashed lines in FIG. 15A and 15B represent the threshold intensity or threshold FLT that provides the highest sensitivity and specificity. Mann-Whitney U test (left and right tailed): *** p<0.001

[0091] Figures 16A-16D are a series of graphs showing fluorescence lifetime enhancement in head and neck cancer. Figures 16A and 16B are violin plots showing the distribution of fluorescence intensity (Figure 16A) and FLT (Figure 16B) in head and neck cancer and various normal oral tissues from multiple patients (n=8) who received indocyanine green (ICG) at least 24 hours before surgery. Here, NE is normal epithelium, NS is normal stroma, DS is desmoplastic stroma, and SG is salivary gland. Figure 16C shows the average fluorescence intensity and Figure 16D shows the FLT in six patients. In both figures, the tumor is represented by the circle on the left side of each graph, and the normal tissue is represented by the circle on the right side. The filled circle itself represents the average of multiple (>20) ROIs of tumor or normal tissue of each patient identified by histology. Mann-Whitney U test (left and right tails): *** p<0.001

[0092] Figure 17A is a graph of a receiver operating characteristic (ROC) plot of the sensitivity and false positive rate (1-specificity) of fluorescence lifetime (FLT) and intensity-based tumor / normal classification in 10 sarcoma patient specimens. Figure 17A shows accuracies (measured as area under the curve (AUC)) of 0.96 and 0.56, respectively. Figure 17B is a similar ROC graph in 8 head and neck cancer patients, showing an accuracy (AUC) of 0.96 for FLT-based tumor / normal classification and an accuracy (AUC) of 0.61 for intensity-based tumor / normal classification.

[0093] Figures 18A-18H are graphs and images showing in vitro and "in vivo" lifetime multiplexing of near-infrared fluorescent proteins (iREFs). All fluorescence data was acquired using one excitation / emission filter pair: ex (excitation): 650 / 40 nm, em (emission): 700 nm long pass. Figure 18A is a fluorescence intensity image of a microorganism expressing five iRFP variants (iRFP670, 682, 702, 713 and 720), and Figure 18B is a fluorescence lifetime image of the same microorganism. Figure 18C shows the normalized excitation spectrum, and Figure 18D shows the emission spectrum of iRFP670, 702 and 720. Figure 18E is a time domain (TD) fluorescence signal graph of iRFP670, 702 and 720 in the microorganism. Figure 18F is a lifetime histogram derived from Figure 18B. FIG. 18G is an image showing the fluorescence intensity and FIG. 18H is an image showing the fluorescence lifetime of three MTLn3 tumors expressing iRFP670, 702, and 720 in the mammary fat pad of female nude mice.

[0094] Figures 19A-19D show images of a mouse using intensity imaging (Figure 19A) and tomographic fluorescence lifetime multiplexing of anatomically targeted fluorophores (Figures 19B-19D). The mouse in Figures 19A-19D was intravenously injected with Osteosense800 (0.65ns, skeletal target) and ZE169 (0.5ns, kidney target). The tumor is clear in Figures 19B-19D, but difficult to distinguish in Figure 19A.

[0095] The present invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof, and the above-described embodiments are in all respects merely illustrative and should not be construed as limiting the present invention. The scope of the present invention is therefore defined by the appended claims rather than the above description. All changes that come within the meaning and range of equivalency of the claims are to be embraced within their scope.

Claims

1. 1. A method for evaluating a tissue to determine the presence or absence of cancer cells, comprising: acquiring fluorescence lifetime (FLT) data from the tissue that received the fluorescent compound; processing the FLT data to identify an FLT signal at each of a plurality of sites in the tissue; Identifying FLT data above a threshold indicating the presence of cancer cells at any one of the plurality of sites; generating a report indicating that any site among the plurality of sites exceeds the threshold indicative of the presence of cancer cells; The method according to claim 1, further comprising:

2. acquiring the FLT data includes acquiring fluorescence data from the tissue at a sequence of a plurality of consecutive time points from a selected time origin to a selected end time point. The method of claim 1.

3. The FLT data is y ＴＤ (t)=a ０ e －ｔ／τ is given by Where: y ＴＤ (t) is the time domain data as a function of time t from the selected time origin to the selected end time; a ０ is the attenuation amplitude related to the fluorophore concentration, quantum yield, and an empirical scaling constant, τ is the lifetime of the fluorophore applied to the tissue to derive the FLT data; The method of claim 2.

4. Linearizing the FLT data based on a logarithmic expansion or a series expansion, and fitting the linearized data using a linear least squares based method to obtain the lifetime. The method of claim 3.

