SEARCH

SEARCH BY CITATION

Keywords:

  • 3D reconstruction;
  • hepatic vasculature;
  • image processing;
  • morphology;
  • tree analysis

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Concluding remarks
  8. Acknowledgements
  9. Conflict of interest
  10. References

Although a full understanding of the hepatic circulation is one of the keys to successfully perform liver surgery and to elucidate liver pathology, relatively little is known about the functional organization of the liver vasculature. Therefore, we materialized and visualized the human hepatic vasculature at different scales, and performed a morphological analysis by combining vascular corrosion casting with novel micro-computer tomography (CT) and image analysis techniques. A human liver vascular corrosion cast was obtained by simultaneous resin injection in the hepatic artery (HA) and portal vein (PV). A high resolution (110 μm) micro-CT scan of the total cast allowed gathering detailed macrovascular data. Subsequently, a mesocirculation sample (starting at generation 5; 88 × 68 × 80 mm³) and a microcirculation sample (terminal vessels including sinusoids; 2.0 × 1.5 × 1.7 mm³) were dissected and imaged at a 71-μm and 2.6-μm resolution, respectively. Segmentations and 3D reconstructions allowed quantifying the macro- and mesoscale branching topology, and geometrical features of HA, PV and hepatic venous trees up to 13 generations (radii ranging from 13.2 mm to 80 μm; lengths from 74.4 mm to 0.74 mm), as well as microvascular characteristics (mean sinusoidal radius of 6.63 μm). Combining corrosion casting and micro-CT imaging allows quantifying the branching topology and geometrical features of hepatic trees using a multiscale approach from the macro- down to the microcirculation. This may lead to novel insights into liver circulation, such as internal blood flow distributions and anatomical consequences of pathologies (e.g. cirrhosis).


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Concluding remarks
  8. Acknowledgements
  9. Conflict of interest
  10. References

The liver is a fascinating but complex multifunctional organ, characterized by its intricate vascular architecture. Compared with other organs, the liver's vasculature is unique due to the two blood supplies. Traditionally, the hepatic artery (HA) is regarded to supply the liver with oxygenated blood. The portal vein (PV) collects partially deoxygenated blood from the intestinal tract including spleen and pancreas. HA and PV blood mixes in the sinusoids, the hepatic microcirculation, often portrayed as a lattice of small vessels in between rows of hepatocytes, determining the smallest functional unit or liver lobule. Blood subsequently leaves liver lobules through central veins and eventually drains into the vena cava inferior (VCI) via the hepatic venous (HV) system (Roskams et al. 2007). All vascular trees as well as the microcirculation have their own morphological and functional characteristics. Together, they determine the hepatic hemodynamic behavior, such as the arterial buffer response (Eipel et al. 2010) or less physiological phenomena such as flow competition between the HA and PV (Monbaliu et al. 2012).

The clinical relevance to understand the exact hepatic vascular anatomy is illustrated by the disturbed vascular architecture in case of liver pathology, such as cirrhosis with fibrosis (overproduction of extracellular matrix), regenerative nodules, increased vascular resistance, neo-angiogenesis, and vascular remodeling leading to portal hypertension and intrahepatic shunt vessels (Anthony et al. 1978; Chen et al. 2009; Thabut & Shah, 2010). Moreover, the hepatic vasculature is subject to intraspecies anatomical differences, relevant for transplantation. Patient-specific analysis of the hepatic vascular topology through computed tomography (CT) and magnetic resonance angiography may facilitate surgical planning and improve its outcome (Selle et al. 2002; Yamanaka et al. 2006; Mutter et al. 2009). Furthermore, topology information and geometrical characteristics may be fed to numerical models to simulate hepatic hemodynamics, useful to model, for example, surgical procedures or the performance of new preservation techniques (e.g. machine perfusion) for transplant livers (Rani et al. 2006; Bonfiglio et al. 2010; Ricken et al. 2010; Debbaut et al. 2011, 2012a,b).

