This PDF file contains the front matter associated with SPIE Proceedings Volume 11943, including the Title Page, Copyright information, Table of Contents, and Conference Committee listings.

Colorectal cancer (CRC) is the third most common and the second most deadly type of cancer worldwide. Developing new technologies for accurate CRC detection/delineation for resection during microsurgery requires unveiling tissue biochemical and microstructural changes associated with carcinogenesis. These changes can be probed by diffuse reflectance spectroscopy (DRS), which is capable of extracting tissue chromophore concentrations and scattering parameters. Previous CRC studies have been mostly restricted to chromophores in the visible region and analytical light diffusion models. In this study, we extended this wavelength range to 350–1919 nm and used the range between 450–1590 nm to extract tissue biochemical and microstructural parameters. This extraction was performed by using DRS spectral fitting based on a reflectance look-up table built using Monte Carlo simulations of light propagation in tissues. Tissue parameters were used as an input to classification and regression tree algorithm to estimate parameter thresholds leading to best tissue differentiation for CRC detection/delineation. Differentiation between mucosa and tumor tissues was based on 2889 diffuse reflectance spectra from fresh ex vivo tissue samples from 47 subjects. All analyses were performed to investigate data of superficial tissue up to 1.1 mm and deeper tissue layers up to 1.8 mm. The most important parameters for CRC detection were total lipid content, water content, reduced scattering amplitude, Mie scattering power, and microvascular parameters. We not only confirmed the importance of these parameters with metrics in addition to statistical tests and classification models of our previous studies, but also extended the motivation of achieving successful tissue classification with an area under the receiver operating characteristic curve (AUC) higher than 90% with interpretable DRS spectral fitting parameters. Our analysis may have important clinical applications for the rapid diagnosis of colorectal neoplasia.

As rapidly accelerating technology, fluorescence guided surgery (FGS) has the potential to place molecular information directly into the surgeon’s field of view by imaging administered fluorescent contrast agents in real time, circumnavigating pre-operative MR registration challenges with brain deformation. The most successful implementation of FGS is 5-ALAPpIX guided glioma resection which has been linked to improved patient outcomes. While FGS may offer direct in-field guidance, fluorescent contrast agent distributions are not as familiar to the surgical community as Gd-MRI uptake, and may provide discordant information from previous Gd-MRI guidance. Thus, a method to assess and validate consistency between fluorescence-labeled tumor regions and Gd-enhanced tumor regions could aid in understanding the correlation between optical agent fluorescence and Gd-enhancement. Herein, we present an approach for comparing whole-brain fluorescence biodistributions with Gd-enhancement patterns on a voxel-by-voxel basis using co-registered fluorescent cryo-volumes and Gd-MRI volumes. In this initial study, a porcine-human glioma xenograft model was administered 5-ALA-PpIX, imaged with MRI, and euthanized 22 hours following 5-ALA administration. Following euthanization, the extracted brain was imaged with the cryo-macrotome system. After image processing steps and non-rigid, point-based registration, the fluorescence cryo-volume and Gd-MRI volume were compared for similarity metrics including: image similarity, tumor shape similarity, and classification similarity. This study serves as a proof-of-principle in validating our screening approach for quantitatively comparing 3D biodistributions between optical agents and Gd-based agents.

Thorough tumor resection is crucial for successful treatment of squamous cell carcinomas (SCCs) because positive surgical margins are associated with poor patient prognosis. Current methods of margin analysis, however, are limited by inefficient pathological read-times that increase exponentially with tissue size. Here, a fluorescence paired-agent imaging (PAI) approach is presented to identify regions of tumor burden in whole, thick tissue margins to act as a rapid screening tool and help focus pathological evaluation. The approach was applied to mouse models of head and neck SCC, and positive tumor burden was detected and localized in deep tissue margins up to 1.3 mm thick. Serial sections with hematoxylin and eosin and EGFR-immunostaining demonstrated good correlation with binding potential (BP: proportional to targeted biomolecule concentration) maps generated from PAI fluorescence slices and confirmed the presence of positive margins suggested by high intensity regions in the whole tissue BP maps. Findings support the use of PAI as a rapid screening method for detecting regions of tumor burden in large, en face tumor margin sections.

