CC BY
31DOI 10.24412/cl-37136-2023-1-7-15
ORGAN-PAM: PHOTOACOUSTIC MICROSCOPY OF WHOLE-ORGAN MULTISET VESSEL
SYSTEMS
JUNNING ZHANG \ DINGLU PENG1, WEI QIN1, WEIZHI QI1, XIAOYA LIU1, YUCHENG
LUO1, QIONGYU GUO1 AND LEI XI1, 2 3
1 Department of Biomedical Engineering, Southern University of Science and Technology, China 2Guangdong Provincial Key Laboratory of Advanced Biomaterials, Southern University of Science and
Technology, China 3Shenzhen Bay Laboratory, China
ABSTRACT
Organs maintain their unique functions through a variety of distinctive vessel systems. However, examination of multiset vessels over the entire organ still remains challenging due to the difficulties of assigning multiple vessel systems and labor-intensive imaging reconstruction technologies [1-3]. Micro X-ray computed tomography (p,CT) is one of the most used non-optical imaging modalities for in vitro visualization and analysis of organ vessel systems due to its high spatial resolution three-dimensional (3D) imaging capability [4]. However, the drawbacks of ^CT are its insufficient signal-to-noise ratio (SNR) and field of view (FOV) as well as the complicated angiographic and data processing protocols [5]. Optical-based imaging modalities also show promising application in revealing organ vessel systems due to their superior resolution and sensitivity. Whole-organ optical clearing technique combined with cutting-edge optical microscopies, such as light-sheet microscopies, can resolve organ vessel systems with cellular-level spatial resolution [6]. Nevertheless, the complex and time-consuming organ optical clearing protocols as well as the tradeoff among temporal resolution, imaging FOV and spatial resolution remains major challenges. Sectioning tomography assisted by tissue microtomy technique can also achieve cellular-level imaging of whole-organ vessel systems [7]. However, correction of unavoidable distortions on the sectioning surfaces, complicated labeling and the unavoidable trade-off between spatial resolution, temporal resolution and imaging scale remain unsolved. Briefly, current commonly applied organ vessel system imaging platforms still face shared technical barriers: (1) unbalance between large-scale organ compatibility, spatial resolution and temporal resolution and (2) complex labeling procedures, complicated labor-assist-required segmentation algorithms and resource-consuming post-imaging processes in distinguishing multiset vessel systems. So far, an organ vessel system imaging platform with a good balance between spatial resolution, temporal resolution and multiscale-organ compatibility that applies simplified protocols to easily distinguish multiset vessel systems has not yet been reported.
Photoacoustic imaging (PAI), an emerging hybrid imaging modality utilize the phenomenon of generating ultrasound waves from a pulsed laser-excited optical absorber, exhibits the advantageous capability of vasculature characterization due to its rich optical imaging contrast and deep ultrasound sensing depth [8]. Photoacoustic microscopy (PAM), a major sub-modality of PAI, features a high spatial resolution of microns, rich optical absorption sensitivity, and 3D imaging ability, making it accessible and promising to investigate vessel systems in organs [8]. However, there remains several major challenges for PAM to achieve whole-organ multiset vessel system imaging: (1) the severe optical scattering of biological tissues limits the penetration depth to ~1.5 mm, preventing PAM from reaching the whole-organ depth [8]; (2) the limited spatial FOVs of existing PAMs, especially the axial FOV, are insufficient for imaging organs of large rodents, non-human primates, and humans; and (3) it is difficult for PAM to differentiate different vessel systems in the organ. As a result, so far, there is no systematic description of using the superior imaging capabilities of PAM to achieve whole-organ imaging of vessel systems.
In this study, we resolved such challenges hindering the feasibility of whole-organ vessel system visualization using PAM and further overcome the technical barriers of imaging scale and multiset vessel
distinguish complexity faced by current approaches. We developed an organ-level photoacoustic microscopic imaging platform (Organ-PAM), which utilizes a fast optical-clearing-assisted organ decellularization technique, a dual-illuminant ultra-large FOV optical-resolution PAM system, and a well-designed photoacoustic angiographic imaging protocol.