Morphological analysis of vascular trees in different organs and species is not new [e.g. rodent pulmonary trees (Gomes & Bates, 2002), mouse placental arteries (Rennie et al. 2011), human coronary arteries (Finet et al. 2007), mouse kidneys (Wagner et al. 2011)]. However, to our knowledge, only a few papers were published on the liver's vasculature. Op Den Buijs et al. (2006) classified the PV tree of the rat liver, but not the HA or HV tree. Selle et al. (2002) focused on analyzing the macrocirculation for surgical planning based on vascular territories. We previously performed studies on modeling liver perfusion based on the macrovessel architecture of the human (Debbaut et al. 2011) and rat liver (Debbaut et al. 2012a; Francque et al. 2012). A larger number of studies has been published on liver microcirculation using, for example, microscopy, histological techniques (Rappaport et al. 1954; McCuskey, 1966; Greenway & Stark, 1971; Matsumoto et al. 1979; Matsumoto & Kawakami, 1982; Ekataksin & Wake, 1991, 1997; McCuskey & Reilly, 1993; Teutsch et al. 1999), or performing 3D reconstructions by registering 2D serial cryosections as done by Teutsch (2005).

Until now, the total spectrum of the human liver vascular architecture has not been analyzed in a systematic way on single liver samples. Therefore, the aim of this study was to analyze the complete human liver vascular architecture using a combination of vascular corrosion casting, micro-CT scanning and image processing. Hereby, a novel multiscale approach was applied to consecutively analyze the macro-, meso- and microcirculation.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Concluding remarks
  8. Acknowledgements
  9. Conflict of interest
  10. References

In this study, a human liver was used after being discarded for transplantation due to failed reallocation. The protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki, and was approved by the Ethical Committee of the University Hospitals Leuven, Belgium, and by the Belgian Liver and Intestine Committee as foreseen by the initial protocol.

Vascular corrosion casting

After hepatectomy, the liver (weight ± 1.9 kg) was connected to a machine perfusion preservation device (Organ Recovery Systems, Zaventem, Belgium) during a 24-h period, and was continuously perfused at 4–6 °C with pressure-control through the HA (25 mmHg; unlimited flow) and PV (7 mmHg; flow limitation of 300 mL min−1). The pressure-controlled feedback system, incorporated in the perfusion machine, ensured that perfusion pressures did not exceed the maximum pressure settings. Machine perfusion allowed preserving the vasculature and parenchyma, and keeping the blood vessels open whilst preparing the necessary logistics. Subsequently, the liver was prepared for the vascular corrosion casting procedure by cannulation of the HA, PV as well as the VCI. Casting resin was prepared by mixing Batson's #17 monomer solution, catalyst and promoter (Polysciences, Warrington, PA, USA) with monomeric methyl methacrylate (Merck, Darmstadt, Germany) and color dyes (red and blue for the HA and PV, respectively). In addition, barium sulfate (50 mg mL−1; Micropaque, Delpharm, France) was added to the HA injection to amplify the contrast between arterial and venous vessels on CT images, facilitating to distinguish different contributing vessels. The HA and PV were simultaneously and manually injected until the resin emerged sufficiently from the VCI. Afterwards, inlet and outlet vessels were clamped to avoid resin leakage during polymerization of the injected mixture (approximately 2 h). Next, the liver was macerated in a potassium hydroxide bath (approximately 2 days). More detailed information on the vascular corrosion procedure is available in Debbaut et al. (2011).

Micro-CT imaging

Macrocirculation

The resulting human liver cast (Fig. 1a) was imaged in globo to acquire data of the first blood vessel generations, being a similar number of generations as obtained with traditional angiography techniques (macrocirculation; Fig. 1a). This was done using a state-of-the-art in-house developed high-resolution micro-CT scanner (Fig. 1d). Two-thousand images were recorded during a 360° rotation of the cast. The image dataset (resolution of 102 μm) was reconstructed with Octopus software (Ghent University, Gent, Belgium) and converted to the DICOM format. More information on the scanning procedure is available in Debbaut et al. (2011).

image

Figure 1. Human liver vascular corrosion cast and micro-computer tomography (CT) scanner. (a) Total liver with indication of the dissection location of the mesocirculation sample; (b) mesocirculation sample; (c) scanning electron microscopic image of the microcirculation sample; (d) micro-CT scanner with a static X-ray tube, a static flat panel detector and a rotating liver cast to capture images during a 360 ° rotation. HA, hepatic artery; HV, hepatic venous system; PV, portal vein.

Download figure to PowerPoint

Mesocirculation

A smaller sample (wedge-shaped; approximately 88 × 68 × 80 mm3) was dissected from the inferior part of the right lobe (Fig. 1a and b) to investigate the morphology of vessel generations distal to the macrocirculation. This sample was imaged at a resolution of 71 μm.