Introduction The 5 year survival rate of pancreatic cancer is <10%. Most patients have metastatic disease at time of diagnosis, often to the liver. Innovative imaging modalities, i.e. fluorescence guided surgery (FGS), may better appreciate metastatic disease and guide treatment. Mucin 4 (MUC4), a glycoprotein, is found in 89% of pancreatic cancers and absent in normal pancreatic tissue making it a candidate for tumor targeting in FGS. In the present study, a fluorescently-labeled MUC4 antibody preferentially targets patient pancreatic cancer in a mouse model. Methods and Materials A MUC4 antibody was conjugated to the infrared dye IRDye800CW (LICOR, Lincoln, NE) to synthesize MUC4-IR800. A high MUC4 expressing patient-derived hepatic metastatic pancreatic tumor (Panc Met) was divided into 1mm3 tumor fragments and implanted under the skin of the nude mouse. After the tumors grew ~5mm3, two mice received 50 μg and two mice received 75 μg of MUC4-IR800 via tail vein injection. Daily in-vivo imaging was performed with the Pearl Trilogy Imager (LICOR, Lincoln, NE) for 3 days. Tumor to background ratios (TBR) were calculated using skin as background. Results MUC4-IR800 selectively imaged the Panc Met tumors (see figure below). TBRs for all time points and doses were <2. The 75 μg arm had higher TBRs at 24 and 72 hours. At 48 hours, the TBRs were the same. Conclusion This present study demonstrated the successful targeting of a patient hepatic metastatic pancreatic cancer mouse model with MUC4-IR800. This has potential to improve metastatic pancreatic cancer detection. Future studies will be conducted with orthotopic models.

Background: Colon-cancer liver metastases is the frequent cause of death due to difficulties in visualizing margins of the metastases resulting in incomplete resection. To perform safer and more reliable liver surgery, indocyanine green (ICG) labeling has been used to visualize liver tumors and liver segment, but it is difficult to distinguish between a liver metastasis and its adjacent liver segment with traditional use of ICG alone. We have previously developed a method to label a liver metastasis with a tumor-specific fluorescent conjugated antibody and the adjacent liver segment with ICG in order to perform image guided metastasectomy. Methods: Nude mice were surgically orthotopically implanted with a human coloncancer cell-line or colon-cancer liver metastases derived from patients. After liver tumor growth, mice received near-infrared conjugated anti-CEA or anti-CEACAM antibody to label the liver metastases. ICG was intravenously injected after ligation of the left or left lateral Glissonean pedicle resulting in specific labeling of the segment adjacent to the tumor with preserved blood-flow in the liver. Imaging was performed with the FLARE Imaging Systems. Results: The liver metastasis was brightly labeled with near infrared fluorescence with selective tumor targeting by the fluorescent anti-CEA or anti-CEACAM antibody, which was imaged on the 700 nm channel. The adjacent liver segment with preserved bloodflow in the liver had a bright fluorescence ICG 800 nm signal, while the left or left lateral segment had no fluorescence signal. Overlay of the images showed clear color-coded differentiation between the tumor and the liver segment, enabling image guided metastasectomy. Conclusions: Color coded imaging of the liver metastasis and adjacent liver segment in the present review can be used in the future for improved liver metastasectomy in the clinic.

Background/purpose. Multi-modality imaging is a major diagnostic component for cancer management, enabling detection, staging of disease, targeting therapy, and monitoring the therapeutic response. The development of a single agent for real-time non-invasive immunoPET imaging and fluorescence guided surgery (FGS) will provide the next generation tool in the surgical management of cancer. Methods. The humanized anti-CEA M5A-IR800 “sidewinder” (M5A-IR800-SW) antibody-dye conjugate has a NIR 800nm dye incorporated into a PEGylated linker and conjugated with the metal chelate p-SCN-Bn-deferoxamine (DFO) for zirconium-89 (⁸⁹Zr, half-life 78.4 h) PET imaging. The dual-labeled ⁸⁹Zr-DFO-M5A-SW-IR800 was evaluated for NIR fluorescence imaging, PET/MRI imaging, terminal tissue biodistribution and blood clearance in a human colorectal cancer LS174T xenograft mouse model. Results. The ⁸⁹Zr-DFO-M5A-SW-IR800 NIR fluorescence imaging showed high tumor and liver localization that was confirmed by PET imaging. Serial PET/MRI imaging was performed at 24 h, 48 h and 72 h and showed tumor localization visible at 24 h that persisted throughout the experiment. However, the PET scans showed higher activity for the liver than the tumor, compared to the NIR fluorescence imaging. This difference is an important finding as it quantifies the expected difference due to the sensitivity and depth of penetration between the 2 modalities.

