Добавил:
kiopkiopkiop18@yandex.ru t.me/Prokururor I Вовсе не секретарь, но почту проверяю Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:

Ординатура / Хирургия / Библиотека им академика М.И. Перельмана / Книга_3835_Библиотеки_им_академика_М_И_Перельмана

.pdf
Скачиваний:
0
Добавлен:
15.09.2026
Размер:
13 Мб
Скачать
☆
6 Dual-Modality Fluorescence Lifetime and Intravascular … 155
https://t.me/medicina_free
Fluorescence measurements from en face frozen sections of human coronary arteries and aorta were acquired with an Olympus BH-2 metallurgic bright-field microscope and an argon ion laser (476 nm excitation) (Fitzmaurice et al. 1989). The fluores­cence properties of collagen, elastin, lipid, and calcium allowed for normal arteries and atherosclerotic arteries to be differentiated. Building on this work, Verbunt et al. more carefully studied the fluorescence of lipid and calcium-rich plaques by ana­lyzing ceroid deposits in these specimens (Verbunt et al. 1992). Ceroid, a type of lipopigment, was found to have an autofluorescence spectrum that was wider and red-shifted in comparison with both collagen and elastin. In 1994, a study was done that showed autofluorescence spectra could distinguish fibrous, aneurysmal, and nor­mal samples but not lipid and calcium-rich plaques. This work was performed on samples of the abdominal aorta from cadavers and femoral artery from bypass surg­eries with an He-Cd laser (442 nm excitation) (Papazoglou et al. 1994). In 2000, a fluorescence spectroscopy study was performed in vivo to observe plaque disruption in a rabbit model (Christov et al. 2000).
During the 1990s, the characteristics of plaque that indicate higher levels of risk for causing heart attack were meticulously studied (Virmani et al. 2006). Subse­quently, studies that focused on using fluorescence to assess features of plaque vul­nerability were initiated. The first study to use fluorescence to look for a thin-capped fibroatheroma was performed by Arakawa et al. in human femoral and coronary arter­ies obtained from cadavers (Arakawa et al. 2002). Autofluorescence was induced with a He-Cd laser (excitation 325 nm); autofluorescence properties of collagen, elastin, and lipid were compared to the measurements from the human tissue to determine what fluorophores were fluorescing in the tissue samples. Plaques with lipid cores had fluorescence spectra that were red-shifted and broader compared to normal tissue and collagenous plaques. Lipid core fluorescence properties were similar to those of oxidized low-density lipoprotein (oxLDL), indicating the presence of this molecule in lipid cores. The fluorescence properties of fibrous plaques were similar to collagen and those of normal artery were similar to elastin. A classification algorithm was able to use the fluorescence spectra from these molecules (collagen, elastin, oxLDL) to distinguish normal artery, lipid cores, atheroma, or pre-atheroma with 86% accu­racy. This work also showed that fibrous cap thickness correlated well with spectral collagen content (R = 0.65, P < 0.0001) (Arakawa et al. 2002).
Time-Resolved Fluorescence Spectroscopy and Imaging Studies of Atherosclerotic Lesions
Point Measurements of Time-Resolved Fluorescence Spectroscopy
Single wavelength measurements Due to the increased ability of time-resolved
fluorescence spectroscopy (TRFS) to resolve fluorophores with overlapping fluores­cence intensity spectra, these studies were used to determine if they could more
156 J. E. Phipps et al.
https://t.me/medicina_free
accurately identify plaque composition than fluorescence intensity spectroscopy alone. The TRFS point measurement technique involved for this work consisted of exciting at a single wavelength and recording the time-resolved fluorescence decay of the emitted light at a single wavelength band. Fluorescence emission intensity from normal and atherosclerotic aorta was found to differ at 340 and 380 nm with 308 nm ultraviolet laser excitation (Baraga et al. 1989). A binary classification scheme was used to distinguish normal from the atherosclerotic aorta in 56 of 60 total cases by using these wavelength bands that characterized the tryptophan content (340 nm) and the elastin-to-collagen ratio (380 nm) (Baraga et al. 1990). Part of this study also used a time-correlated single photon counting technique (TCSPC). With this technique, fluorescence lifetime was shorter in the tryptophan region (340 nm) and longer at the collagen/elastin region (380 nm) (Baraga et al. 1989). A frequency­doubled mode-locked and cavity-dumped continuous wave dye laser for picosecond pulse generation at 320 nm was used by Andersson-Engels et al. with TCSPC to acquire time-resolved signal from the human aorta samples in vitro. Both collage­nous and calcified plaques exhibited longer fluorescence lifetimes than normal artery at 400 and 480 nm (Andersson-Engels et al. 1990). Several studies by Maarek et al. reported the spectrally resolved time-resolved fluorescence spectroscopy character­ization of arterial wall constituents, including collagen, elastin, and lipid as well as characterizing specific lesion types (Maarek et al. 1997, 1998; Marcu et al. 1998).