Microcirculation

Thirdly, a microvascular sample (approximately 2.0 × 1.5 × 1.7 mm3) was dissected from the superior part of the right lobe to study the smallest vessels of the liver (Fig. 1c). To assure that the sinusoids were filled during casting, this sample was imaged by scanning electron microscopy (Fig. 1c). Afterwards, the sample was imaged using the micro-CT scanner at a 2.6-μm resolution.

Image processing and vascular architecture analysis

Macrocirculation

The resulting micro-CT dataset was processed using Mimics (Materialise, Leuven, Belgium). The vascular trees were segmented based on the gray values of the images. Separating arterial from venous vessels was straightforward due to the arterial contrast agent. It was, however, challenging to segment the PV and HV trees (having similar gray values), which were manually separated at locations where they touched each other. Therefore, each vascular tree was followed starting at the first generation vessel going down to smaller vessels. When a touching vessel was detected, it was separated by removing the pixel(s) of the connection. After segmentation, a 3D reconstruction of each tree was calculated.

After image processing, the vascular tree centerlines were calculated using the centerline algorithm in Mimics. Using the concept of blood vessel generations, the centerlines were used to determine the branching topology and geometrical features of all vascular trees. This was done by classifying vessels based on their branching pattern (see Debbaut et al. 2011 for more detailed information), being either n-furcations (parent vessels splitting in n similar daughter vessels; dichotomous bifurcations when n = 2) or monopodial vessels (small side branches coming from parent vessels at an angle close to 90°; Gordon et al. 2007). Hereby, the inlet vessel of each vascular tree is assigned to generation 1. Daughter vessels have higher generation numbers than their parent vessel. For instance, if a generation n vessel splits into two daughter vessels with a similar smaller diameter, the daughter vessels are assigned to generation n + 1. In addition, the radius (r ; best fit radius averaged over the vessel) and length (l; centerline length) of each vessel were measured. After data acquisition, exponential trend lines were fitted to the data (mean radius, mean length, number of vessels) as a function of the generation number (Eq. (1) with y the geometrical feature, x the generation number, a and b the coefficients to be fitted). Exponential trend lines were selected, as they showed a better goodness of fit compared with linear and power law trend lines.

  • display math(1)
Mesocirculation

Again, segmentations and 3D reconstructions were performed. As this sample contains vessel generations distal to the macrocirculation, while the resolution is insufficient to capture the microvessels, it is labeled ‘mesocirculation’.

Vascular tree analysis of this sample was similar to that of the macrocirculation. However, for labor intensity reasons, we did not analyze the total sample but selected four representative subsamples (brighter subsamples in Fig. 2f–h). To register the mesocirculatory dataset with the macrocirculatory dataset, the ingoing HA, PV and HV vessel of each mesocirculation subsample were identified by their corresponding vessel in the macrocirculatory dataset. Doing so, it was straightforward to assign generation numbers and geometrical feature measurements (radii and lengths) to the mesovessels. However, because we did not quantify the total liver mesovasculature, the number of vessels vs. the generation number [nmeso,estimated (x); x is the generation number] had to be estimated. This was done by multiplying the measured number of vessels for each mesocirculation generation [nmeso (x)] by a factor. This factor is determined by dividing the number of macrocirculation vessels of the f  th generation [nmacro (f ); f is the number of the first mesocirculation generation included in all four subsamples] by the number of measured vessels of generation f in the mesocirculation [nmeso (f )] (Eq. (2)).

  • display math(2)
image

Figure 2. Three-dimensional reconstructions of the macrocirculation and mesocirculation. (a) Superposition of three macrovascular trees with indication of the dissection location of the mesocirculation sample; macrovascular HA (b), PV (c) and HV trees (d) with arrows indicating monopodial branches; (e) superposition of three mesovascular trees; mesovascular HA (f), PV (g) and HV trees (h) with brighter parts indicating four subsamples used to acquire geometrical data.