5. acquiring the FLT data includes acquiring fluorescence data from the tissue cumulatively in time starting from a selected time origin and ending at a selected time point or at multiple time points; The method of claim 1.

6. The FLT data is [0010] is given by Here, y ＱＴＤ is the quasi-time domain data as a function of time t from the selected time origin to the end time point or at multiple time points; The method of claim 5.

7. The FLT data is [0025] is given by Here, y ＱＴＤ is the quasi-time domain data as a function of time t from the selected time origin to the selected end time point or times, The FLT data is a function a ０ τ(1-e －ｔ／τ ) is fitted, τ is the lifetime of the fluorophore applied to the tissue to derive the FLT data; a ０ is the attenuation amplitude related to the fluorophore concentration, quantum yield, and an empirical scaling constant, The method of claim 5.

8. Linearizing the FLT data based on a logarithmic expansion or a series expansion, and fitting the linearized data using a linear least squares based method to obtain the lifetime. The method of claim 7.

9. Using continuous wave (CW) fluorescence intensity data and one time gate, [0030] generating a FLT image using where [0045] is the gate width T １ The FLT data obtained in The method of claim 7.

10. acquiring said FLT data includes acquiring fluorescence data from the tissue at one or more modulation frequencies and phases; The method of claim 1.

11. The threshold is a cutoff lifetime selected for one of a particular cancer type or anatomical region. The method of claim 1.

12. generating the report includes generating an overlay of the FLT data spatially registered to the tissue. The method of claim 1.

13. The overlay may include a mask defining a percentage probability of cancer, or may include color coding to display the FLT data as a lifetime map.

13. The method of claim 12.

14. generating the report includes generating a lifetime histogram. The method of claim 1.

15. obtaining the FLT data includes placing an optical probe in proximity to the tissue to obtain the FLT data; The method of claim 1.

16. The tissue is located at a surgical site in a living body. The method of claim 1.

17. 1. A medical imaging system comprising: a light source configured to deliver light to the tissue; a detector configured to receive light fluorescently emitted by the tissue and generate fluorescence lifetime (FLT) data; A processor; It is equipped with The processor, determining the presence or absence of cancer in the tissue by analyzing the FLT data; Generate a report indicating the spatial location of any cancer determined to be present in said tissue. It is structured as follows: The medical imaging system further comprises: a display configured to display the report to guide a surgical procedure to remove the cancer; and A medical imaging system comprising:

18. The processor further comprises: analyzing the FLT data to identify an FLT signal at each of a plurality of sites in the tissue; analyzing the FLT data by identifying FLT data above a threshold indicative of the presence of cancer cells at any one of the plurality of sites; 20. The medical imaging system of claim 17, configured to:

19. The threshold is a selected cutoff lifetime for one of a particular cancer type or anatomical region.

20. The medical imaging system of claim 18.

20. The processor is further configured to generate the report by generating an overlay of the FLT data spatially registered to the tissue.

20. The medical imaging system of claim 17.

21. The overlay may include a mask defining a percentage probability of cancer, or may include color coding to display the FLT data as a lifetime map.

21. The medical imaging system of claim 20.

22. The processor is further configured to generate the report by generating a lifetime histogram.

20. The medical imaging system of claim 17.

23. The processor is configured to assemble the FLT data by receiving fluorescence data from the detector cumulatively over time, either continuously or at multiple time points starting from a selected time origin and ending at a selected time point.

20. The medical imaging system of claim 17.

24. The FLT data is y ＴＤ (t)=a ０ e －ｔ／τ is given by Where: y ＴＤ (t) is the time domain data as a function of time t, which may range from the selected time origin to an end time point; a ０ is the attenuation amplitude related to the fluorophore concentration, quantum yield, and an empirical scaling constant, 20. The medical imaging system of claim 17.

25. The FLT data is [0050] is given by Here, y ＱＴＤ is the quasi-time domain data as a function of time t from a selected time origin to an end time point or at multiple time points; 20. The medical imaging system of claim 17.

26. The FLT data is [006] is given by Here, y ＱＴＤ is the quasi-time domain data as a function of time t from a selected time origin to an end time point or at multiple time points; The FLT data is expressed as a function τ(1-e －ｔ／τ ) is fitted, τ is the lifetime of the fluorophore applied to the tissue to derive the FLT data; 20. The medical imaging system of claim 17.

27. and a sample chamber configured to receive an in vitro sample for data acquisition.

20. The medical imaging system of claim 17.

28. The detector is further configured to receive light fluorescently emitted by the tissue and generate intensity data.

20. The medical imaging system of claim 17.

Images