Multiple wavelength measurements Following the TRFS studies that interro­gated single wavelength bands of emitted fluorescence light per measurement, newer studies employed a technique that allowed for multiple bands of emitted fluorescence to be temporally resolved. These newer studies could be faster and provide more information than earlier TRFS studies. Human aortic samples were first studied with this technique (94 samples, postmortem) (Maarek et al. 2000; Marcu et al. 2003). Excitation was performed with a 337 nm nitrogen laser, and emission was collected through silica fibers combined in a single probe. Wavelength selection occurred with a scanning monochromator, a gated micro-channel plate photomultiplier was used to measure the emission waveform, and sampling was performed with a digitiz­ing oscilloscope. The intrinsic fluorescence decays were computed with a Laguerre deconvolution method, from which the time-integrated fluorescence emission spec­trum and fluorescence lifetime were derived. TRFS spectra from the aortic samples varied with the progression of atherosclerosis–fluorescence intensity and average lifetime. The substantial differences in time-resolved properties between clinically relevant plaque compositions suggested that TRFS parameters could translate to arterial disease diagnostic markers. TRFS was also found sensitive to macrophage infiltration in plaques (Marcu et al. 2005) and the presence or absence of matrix met­alloproteinases (Phipps et al. 2011). A TRFS study was also conducted in vivo in an atherosclerotic rabbit model, demonstrating that TRFS could discriminate between macrophage and collagen content in atherosclerotic plaques (Marcu et al. 2005).
A similar system as described above was also used to study time-resolved fluores­cence emission of normal and atherosclerotic human coronary arteries (58 coronary segments from 11 subjects, postmortem) (Marcu et al. 2001, 2003). Similar to the study in the human aorta, this work showed parameters derived from time-resolved
6 Dual-Modality Fluorescence Lifetime and Intravascular … 157
https://t.me/medicina_free
fluorescence spectra could be used to enhance discrimination between normal and different grades (I, II, III, IV, Va, and Vb) of atherosclerotic lesions as defined by the American Heart Association (AHA). Lipid-rich lesions were distinguishable from other lesion types, particularly fibrous lesions and normal artery wall. The fluo­rescence emission of lipid-rich lesions matched that of lipid components and the fluorescence emission of fibrous lesions matched that of collagen type I. This study demonstrated that time-resolved spectra varied significantly between lipid-rich (more unstable) and fibrous lesions (more stable), further validating the potential impact TRFS could have if implemented intravascularly.
Fluorescence Lifetime Imaging Microscopy (FLIM) Studies
Fluorescence lifetime imaging microscopy (FLIM) allows for time-resolved spectra to be acquired for an image versus a single point. An early fluorescence lifetime imag­ing microscopy (FLIM) system with 10 ps temporal resolution was introduced for biomedical applications in 1998. The system acquired fluorescence lifetime images to be acquired through excitation with a commercial Ti: Sapphire laser (415 nm), a time-gated image intensifier, and an intensified CCD camera (Dowling et al. 1998). Lifetime differences were found between elastin, collagen, and aorta samples. More recent advances in FLIM technology revolutionized the potential of this technique positive impact in clinically relevant biomedical applications (De Beule et al. 2007; Munro et al. 2005).