Download figure to PowerPoint

Microcirculation

Because, at the terminal microcirculation level (interconnected sinusoids), the difference between arterial and portal vessels becomes unclear and the tree structure is lost, the dataset was segmented as one volume, instead of being separated in HA, PV and HV trees. Consequently, this dataset was used to investigate the hepatic vascular microstructure and its dimensions by calculating 3D visualizations, and measuring lobule and sinusoid diameters.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Concluding remarks
  8. Acknowledgements
  9. Conflict of interest
  10. References

The casting procedure resulted in a replica of the human hepatic vascular system (Fig. 1a). HA vessels were red, while PV and HV were blue, due to the relatively high PV flow (75% of the outflow) compared with HA flow, resulting in dominantly blue HV flow.

Macrocirculation

Figure 2 shows 3D reconstructions of the total liver micro-CT scan. The HA, PV and HV vascular beds were clearly distinguishable. Next to n-furcations (such as bifurcations and trifurcations), the trees count a high number of monopodial vessels, sprouting from parent vessels at angles close to 90° (Fig. 2b–d). After the first generations, HA vessels run parallel to PV vessels. In a few cases, one PV vessel is even flanked by two HA vessels. Moreover, HA vessels have predominantly circular cross-sections compared with elliptical cross-sections of PV and HV vessels. This anatomical feature is probably due to the structural differences between veins (thin vessel walls) and arteries (thick vessel walls including a thick muscle layer). Each tree was classified according to its branching topology, resulting in six, six and five generations for the HA, PV and HV tree, respectively. Mean HA radii drop from 3.45 mm to 5.92 × 10−1 mm, PV radii from 7.34 mm to 1.08 mm, and HV radii from 13.2 mm (VCI) to 1.13 mm (Fig. 3a–c; see also tables 1–4 in Debbaut et al. 2011). As anticipated, portal and hepatic venous vessels (PV and HV) have larger diameters compared with HA vessels. Exponential trend lines, fitted to radii measurements, have high determination coefficients (R² ≥ 0.97; see also tables 1–4 in Debbaut et al. 2011). Mean lengths do not show a clear-cut decreasing trend in the first generations (especially for PV vessels), but decrease in higher generations (Fig. 3d–f). This is partially due to the first generation vessel of each tree being cut to resect the liver, implying an underestimated length. Therefore, the first generation length is not accounted for when calculating length trend lines. The PV length trend line has the lowest R² value (0.77). Length standard deviations are higher than for radii. Numbers of vessels per generation clearly increase exponentially towards higher generation numbers: from 1 to 271, 1 to 216, and 1 to 76 vessels for the HA, PV and HV tree, respectively (Fig. 3g–i; see also tables 1–4 in Debbaut et al. 2011).

image

Figure 3. Results of the vascular tree analysis of the macro- and mesocirculation. Radius (a–c), length (d–f) and number of vessels (g–i) as a function of the generation number for the hepatic artery (HA), portal vein (PV) and hepatic venous system (HV) trees. Macrocirculation measurements (as obtained in Debbaut et al. 2011) and mesocirculation measurements are indicated by black and white markers, respectively. [Original and estimated numbers of vessels for the mesocirculation (Table 1) are indicated by gray and white dots, respectively.] Exponential trend lines are depicted by dashed lines when fitted to macro- and mesocirculation data. Equations and coefficients of determination (R²) of the exponential functions are given. The first HV generation (VCI) was not taken into accounted when fitting the radius trend line, as the VCI has a much larger diameter and is not really part of the hepatic vasculature. Length trend lines did not incorporate the first generation, as these vessels were cut to resect the liver, resulting in an underestimated length.

Download figure to PowerPoint

Mesocirculation

Segmentations of the mesocirculation subsample (Fig. 1b) were more challenging than the macrocirculation, because more vessels were touching each other (Fig. 2e–h). Figures 1b and 2e show that the PV inflow runs parallel to two HA vessels. Similar to the macrocirculation, this sample shows elliptical PV and HV vessels compared with circular HA vessels. Analyzing four subsamples (Fig. 2f–h) resulted in the visualization of higher generation vessels, going up to generation 13, 13 and 10 for the HA, PV and HV trees, respectively (Table 1; Fig. 3). Mean radii decrease to 8.00 × 10−2, 1.23 × 10−1and 1.60 × 10−1 mm for the HA, PV and HV trees, respectively (Fig. 3a–c). Radii trend lines were fitted to the pooled macro- and mesocirculation measurements (R² ≥ 0.98). Mean lengths also decrease for all trees (R² ≥ 0.94). The numbers of vessels per generation again increases exponentially, almost exactly following extrapolations of the macrocirculation trend lines. All trend lines fitted to the combination of macro- and mesocirculation measurements show equal or higher goodness of fit than those fitted to only the macrocirculation, except for the HA radii (Fig. 3).