The detailed design of the imaging interface of our designed dual-illuminant ultra-large FOV optical-resolution PAM system consists of four separate units from top to bottom (Figure 1a), both the upper-side and bottom-side excitation units are made up of a 532 nm pulsed laser to induce photoacoustic waves, a two-dimensional galvanometer to perform rotary scanning of the light beam, and an F-theta lens to converge the light beam into the sample. The transducer rotates coordinately with the light scanning via a specially designed rotary scanning mechanism to achieve an effective imaging domain with a diameter of 4 cm in lateral (Figure 1b), the unit of the motorized sample holder moves the oversized sample in a raster pattern to obtain an extended lateral FOV via an imaging stitching algorithm (Figure 1c). We achieved a lateral resolution perpendicular to the acoustic beam of ~12.1 ^m and an axial resolution along the acoustic axis of ~172.8 ^m with the dual-illuminant ultra-large FOV optical-resolution PAM system (Figure 1d) and the current maximum lateral FOV can be achieved up to ~81 cm2. Based on the distribution and a tolerable spatial resolution limit of 25 ^m as the worst resolution (Figure 1e), we estimated the best double-sided illumination scenario where an axial FOV of ~12 mm can be achieved with an acceptable lateral resolution (Figure 1f).
Figure 1: Organ-PAM imaging interface configuration and lateral and axial FOVs. (a) 3D rendered configuration of the Organ-PAM imaging interface. (b) Close-up schematic of imaging target placed on the object holder and rotationally scanned by dual-sided scanning laser beams axially oriented to each other. (c) Maximum lateral FOV evaluation using an ultra-large planner phantom. (d) The best lateral and axial resolution of Organ-PAM. (e) The distribution of Organ-PAM's lateral resolution in the off-focus range of -5 mm to +5 mm along the axial direction. (f The calculated Organ-PAM's lateral resolutions over the axial FOV of ~12 mm. 2D-MOH, two-dimensional movable object holder; SWRG, sound wave reflection glass; UST-RC, ultrasonic transducer rotation chamber; MOA, mutual
overlap area; OHMT, object holder moving trajectory; SCIA, single circular imaging area; ESF, edge spread function; EXDP, experimental data point; PSF, point spread function; DTFP, distance to focal plane; LR, lateral resolution; OOFD, out-of-focus distance; CDOF, the center of depth-of-focus; CDP, calculated data point; DT-CDOF, distance to the center of depth-of-focus; L-DOF, lower side laser depth-of-focus; LLFP, lower side laser focal plane; U-DOF, upper side laser depth-of-focus; ULFP, upper side laser focal plane.
A photoacoustic gradient concentration differential angiography was proposed for the first time to enable complex-procedure-free labeling, imaging, and segmentation of multi-set vessel systems. This angiography is developed based on a fundamental photoacoustic phenomenon, which presents the amplitude of a photoacoustic wave linearly proportional to the concentration of the chromophore inside the absorber. According to the photoacoustic effect, samples labeled with different concentrations of chromophore can be distinguished by linear mapping between chromophore concentration and photoacoustic amplitude. By imaging a five-channel microfluidic chip injected using photoacoustic
contrast agents with different concentration, we observed varying photoacoustic intensities for different channels (Figure 2a-f). Based on the microfluidic phantom experiment and combined with organ decellularization clearing technique to produce acellular organs with enhanced optical transmittance following a decellularization protocol, we developed photoacoustic gradient concentration differential angiography for imaging multiple vessel systems in a single organ (Figure 2g-h).
Figure 2: Performance demonstration of photoacoustic gradient concentration differential angiography. (a) Colorimetric test of solvent soybean oil andfive PAI contrast agents with gradient concentration formulated by
mixing soybean oil and black oily ink in different ratios. (b) The PAI contrast agents were injected into the corresponding channels on a vessel phantom. (c) Amplitude normalized imaging results of the contrasted vessel phantom using Organ-PAM. (d) The zoomed-in visualization of the area within the white square in (c). (e) The normalized amplitude profile corresponds to the white dotted profile line marked in (d) with AG indicated. (f The
workflow of applying photoacoustic concentration differential angiography to segment PAI signal from the five contrasted microfluidic channels in a vessel phantom. (g) A demonstration workflow of applying the photoacoustic gradient concentration differential angiographic method to imaging and separate PAI signals from two vessel systems located at different depths inside an organ. (h) Re-merged color-labeled image of two vessel systems separated through the workflow shown in (g). AG, amplitude gap; C&I, contrast-and-imaging; DMI, denoised merged image; N. Amplitude, normalized amplitude; SI, segmented image; THV, threshold value; CA, contrast agent; DLV, deep-layer vessel system; DTT, decellularized transparent tissue; SL, scanning laser; SLID, scanning laser incident direction; SLV, shallow-layer vessel system; VS, vessel system.