In 2011, atherosclerotic aortas were studied with a FLIM system implemented in a wide-field fiber bundle (Phipps et al. 2011). N = 11 human aorta samples (N = 48 locations) were imaged. A pulsed nitrogen laser was used for excitation (337 nm, 700 ps width), an imaging bundle with 10,000 optical fibers in a 0.6 mm outer diame­ter probe was used to collect fluorescence, and a fast-gated (up to 400 ps) intensified CCD camera generated images. Two emission wavelength bands were detected, F377: 377/50 nm and F460: 460/60 nm (center wavelength/bandwidth). This combi­nation of wavelengths provided discrimination between intrinsic fluorophores related to plaque vulnerability. Different tissue types were identified with average lifetime and Laguerre deconvolution parameters (Jo et al. 2005). N = 4 distinct regions of interest (ROIs): Lipid-rich (LR), collagen-rich (CR), elastin-rich (ER), and elastin + macrophage-rich (E + M) were identified based on histopathology from N = 81 regions of interest (ROIs) within the FLIM images. FLIM-derived parameters were distinct in each group.
Another FLIM investigation with a wide-field fiber bundle was used to study ex vivo human coronary arteries and found discrimination between collagen-rich plaques and normal (elastin-rich) artery (Thomas et al. 2010). However, this study did not include lipid-rich plaques, and thus, the extensive comparison with TRFS results could not be made.
158 J. E. Phipps et al.
https://t.me/medicina_free
Wavelength-Multiplexed Fluorescence Lifetime Imaging (FLIm) of Planar Samples
FLIM enables the acquisition of whole images of the field of view, but r elies on bulky and stiff fiber bundles. Multispectral TRFS measurements can be performed using a single fiber for excitation and collection, but the slow acquisition time is prohibitive for the acquisition of images. Here, we present a fluorescence lifetime imaging (FLIm) method that relies on a single fiber for excitation and collection and is able to acquired data at high speed (Sun et al. 2011). This approach was initially reported in 2008 (Sun et al. 2008) and allowed simultaneous acquisition and stor­age of fluorescence data from three wavelength bands in a few microseconds. When coupled with a compact x-y-z 3D positioning stage, this system also allowed the collection of point measurements from these wavelength bands in a fast-scanning mode that could then be reconstructed to create fluorescence lifetime imaging (FLIm) images (Sun et al. 2011). This scanning system was validated in ex vivo atheroscle­rotic aortas (Sun et al. 2011). Using the fast-scanning system, FLIm images were created that demonstrated the distinction between plaque types. A similar system has also been incorporated within optical coherence tomography (OCT) for arterial imaging (Jo et al. 2015). We note here the difference between FLIM and FLIm. The FLIm technique described here acquires multiple wavelengths of emitted fluores­cence simultaneously through a single fiber in contrast to the FLIM technique that acquires high-resolution images of the emitted fluorescence (microscopy) at a single wavelength band per measurement. Moving forward, we will show how this single fiber, high-speed technique has been further developed to allow for the technology to be translated to an intravascular catheter.
Challenges for Implementing FLIm in an Intravascular Catheter
This chapter will discuss the necessary qualities of a catheter-based FLIm system and the challenges to achieving these specifications that include:
• Fast-scanning speed to account for the fact that blood will need to be flushed from
the field of view, which can only be done for short amounts of time for patient safety
• Pairing with a morphologic imaging modality that will guide FLIm measurements
from within the coronary artery tree
• A flexible fiber that is able to reach the coronary arteries and acquire images from
tortuous vessels.
FLIm excitation and signal cover the near-UV to visible range. Over this wavelength range, blood presents strong absorption, it is therefore necessary to displace blood from the field of view during imaging. In a similar way to OCT, this can be achieved
6 Dual-Modality Fluorescence Lifetime and Intravascular … 159
https://t.me/medicina_free
using a balloon located proximally to the imaging catheter field of view in combi­nation of i njection of clear fluid, or bolus injection of clear fluid without occlusion. In both cases, this restricts the imaging duration as blood flow interruption or fluid injection duration need to be limited to a few seconds. A system able to acquire FLIM images in vivo must therefore be able to measure locations with speeds in the kHz range.
Coronary arteries, in particular when diseased, present a small cross section and notable tortuosity.Successful navigation through lesions places a constraint on device cross section and flexibility. Additionally, FLIm is not depth-resolved and only pro­vides information down to 200 µm from the lumen surface. To provide navigational information (lumen dimension, localization in the arterial tree) and structural and anatomical information (plaque burden, presence of necrotic or calcified cores, etc.), it needs to be combined with a morphological imaging technique such as IVUS. While the biochemical information provided by FLIm is ideally complementing the morphological information provided by IVUS, integration of FLIm with IVUS cre­ates constraints for the design of the instrument. Namely, both FLIm and IVUS elements need to be integrated into a low profile catheter, and the imaging section needs to be both optically transparent in the near-UV to visible and present low sound impedance in addition to being flexible.