Table 1. Overview of the mesoscale HA, PV and HV measurements
 Generation numberMean radius (mm)Standard deviation of radius (mm)Mean length (mm)Standard deviation of length (mm)Number of vesselsEstimated number of vessels
  1. Mean radii, mean lengths and numbers of vessels are reported per generation number. (Estimated numbers of vessels of mesocirculation generations are calculated using Eq. (2).)

  2. HA, hepatic artery; HV, hepatic venous system; PV, portal vein.

HA MesoscaleHA 58.62 × 10−13.54 × 10−354.425.42 
HA 65.21 × 10−11.54 × 10−112.28.296 
HA 73.45 × 10−13.90 × 10−27.224.0412542
HA 82.77 × 10−14.70 × 10−23.512.62341536
HA 92.41 × 10−14.19 × 10−23.092.08763433
HA 102.08 × 10−12.73 × 10−22.361.381988943
HA 111.60 × 10−12.62 × 10−21.481.0534215 447
HA 121.19 × 10−12.47 × 10−21.005.45 × 10−122510 163
HA 138.00 × 10−22.46 × 10−27.41 × 10−12.40 × 10−19407
PV MesoscalePV 51.41019.901 
PV 68.86 × 10−12.25 × 10−114.87.945 
PV 77.18 × 10−11.72 × 10−18.144.3310432
PV 84.82 × 10−11.12 × 10−15.242.80311339
PV 93.09 × 10−16.16 × 10−23.161.80863715
PV 102.31 × 10−13.92 × 10−22.681.3924410 541
PV 111.75 × 10−12.61 × 10−21.488.75 × 10−155523 976
PV 121.38 × 10−12.39 × 10−21.034.91 × 10−1733154
PV 131.23 × 10−12.71 × 10−29.50 × 10−13.79 × 10−16259
HV MesoscaleHV 51.274.42 × 10−132.512.74 
HV 67.41 × 10−19.27 × 10−28.565.309171
HV 74.56 × 10−19.66 × 10−26.593.2326494
HV 82.46 × 10−17.65 × 10−24.091.971112109
HV 91.71 × 10−15.59 × 10−22.611.47741406
HV 101.60 × 10−16.10 × 10−22.171.0620380

Microcirculation

Scanning electron microscopy of the microcirculation sample shows that casting resin was observed in the sinusoids (Fig. 1c). The sample represents a complex network of interconnected and tortuous sinusoids (Fig. 4). Measurements of 100 sinusoids resulted in a mean diameter of 13.23 ± 2.36 μm.

image

Figure 4. Liver microcirculation. (a) Single micro-CT slice showing bright spots (probably contrast agent particles); (b) indication of the most likely lobule borders; (c) 3D reconstruction of a liver lobule; and (d) of a virtually dissected subsample.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Concluding remarks
  8. Acknowledgements
  9. Conflict of interest
  10. References

For the first time, the total spectrum of the (human) hepatic vasculature (from the largest vessels down to the sinusoidal microcirculation) is visualized and 3D reconstructed. Therefore, this pilot study used a combination of state-of-the-art vascular corrosion casting and micro-CT scanning techniques (at resolutions up to the order of 2.6 μm) with a novel image processing and analysis technique.

However, one has to be careful when generalizing these findings based on a single liver because of potential individual differences such as different vessel dimensions, different liver sizes according to the body weight (the liver represents approximately 2% of the total body weight), different liver shapes (e.g. a relatively small or large left lobe) and the potential presence of liver pathology (e.g. fibrotic tissue and regenerative nodules in liver cirrhosis). More livers should thus be similarly analyzed to get a general overview of the human hepatic vascular architecture. As such, this pilot study should be interpreted as the vascular analysis of a specific liver. This human liver was offered to our lab after failed rescue allocation and there was no obvious macroscopic abnormality. The casting procedure was done by manual injection, having the advantage of sensing how the organ reacts while injecting. Also, it has been described that casting may lead to slight shrinkage of the resin, which would imply slightly smaller diameters (Kratky & Roach, 1984). However, our diameter measurements were comparable with literature data on the largest hepatic vessels and sinusoidal diameters (Matsumoto & Kawakami, 1982; Warren et al. 2008; Vollmar & Menger, 2009). Moreover, the casting procedure was performed according to standardized procedures by a team having many years of expertise (Casteleyn et al. 2010; Debbaut et al. 2011, 2012a).