Utilizing our dual-illuminant ultra-large FOV optical-resolution PAM system and photoacoustic gradient concentration differential angiography organ multiple vessel systems imaging workflow, we first achieved panoramic visualization of the hepatic vein (HV), portal vein (PV), hepatic artery (HA), and bile duct (BD) systems in a rat liver. We sequentially infused the oil-based ink solutions with increasing concentrations of 1%, 4%, 20%, and 100% into HV, PV, BD, and HA, respectively, and carried out Organ-
PAM scans after each infusion (Figure 3). To show the detailed 3D spatial distribution of these vessel systems, we zoomed in on three sub-views from different lobes of the liver with a FOV of 5 mm x 5 mm x 4 mm (x-, y- and z-axis, respectively), shown in Figure 3f, and observed the portal triad of PV, BD, and HA as well as the discrete HV in the three corresponding cross-sectional slices (Figure 3g).
Figure 3: 2D and 3D visualization of hepatic vessel systems inside an SD rat liver using Organ-PAM. Amplitude normalized 2D MAP images of (a) hepatic vein, (b) portal vein, (c) bile duct, and (d) hepatic artery. (e) 3D color-labeled and merged reconstruction results of whole-liver vessel systems. (f Close-up volumetric view of three
sections corresponding to the dashed white squares marked in (e). (g) Selected longitudinal cross-section reconstructions corresponding to the area within the dashed white rectangles in (f). BD, bile duct; HA, hepatic
artery; HV, hepatic vein; PV, portal vein.
The lateral FOV is essential for Organ-PAM to investigate the vessel systems inside organs of large animals and even humans. To show the ultra-large FOV of Organ-PAM, we harvested a New Zealand rabbit liver with a size of 12 cm * 12 cm and decellularized the entire organ. Due to the fact that the size of the liver is far beyond the one-scan FOV (4 cm in diameter) of Organ-PAM, we divided the liver into nine partially overlapping circular regions (Dia. 4 cm) that covered the entire liver (Figure 4). Following the nine single-scans of the divided regions, the image registration algorithm recognizes the overlapped areas between adjacent images and achieves image tiling by matching the overlapped areas (Figure 4b). Using photoacoustic concentration differential angiography, we achieved the visualization and segmentation of whole-set HV and PV systems of the healthy rabbit liver (Figure 4c).
For organs with planar sides that can be flattened on the imaging plane, such as the liver, lung, and spleen, as well as organs from small animals that possess relatively thin tissue, single-sided illumination is sufficient to cover the entire organ in depth. Unfortunately, the thickness of some important organs, such as the heart, stomach and kidney, are beyond the best effective axial FOV of single-sided illumination. The Organ-PAM we designed can achieve a doubled axial FOV of up to ~30 mm using a specially designed double-sided illumination. To highlight the superiority of the extended axial FOV, we tested the panoramic volumetric visualization of whole-set renal artery (RA) and renal vein (RV) systems of a rabbit kidney with a thickness of 11 mm (Figure 5).
Figure 4: Visualization of large-scale whole-hepatic vein systems inside a New Zealand rabbit liver by Organ-PAM.
(a) Multi-regional imaging of large-scale New Zealand rabbit liver to reconstruct complete vessel system visualization. (b) Image tiling workflow demonstration of two adjacent imaging units corresponding to the imaging regions marked by black circles in (a). (c) The complete 2D reconstruction of the large-scale New Zealand rabbit liver whole-hepatic HV and PV and two ROI images were obtained by re-merging and color-labeling the two vessel systems inside the areas corresponding to the areas within the dotted white squares. HV, hepatic vein; PV, portal vein; WH-HV, whole-hepatic HV; WH-PV, whole-hepatic PV.
