3D-Printed Tumor Phantoms for Assessment of In Vivo Fluorescence Imaging Analysis Methods.

PURPOSE: Interventional fluorescence imaging is increasingly being utilized to quantify cancer biomarkers in both clinical and preclinical models, yet absolute quantification is complicated by many factors. The use of optical phantoms has been suggested by multiple professional organizations for quantitative performance assessment of fluorescence guidance imaging systems. This concept can be further extended to provide standardized tools to compare and assess image analysis metrics. PROCEDURES: 3D-printed fluorescence phantoms based on solid tumor models were developed with representative bio-mimicking optical properties. Phantoms were produced with discrete tumors embedded with an NIR fluorophore of fixed concentration and either zero or 3% non-specific fluorophore in the surrounding material. These phantoms were first imaged by two fluorescence imaging systems using two methods of image segmentation, and four assessment metrics were calculated to demonstrate variability in the quantitative assessment of system performance. The same analysis techniques were then applied to one tumor model with decreasing tumor fluorophore concentrations. RESULTS: These anatomical phantom models demonstrate the ability to use 3D printing to manufacture anthropomorphic shapes with a wide range of reduced scattering (µs': 0.24-1.06 mm-1) and absorption (µa: 0.005-0.14 mm-1) properties. The phantom imaging and analysis highlight variability in the measured sensitivity metrics associated with tumor visualization. CONCLUSIONS: 3D printing techniques provide a platform for demonstrating complex biological models that introduce real-world complexities for quantifying fluorescence image data. Controlled iterative development of these phantom designs can be used as a tool to advance the field and provide context for consensus-building beyond performance assessment of fluorescence imaging platforms, and extend support for standardizing how quantitative metrics are extracted from imaging data and reported in literature.

Theoretical lateral and axial sensitivity limits and choices of molecular reporters for Cherenkov-excited luminescence in tissue during x-ray beam scanning.

PURPOSE: Unlike fluorescence imaging utilizing an external excitation source, Cherenkov emissions and Cherenkov-excited luminescence occur within a medium when irradiated with high-energy x-rays. Methods to improve the understanding of the lateral spread and axial depth distribution of these emissions are needed as an initial step to improve the overall system resolution. METHODS: Monte Carlo simulations were developed to investigate the lateral spread of thin sheets of high-energy sources and compared to experimental measurements of similar sources in water. Additional simulations of a multilayer skin model were used to investigate the limits of detection using both 6- and 18-MV x-ray sources with fluorescence excitation for inclusion depths up to 1 cm. RESULTS: Simulations comparing the lateral spread of high-energy sources show approximately 100 × higher optical yield from electrons than photons, although electrons showed a larger penumbra in both the simulations and experimental measurements. Cherenkov excitation has a roughly inverse wavelength squared dependence in intensity but is largely redshifted in excitation through any distance of tissue. The calculated emission spectra in tissue were convolved with a database of luminescent compounds to produce a computational ranking of potential Cherenkov-excited luminescence molecular contrast agents. CONCLUSIONS: Models of thin x-ray and electron sources were compared with experimental measurements, showing similar trends in energy and source type. Surface detection of Cherenkov-excited luminescence appears to be limited by the mean free path of the luminescence emission, where for the given simulation only 2% of the inclusion emissions reached the surface from a depth of 7 mm in a multilayer tissue model.

Indocyanine green matching phantom for fluorescence-guided surgery imaging system characterization and performance assessment.

SIGNIFICANCE: Expanded use of fluorescence-guided surgery with devices approved for use with indocyanine green (ICG) has led to a range of commercial systems available. There is a compelling need to be able to independently characterize system performance and allow for cross-system comparisons. AIM: The goal of this work is to expand on previous proposed fluorescence imaging standard designs to develop a long-term stable phantom that spectrally matches ICG characteristics and utilizes 3D printing technology for incorporating tissue-equivalent materials. APPROACH: A batch of test targets was created to assess ICG concentration sensitivity in the 0.3- to 1000-nM range, tissue-equivalent depth sensitivity down to 6 mm, and spatial resolution with a USAF test chart. Comparisons were completed with a range of systems that have significantly different imaging capabilities and applications, including the Li-Cor® Odyssey, Li-Cor® Pearl, PerkinElmer® Solaris, and Stryker® Spy Elite. RESULTS: Imaging of the ICG-matching phantoms with all four commercially available systems showed the ability to benchmark system performance and allow for cross-system comparisons. The fluorescence tests were able to assess differences in the detectable concentrations of ICG with sensitivity differences >10× for preclinical and clinical systems. Furthermore, the tests successfully assessed system differences in the depth-signal decay rate, as well as resolution performance and image artifacts. The manufacturing variations, photostability, and mechanical design of the tests showed promise in providing long-term stable standards for fluorescence imaging. CONCLUSIONS: The presented ICG-matching phantom provides a major step toward standardizing performance characterization and cross-system comparisons for devices approved for use with ICG. The developed hybrid manufacturing platform can incorporate long-term stable fluorescing agents with 3D printed tissue-equivalent material. Further, long-term testing of the phantom and refinements to the manufacturing process are necessary for future implementation as a widely adopted fluorescence imaging standard.

Probe-based fluorescence dosimetry of an antibody-dye conjugate to identify head and neck cancer as a first step to fluorescence-guided tissue preselection for pathological assessment.

BACKGROUND: Despite the rapid growth of fluorescence imaging, accurate sampling of tissue sections remains challenging. Development of novel technologies to improve intraoperative assessment of tissue is needed. METHODS: A novel contact probe-based fluorescence dosimeter device, optimized for IRDye800CW quantification, was developed. After evaluation of the device in a phantom setup, its clinical value was defined ex vivo in patients with head and neck squamous cell carcinoma who received panitumumab-IRDye800CW. RESULTS: Ten patients were enrolled with a total of 216 data points obtained. Final histopathology showed tumor in 119 spots and normal tissue in 97 spots. Fluorescence-to-excitation ratios in tumor tissue were more than three times higher than those in normal tissue. The area under the curve was 0.86 (95% CI: 0.81-0.91) for tumor detection. CONCLUSIONS: Fluorescence-guided tissue preselection using a fluorescence dosimeter could have substantial impact on tissue sampling for frozen section analysis and potentially reduce sampling errors.

Modeling PpIX effective light fluence at depths into the skin for PDT dose comparison.

BACKGROUND: Daylight-activated PDT has seen increased support in recent years as a treatment method for actinic keratosis and other non-melanoma skin cancers. The inherent variability observed in broad-spectrum light used in this methodology makes it difficult to plan and monitor light dose, or compare to lamp light doses. METHODS: The present study expands on the commonly used PpIX-weighted effective surface irradiance metric by introducing a Monte Carlo method for estimating effective fluence rates into depths of the skin. The fluence rates are compared between multiple broadband and narrowband sources that have been reported in previous studies, and an effective total fluence for various treatment times is reported. A dynamic estimate of PpIX concentration produced during pro-drug incubation and treatment is used with the fluence estimates to calculate a photodynamic dose. RESULTS: Even when there is up to a 5x reduction between the effective surface irradiance of the broadband light sources, the effective fluence below 250 µm depth is predicted to be relatively equivalent. An effective threshold fluence value (0. 70Jeff/cm2) is introduced based on a meta-analysis of previously published ALA-PpIX induced cell death. This was combined with a threshold PpIX concentration (50 nM) to define a threshold photodynamic dose of 0.035 u M Jeff/cm2. CONCLUSIONS: The threshold was used to generate lookup tables to prescribe minimal treatment times to achieve depth-dependent cytotoxic effect based on incubation times and irradiance values for each light source.