Accidental damage of vital nerve structures remains a significant surgical morbidity. Patient-to-patient neuroanatomical variability requires considerable dependence on a surgeon’s first-hand experiences that primarily rely on proximal features for orientation, which can be further complicated in patients with nerve damage. As such, enhanced nerve visualization proves to be a vital avenue for advancing surgical precision and patient outcomes. Fluorescence guided surgery (FGS) has the potential to improve surgical guidance, but there are no current nerve-specific fluorophores approved for clinical use. Previous work has identified the oxazine scaffold as a promising avenue for nerve-specific contrast agent development, due to its sufficiently low molecular weight to cross the blood-nerve-barrier (BNB), tunable photophysical properties, and high nerve specificity. Herein we report our efforts to investigate the structure-function relationship of Oxazine-4 through fine-tuned terminal alkylamino modifications, both based on optical and physicochemical properties as well as their affected nerve specificities.

Rapid expansion in the field of fluorescence guided surgery (FGS) has yielded a wide range of contrast agent types under development with diverse characteristics. Currently, 105 active or completed clinical trials are registered in clinicaltrials.gov studying 39 unique novel FGS contrast agents. For these 39 contrast agents, there exists 6 distinct classes or types of probes: nanoparticle, antibody, protein, affibody, peptide, and small molecule. This diversity yields unique advantages and disadvantages among each type of contrast agent, which change throughout the various stages of development and clinical translation. In this review, we outline the relevant advantages and disadvantages for each type of FGS contrast agent at each stage of development. As the field continues to progress and expand, this diversity in FGS contrast agent type and their respective unique characteristics will enable broad applicability to shift the surgical paradigm and improve outcomes for patients across all surgical specialties.

Iatrogenic nerve injury remains one of the most common surgical complications, often resulting in permanent disabilities that severely impact patient quality of life following surgery. Current means of intraoperative nerve identification are limited beyond white light visualization and neuroanatomical knowledge but include ultrasound and the gold standard electromyography (EMG). However, nerve identification in the surgical field of view often remains inadequate. Though fluorophores like rhodamine, cyanine, and others have found extensive and diverse uses in the life sciences, in the realm of fluorescence-guided surgery (FGS), fluorophores that absorb and emit in the NIR region (650-900 nm) have the highest potential for clinical translation. Combining the structural characteristics of a long wavelength emitting fluorophore cyanine like indocyanine green (ICG) with those of a topically nerve-specific fluorophore, like rhodamine B, could offer a strategy for generating NIR-emissive and nerve-specific fluorophores. This study investigated whether the topical nerve-affinity observed in rhodamines extends to systemic administration and whether the structural hybridization strategy used in the previously published Changsha dyes could prove useful in generating long-wavelength nerve-specific contrast agents for use in FGS.

Curative surgery for other many cancers requires that the tumor be removed with a zone of normal tissue surrounding the tumor with ‘negative’ margins. Sarcomas, cancers of the bones, muscles, and fat, require WLE for cure. Unfortunately, ‘positive’ margins occur in 20-25% of sarcoma surgeries, associated with cancer recurrence and reduced survival. Our group successfully tested a small-molecule fluorophore (ABY-029) in sarcomas that targets the epidermal growth factor receptor. We sought to evaluate human sarcoma xenografts for epidermal growth factor receptor expression and binding of ABY-029 with and without exposure to standard presurgical chemotherapy and radiation. We inoculated groups of 24 NSG mice with five cell lines (120 mice total). Eight mice from each cell line received: 1) radiation alone; 2) chemotherapy alone; or 3) chemotherapy and radiation. We administered ABY-029 2-4 hours before surgery. Tumor and biopsy portions of background tissues were removed. All tissues were imaged on a LI-COR Odyssey and processed in pathology. There were no significant reductions in epidermal growth factor receptor expression or in ABY-029-mediated fluorescence in tumors exposed to chemotherapy, radiation, or both. fluorescence-guided surgery demonstrates strong promise to improve curative surgical cancer care, particularly for sarcomas where the positive margin rate is substantial. Fluorophore performance must be evaluated under circumstances that duplicate accurately the biological milieu relevant to a particular cancer. This work shows that human sarcoma xenografts subjected to standard therapies do not demonstrate a change in epidermal growth factor receptor expression or in epidermal growth factor receptor-targeted fluorescence, thereby indicating that epidermal growth factor receptor-targeted fluorescence-guided surgery should be feasible under normal therapeutic conditions in the clinic.