FLIm-IVUS Catheter System Instrumentation
The FLIm instrumentation dedicated to intracoronary imaging is composed of a mul­tispectral high-speed time-domain fluorescence lifetime imaging system, integrated with a clinical 40 MHz IVUS system. The design of the FLIm instrumentation as well as its integration into a bimodal catheter system is described in the following sections.
Principle of the High-Speed Fluorescence Lifetime Imaging Instrumentation
The FLIm instrumentation (see Fig. 6.1) consists of a pulsed UV fiber laser (20 kHz, 80 ps pulses, Fianium, UK), a wavelength selection module (WSM) that directs the laser light from the laser to the sample via the excitation/collection fiber optic. Autofluorescence generated by the sample is then collected by the same fiber and transmitted to the WSM where it is split into four different bands matched with the emission wavelengths of fluorophores expected in diseased and healthy arteries.
Optical temporal multiplexing is performed by routing the signal from each band through different optical delay lines (1, 10, 19, and 28 m). The 45 ns delay between channels introduced by the fibers enables fluorescence contribution from each of
160 J. E. Phipps et al.
https://t.me/medicina_free
Fig. 6.1 Fast FLIm principle of operation: A pulsed UV source delivers excitation light to the vessel wall via a fiber optic channel (100–200 µm core silica fiber) integrated inside of the FLIm- IVUS motor drive and catheter. Autofluorescent light generated by the arterial wall is collected and transmitted via the same fiber and separated in four different wavelength bands. Temporal multiplexing achieved by transmitting the contribution of each channel to the photodetector using different length of fiber optic enables sampling of all four bands using a single channel 12.5 GS/s digitizer. For each excitation pulse, full fluorescence decays over four different wavelengths are acquired
the four spectral bands to be sent to a single multi-channel plate photomultiplier (R3809U-50, Hamamatsu, Japan), without overlap of the signals. The electrical sig­nal is amplified (C5594 Hamamatsu, Japan) and sampled using a high-speed digitizer (12.5 GS/s, 3 GHz, 8-bit, 512 Mbytes, PXIe-5185, National Instruments, Austin, TX).
This detection method enables acquisition of fluorescence decay from multiple wavelength bands from a single excitation pulse, therefore enabling much higher acquisition speeds than alternative fluorescence lifetime methods (Stary et al. 1994). Successful measurements of fluorescence decay using a pulse sampling technique rely on a high enough number of photons reaching the photodetector following each excitation pulse. Therefore, excitation pulse energy, throughput of the excitation and collection optical paths, and width of each spectral band of the instrument need to be optimized.
FLIm-IVUS Intravascular Catheter System
The FLIm system described above is ideally suited to the implementation of FLIm in the intravascular setting since it is able to work at high speed and requires only a single multimode fiber for fluorescence excitation and collection. Nevertheless,
6 Dual-Modality Fluorescence Lifetime and Intravascular … 161
https://t.me/medicina_free
challenges to implement intravascular FLIm highlighted previously, coupled with the high collection efficiency required for successful implementation of the pulse sampling method lead to several engineering challenges. The design of a FLIm­IVUS catheter system suitable for in vivo interrogation of coronary arteries required several design iterations until a configuration able to fulfill all requirements was identified.
The first design, characterized by IVUS and FLIm channels arranged in a parallel configuration, allowed for independent rotation of the ultrasonic and optical channels (using independent motor drives) within an 8 Fr outer diameter catheter sheath (Bec et al. 2012). No significant modification of the commercial IVUS system/catheter was required. The ultrasound imaging core was still introduced in the original IVUS sheath, while the FLIm fiber was housed in an acrylic tube. With this catheter, the twin tube imaging section was enclosed in a compliant siloxane balloon that was inflated with saline solution during imaging and displaced the blood from the field of view. This catheter enabled the initial technical performance evaluations of the rotational FLIm system and the first in vivo measurements in arterial vessels autoflu­orescence in pulsatile blood flow in swine peripheral arteries (Bec et al. 2014). In this configuration, the scanning pullback length was limited to 6 mm.