a Upper tid* b
Figure 5: 2D and 3D visualization of whole-renal vessel systems using Organ-PAM. (a) The schematic of whole-renal concentration differential angiography with Organ-PAM. (b) 2D MAP images of RV and RA of each half of the kidney. (c) 3D color-labeled complete visualizations of RV and RA. (d) Close-up volumetric view of two sub-sections
corresponding to the dashed white boxes marked in (c). (e) The selected longitudinal cross-sections of the sub-visualizations correspond to the dashed white rectangles in (d). (f Quantification of RV and RA vascular density (%) according to the imaging data of both halves within four zones equidistantly divided from the root vessel original point to the renal edge. RA, renal artery; RV, renal vein; LH-RA, lower-half renal artery; LH-RV, lower-half renal vein; UH-RA, upper-half renal artery; UH-RV, upper-half renal vein.
Unlike kidneys and livers, lungs are highly ductile organs with more complex extracellular matrix (ECM) structures rich in various structural proteins, resulting in more structural proteins in decellularized lung tissue and thus more light scattering and less tissue transparency. Hence, it is quite challenging to achieve sufficient light penetration at thick tissue regions of decellularized lungs. To address such limitation, we combined decellularization with a unique intra-organ-perfusion-style optical clearing (OC), which enables faster and better transparency of lungs. To further demonstrate the non-vascular vessel system visualization capability and the multi-organ compatibility of Organ-PAM, we visualized the PT and
pulmonary artery (PA) systems panoramically within a rat lung treated with both decellularization and
intra-organ-perfusion-style OC (Figure 6).
3 Decellularized W/O OC Decellularized ♦ OC b Decellularized W/O OC
Decellularized + OC
s*
P
® \ i II
if \ - 5 mo«
PulfTKrtiacy Trachea (PT)
L&U Dual-aid« Merged MAP View
■ i ttV
Upper-illumination
' \ ,
3&L/
-S^gr^
Upper-illumination
V- A i
1 cm
L&U Dual-Sid* Merged MAP View
m
. T fv ' li
Pulmonary Trachea (PT)
PT & PA Merged Color-labeled 3D X-Y View
Pulmonary Artery (PA)
PT-TB
/ / \
/ I 1
V v _ ✓
/
PT&PA
Cluster
® LL Root Area PT-PB
/ !
Alveolar f Cluster
® RL PT-PB Middle
® RL PT-TB Middle
J PT&PA ~ Cluster
'v J 1 """
® LL PT&PA Cluster
Figure 6: Fast optical-clearing-assisted decellularized tissue transparency and the visualization of pulmonary trachea (PT) and pulmonary artery (PA) systems within a rat lung using Organ-PAM. (a)-(b) The comparation of
tissue transparency and imaging performance between decellularized rat lung and optical clearing enhanced decellularized rat lung. (c)-(d) The 2D MAP visualizations of segmented PT and PA. Visualizations captured using both lower and upper illuminations and merged visualizations are provided. (e) 3D color-labeled visualization of PT and PA with four representative areas selected for zoom-in views. OC, optical clearing; W/O, without; L&U, lower and upper; LL, left lung; PA, pulmonary artery; PT, pulmonary trachea; PT-PB, pulmonary trachea primary bronchi; PT-SB, pulmonary trachea secondary bronchi; PT-TB, pulmonary trachea tertiary bronchi.
We further utilized the Organ-PAM to probe the pathological vessel systems in primary rat liver cancers at distinct stages (Figure 7). The three sets of vasculatures (i.e., HV, PV, and HA) in the decellularized cancerous livers were contrasted and imaged to investigate the role of blood vessels during cancer progression. We further derived the vascular density ratio (VDR, normalized indicator of vascular density), branch point quantity (BPQ) and vessel length (VL) to quantify the difference between cancers at different stages. We note that the artery system shows an increase, while the vein systems reveal decreases from the marginal area to the center area. All the quantification results demonstrate that the artery system plays a significant role in the metabolic activity of hepatic tumor compared with the vein systems.