ICG-based dynamic contrast-enhanced fluorescence imaging (DCE-FI) and intraoperative DCE- magnetic resonance imaging (MRI) have been carried out nearly simultaneously in three lower extremity bone infection cases to investigate the relationship between these two imaging modalities for assessing bone blood perfusion during open orthopedic surgeries. Time-intensity curves in the corresponding regions of interest of two modalities were derived for comparison. The results demonstrated that ICG-based DCE-FI has higher sensitivity to perfusion changes while DCE-MRI provides superior and supplemental depth-related perfusion information. Research applying the depth-related perfusion information derived from MRI to improve the overall analytic modeling of intraoperative DCE-FI is ongoing.

Optimal differentiation between tumor and normal tissues using epidermal growth factor receptor targeted fluorescence guided surgery (FGS) of head and neck cancer (HNC) is complicated by the presence of target receptor in the normal surrounding tissues. We propose the use of radiomics feature analysis to increase the accuracy and efficiency of tumor tissue discrimination based on machine-learning algorithms. Radiomics analysis demonstrates that radiomics analysis reaches a higher identification performance than the traditional intensity threshold method in the preclinical mice. This study proposes that a radiomics approach for fluorescence imaging in preclinical studies is a more accurate tissue type identification method requiring less post-agent-administration waiting time than the traditional fluorescence intensity threshold method.

The field of florescence guided surgery (FGS) has been growing rapidly since the initial market clearance of the SPY SP2000 in 2015. Many of the currently approved exogenous fluorophores have existed for half a century, yet adoption utilizing their florescence properties have only come to fruition in the past two decades. Now, a number of new nearinfrared (NIR) contrast agents are poised to reach the market, pushing the limits of FGS beyond perfusion imaging into the realm of molecularly targeted contrast. The shear number and combinations of imaging systems and fluorophores will become increasingly burdensome for regulatory reviewers if standardized approaches for system characterization are not implemented. In 2017 the American Association of Physicists in Medicine (AAPM) convened Task Group 311 (TG311) to consider standardization criteria, but the implementation may prove difficult. Standardization tools will need to be developed and utilized to implement the recommendations by TG311. The progress of implementing standardized phantoms, augmented by computational aides in the form of image analysis and simulation packages are presented as a means to address future standardization efforts.

Cherenkov-excited luminescence scanned tomography (CELST) is an emerging tomographic optical imaging modality. However, recovering spatial distribution of luminescent source from boundary measurements is a typically ill-posed problem. To improve the performance of CELST reconstruction, an end-to-end reconstruction algorithm is developed by combining dilated convolution and attention mechanism based on Unet (DA-Unet). Its performance is validated with numerical simulations. The results reveal that DA-Unet has superior reconstruction performance with high spatial resolution. It achieves image quality with PSNR of more than 35 dB and SSIM of larger than 0.95. Furthermore, the DAUnet can reconstruct luminescent source even with less boundary measurements.

Cherenkov-excited luminescence scanned imaging (CELSI) is a new emerging imaging modality, which uses linear accelerator (LINAC) to induce Cherenkov radiation, and then secondary excite molecular probes to produce luminescence. The tomographic distribution of the molecular probes can be recovered by a reconstruction algorithm. However, the reconstruction images usually suffer from many artifacts. To improve the image quality for tomographic reconstruction, we propose a reconstruction method based on learned KSVD. Numerical simulation experiments reveal that the proposed algorithm can reduce the artifacts in the reconstructed image. The quantitative results show that the structured similarity (SSIM) is improved more than 8.8% compared to the existing algorithms. In addition, our results also demonstrate that the proposed algorithm has the best performance under different noise levels (0.5%, 1%, 2%, and 4%).

In orthopedic trauma surgery, timely assessment of bone tissue perfusion plays a vital role in the successful treatment outcome. Fluorescence-guidance is gaining increased surgical interest, especially with respect to hemodynamic assessment of bone. Intraoperative dynamic contrast-enhanced fluorescence imaging (DCE-FI) not only enables visualization of the perfused areas of the injured bone, but with subsequent analysis using kinetic models, may also provide a valuable quantitative bone blood flow information to a surgeon. In this study, we are validating this quantitative approach with a modified fluorescent microsphere (FM) technique using a custom-built four-channel imaging cryomacrotome. We demonstrate that FMs of four different colors can be accurately detected in controlled phantoms and evaluate their detection accuracy in real blood samples. In a rabbit model of orthopaedic trauma, we show that blood flow measurements using the DCE-FI technique can be compared with the FM technique. This feasibility pilot study provides the groundwork for investigation of the correlation between bone perfusion measurements using DCE-FI and using fluorescent microspheres, in units of ml/min/100g.