This design was further refined in a subsequent design using a parallel/sequential configuration (Design 2). It consisted of a catheter with a small cross-sectional profile (3.2 Fr) relying on sequential scanning of the field of view by parallel independent FLIm and IVUS imaging cores, advancing sequentially into a single sheath and imag­ing section (Ma et al. 2015) by a dedicated hardware and software interface. The use of polymethylpentene (TPX med-18) for the imaging section provided both high transparency and low autofluorescence in the near-UV-visible range, required for FLIm operation, and low sound impedance, required for IVUS operation. Similarly to Design 1, no significant modification of the commercial IVUS system/imaging core was required. This catheter enabled us to demonstrate for the first time the ability of our FLIm system to rapidly acquire spectrally resolved fluorescence life­time data in helical scanning from tubular and tortuous structures (including ex vivo porcine coronary arteries in intact hearts) (Ma et al. 2014). Moreover, this design enabled the generation of experimental data from diseased human coronary arter­ies supporting the development of methods for IVUS image segmentation, FLIm data acquisition under IVUS guidance, and methods/algorithms for plaque diagnosis (Fatakdawala et al. 2015; Gorpas et al. 2015). However, in this configuration co­registration between FLIm and IVUS is only achieved in the absence of motion, and thus, this was not suitable for in vivo use (Ma et al. 2015).
These limitations were addressed in the third design of the instrument, reported in (Bec et al. 2017). It was built on the knowledge gained from the two earlier designs and consisted of a catheter that fully integrated FLIm imaging capability
TM
into an existing clinical IVUS platform (iLab
Boston Scientific). A custom 3.7 Fr monorail bimodal imaging catheter includes both a 40 MHz ultrasound transducer and optical channel in tandem configuration (see Fig. 6.2). Additionally, the IVUS motor drive was modified to accommodate an optical channel. Thus, both the 40 MHz transducer and the fiberoptic are rotated by the same motor drive. This design enabled
162 J. E. Phipps et al.
https://t.me/medicina_free
Fig. 6.2 Design 1 (Parallel): enabled co-registered FLIm-IVUS data acquisition in vivo in swine peripheral arteries. Design 2 (Sequential): allowed for co-registered FLIm-IVUS data acquisition in ex vivo excised human coronaries but co-registration is susceptible to motion artifacts. Design 3 (Integrated): FLIm and IVUS fully integrated into a single 3.7 Fr imaging core. With Boston Scientific support, their IVUS motor drive (MDU5 PLUS) was modified to incorporate an optical channel. This system was used for co-registered FLIm-IVUS data acquisition in vivo i n swine coronary arteries as well as ex vivo diseased human coronary arteries
the acquisition of directly co-registered FLIm-IVUS data in a single pullback of the catheter. With this system, 25,000 independent point spectroscopy measurements are acquired in 5 s, using an 1800 rpm rotation speed, and 4 mm/s pullback speed. The 20 mm axial field of view can be increased by using higher pullback speeds, thus increasing axial sampling. This novel catheter was found as being fully compatible with in vivo coronary interventions (Bec et al. 2016).
Methods for Data Processing
FLIm data were processed using a Laguerre expansion technique (Upston et al. 2002) due to the speed of this nonparametric technique and its ability to perform decon­volution without a priori information about the fluorescence emission. The acquired waveform of each channel was fitted with a set of Laguerre functions convolved with the instrument response function. The weights obtained from the fitting operation were then applied to the original Laguerre basis functions (without IRF convolution) to estimate the original fluorescence decay. The average fluorescence lifetime was then derived from this decay. For each measurement, the average lifetime and fluores­cence intensity were computed and maps were generated where the location of each measurement corresponds to a distinct angular pullback location of the imaging core of the catheter. IVUS data were also processed to identify the vessel lumen (Spite and Serhan 2011) and to estimate the distance between catheter and vessel wall.
6 Dual-Modality Fluorescence Lifetime and Intravascular … 163
https://t.me/medicina_free
The distance map was used to correct the FLIm intensity images from variations of light excitation/collection due to changes in probe-to-wall distance (Angheloiu et al.