Overall, we present an organ-level photoacoustic microscopic imaging platform, named Organ-PAM, with a micron-scale spatial resolution and ultra-large field of views (FOVs), which can achieve a lateral
and axial FOV of up to ~81 cm2 and ~30 mm, respectively. With the assistance of whole-organ decellularization, optical clearing and a specifically designed photoacoustic gradient concentration differential angiographic pipeline, we successfully achieve the visualization of up to four-set vessel systems inside diversified organs with multiple scales. In addition, we conduct quantitative analyses of vessel systems in both healthy livers and kidneys as well as cancerous livers with exogenous transplanted tumor at different stages. Thus, the platform enables high-efficiency multiset vessel imaging, recognition, and quantification of different organs, providing critical insights into distinct vessel systems under varying pathological conditions.
Advanced Tumor Intermediate-stage Tumor Tumor Location Merged Image Tumor Cross-section H&E Staining Analysis
Figure 7: Visualization and quantitative analysis of whole-hepatic vessel systems inside SD rat livers with exogenous transplanted tumors using Organ-PAM. The substance photos ((a) and (e)) and amplitude normalized 2D MAP images ofHV ((b) and (f)), PV ((c) and (g)), and HA ((d) and (h)) of two cancerous rat livers at advanced and intermediate stages, respectively, with dotted circles indicating the cancer locations. The color-labeled and merged
vessel systems inside the cancer regions at advanced (i) and intermediate stages (j). Histological analysis of the advanced cancer (k) and intermediate-stage cancer (l) using H&E staining with corresponding sectioning location
indicated in (i) and (j). The triangle markers in (i), (j), (k) and (l) indicate vessels can be identified and located mutually in both PAM images ((i) and (j)) and the histological images ((k) and (l)). (m) Local-to-local and local-to-global vascular density ratios of both cancerous livers. (n) Quantification of branch point quantity and vessel length
of the vessel systems at cancer sites in both cancerous livers. (o) Quantification of vessel system complexity for advanced liver cancers using box-counting. (p) The visualization showing the changing trend of vessel weights for the
vessel systems inside the advanced stage liver cancer from margin to center. HA, hepatic artery; HV, hepatic vein; PV, portal vein; AS, advanced stage; IS, intermediate-stage; L/L, local-to-local vascular density ratio; L/G, local-to-global vascular density ratio; BPQ, branch point quantity; VL, vessel length.
REFERENCES
[1] Collaboration GBoDLC, The Burden of Primary Liver Cancer and Underlying Etiologies From 1990 to 2015 at the Global, Regional, and National Level: Results From the Global Burden of Disease Study 2015, JAMA Oncology, 2017;3(12):1683-91.
[2] Lafci B, Hadjihambi A, Konstantinou C, Herraiz JL, Pellerin L, Burton NC, et al, Multimodal Assessment of Non-Alcoholic Fatty Liver Disease with Transmission-Reflection Optoacoustic Ultrasound, bioRxiv, 2022:2022.08.16.504139.
[3] Lugano R, Ramachandran M, Dimberg A, Tumor angiogenesis: causes, consequences, challenges and opportunities, Cellular and Molecular Life Sciences, 2020;77(9): 1745-70.
[4] Epah J, Palfi K, Dienst FL, Malacarne PF, Bremer R, Salamon M, et al, 3D Imaging and Quantitative Analysis of Vascular Networks: A Comparison of Ultramicroscopy and Micro-Computed Tomography, Theranostics, 2018;8(8):2117-33.
[5] Elkhoury JE, Shankar R, Ramakrishnan TS, Resolution and Limitations of X-Ray Micro-CT with Applications to Sandstones and Limestones, Transport in Porous Media, 2019;129(1):413-25.
[6] Yang L, Feuchtinger A, Moller W, Ding Y, Kutschke D, Moller G, et al, Three-Dimensional Quantitative Co-Mapping of Pulmonary Morphology and Nanoparticle Distribution with Cellular Resolution in Nondissected Murine Lungs, ACS Nano, 2019;13(2):1029-41.
[7] Zheng T, Feng Z, Wang X, Jiang T, Jin R, Zhao P, et al, Zheng T, Feng Z, Wang X, Jiang T, Jin R, Zhao P, et al. Review of micro-optical sectioning tomography (MOST): technology and applications for whole-brain optical imaging [Invited], Biomed Opt Express, 2019;10(8):4075-96.
[8] Wang LV, Yao J, A practical guide to photoacoustic tomography in the life sciences, Nat Methods, 2016;13(8):627-38.