2011). This step enables quantification of the amount of fluorescence emitted in each
location. For each channel, intensity ratio maps were created by normalizing each pixel of the intensity images of each channel by the summed intensities of the same pixel across all four channels. Intensity-weighted lifetime maps were then created where colors represent lifetime values, while brightness represents the amount of fluorescence. Spectral ratio weighted lifetime maps, where color represents lifetime while brightness represents spectral contribution, were created. This representation contains both time-resolved and spectral information and is independent of changes in excitation/collection efficiency as they are inherently ratiometric. Finally, fluores­cence lifetime information was projected on the vessel surface recovered from IVUS to generate 3D renderings of the fluorescence lifetime maps.
FLIm-IVUS Imaging of In Vivo Swine Coronary Arteries
The ability of the catheter system to access tortuous anatomy, flush blood from the imaged location, image varioussized vessels, recover the signal with sufficient signal­to-noise ratio, and discriminate targets based on lifetime and spectral information was tested in 60–70 kg swine, which present morphology similar to humans.
Flushing was performed through the catheter guide using a power injector for precise control of the injection parameters. 10% LMD solution (Dextran 40) with 5% glucose was used as a flushing medium. This solution, evaluated for human use as an alternative to iodinated contrast for OCT flushing (Ozaki et al. 2012; Frick et al.
2014), is well suited to FLIm imaging due to its high transparency in the near-UV,
low toxicity, and high viscosity. Flushing flows between 2 and 10 cc/s were evaluated. It was observed that flow above 4 cc/s was sufficient to provide adequate clearing.
The ability of the system to recover lifetime information from the vessel wall in vivo was demonstrated by imaging a healthy section of coronary, expected to have a homogeneous composition, and computing the average lifetime (Bec et al. 2017). In spite of large variations of the amount of signal collected for different regions of the field of view, a low standard deviation was obtained (5.19 ± 0.12 ns). Subsequently, a stent with fluorescent markers was placed and imaged. Both spectral and temporal information enabled easy identification of the marker locations, as demonstrated in Fig. 6.3. Finally, the ability to recover lifetime measurements was evaluated in larger vessels (5 mm diameter). In that case, an increase in standard deviation of the measured lifetime was observed compared to smaller vessels, but the correct lifetime was still retrieved (5.14 ± 0.33 ns).
164 J. E. Phipps et al.
https://t.me/medicina_free
Fig. 6.3 View of large animal surgical facility at UC Davis (top left): Boston Scientific IVUS interface provides real-time IVUS display and enables identification of areas of interest for bimodal imaging. A commercial Medrad deliver 10% LMD (dextran 40) in 5% dextrose solution required to displace blood during bimodal acquisition (4–6 cc/s, 8 s). Close-up of stent deployed in coronary vessels: Fluorescent painted sections on stent struts were used to provide contrast in healthy pig coronary arteries. Bimodal B-scans (from one in vivo image cross section) showing easily differentiated fluorescence signal from painted section of stent and artery wall fluorescence between painted section. Full spectral and average lifetime data: 20 mm section imaged in 5 s. Both spectral and lifetime data sets enable straightforward identification of two marker sets from arterial wall (Bec et al. 2014)
®
Ava n taTMfluid management system (power injector) is used to
FLIm-IVUS Imaging of Ex Vivo Human Coronary Arteries
To image ex vivo human samples, coronary arteries are imaged in circulating heated (37 °C) PBS using the FLIm-IVUS catheter system. IVUS and FLIm data sets are co-registered by shared time stamps during acquisition. IVUS data are processed twofold: (1) RF data are used to extract spectral parameters such as integrated backscatter and energy norm, and (2) gray-scale images are segmented to extract lumen and media boundaries. FLIm data are deconvolved using Laguerre-based techniques (Liu et al. 2012;Suetal.2016) and for each spectral channel, inte- grated intensity, average lifetime, and Laguerre coefficients are computed. FLIm and IVUS data are then jointly displayed (3D rendering, cross and longitudinal sections) and registered with histology sections. Axial registration is achieved using pullback position information and angular registration is achieved using IVUS morphology (lumen shape, presence of calcifications). Classification groups are chosen based on histologic analysis: normal intima, fibrous cap, macrophage infiltration, and lipid core/extracellular lipid, as determined by H&E, Movat’s pentachrome and CD68 citeRN192, RN32.