Ideally, the complete cast should be scanned at once at a sufficiently high resolution to allow visualizing all vessels down to the sinusoids. However, it was technically impossible to scan a total liver at a resolution of a few micrometers due to computational and software limitations to process datasets with an extremely large file size. Consequently, a multiscale approach was used to consecutively study the macro-, meso- and microcirculation, allowing the quantification of 13 (HA and PV) and 10 (HV) blood vessel generations as well as the sinusoids. However, there is still a gap of a number of generations between the 10/13th generation and the sinusoids. Nonetheless, the total number of generations can be approximated by extrapolating the resulting exponential trend lines of the radius (Fig. 3) down to the level of radius values of the terminal vessels, resulting in an estimation of 19–20, 17–18 and 13–14 generations for the HA, PV and HV trees, respectively. Hereby, sinusoids are interpreted as being the connection between the terminal/last generation of the inflow trees (HA and PV) and outflow tree (HV).

Our data reveal a complex hepatic vascular topology covering the macro- down to the microcirculation, showing that hepatic trees do not solely branch according to a dichotomous bifurcating pattern, which is often assumed in vascular tree models. n-Furcations as well as monopodial vessels were clearly visualized in the macro- and mesocirculation. Geometrical characteristics (Fig. 3) show exponential behavior as a function of generation numbers. HA and PV trees show similar behavior and exponential trends, resulting in equal numbers of generations for the macro- and mesocirculation. However, PV radii are larger than HA radii and, starting from generation 4, numbers of HA vessels are higher than those of PV vessels (Fig. 3g and h; Table 1; tables 2–4 in Debbaut et al. 2011), probably due to some PV vessels being flanked by two parallel HA vessels. The HV tree has the largest radii and typically counts less generations than the HA and PV tree.

In contrast, the microcirculation gives evidence of a completely different type of organization compared with vascular trees. Tortuous sinusoids form an interconnected and intertwined network embedded in a matrix of liver cells. Consequently, the exponential behavior of the HA, PV and HV trees probably stops at the level of the terminal microcirculation (where blood is drained from the terminal hepatic arterioles and portal venules into the sinusoids). From our data, macro- and mesovessels may be interpreted as ‘distributing’ vessels, ensuring that blood reaches all liver lobules, and the sinusoids may be interpreted as ‘functional’ vessels, ensuring that blood solutes are able to penetrate through sinusoidal fenestrations into the space of Disse and reach the microvilli of hepatocytes for metabolic exchange. The microvascular sample clearly shows sinusoidal structures (Fig. 1c; meaning that the casting resin was able to fill the sinusoids) and is thought to include three liver lobules. Hereby, the vascular septa of lobules seem to be highlighted by contrast agent particles (bright dots in Fig. 4a). These particles are probably not able to penetrate into the sinusoids because of their size, leaving them trapped at the vascular septa, (partially) delineating the lobule borders (Fig. 4b). Accordingly, the sample represents three liver lobules (Fig. 4a,b) and, based on these contours, lobule diameters were in the order of 700–800 μm. As illustrated, it is not obvious to distinguish between neighboring lobules. This might be due to the fact that the human liver microcirculation is characterized by less connective tissue delineating its lobule borders in comparison with pig or rodent livers, often used as animal liver models. Furthermore, lobules can have more irregular shapes than the traditionally hexagonal prism-shaped lobules. Concerning this, literature gives evidence of an ongoing discussion on the most appropriate functional unit to represent the liver microcirculation (e.g. hexagonal lobule, primary lobule, liver acinus; Roskams et al. 2007).

The novel and detailed morphological data gathered in this study are useful to complement scientific insights into liver morphology and physiology. The new multiscale approach is also applicable in the context of liver pathophysiology, for example to investigate pathology-related microvascular alterations in case of fibrosis and cirrhosis (Vanheule et al. 2008; Chen et al. 2009), hepatocellular carcinoma (Maksan et al. 2003), and portal hypertension (Yagi et al. 2005; Fondevila et al. 2010). Furthermore, this approach may generate input data for the development of numerical models of liver perfusion. These models can be applied to simulate the hemodynamic impact of (ab)normal situations such as isolated organ perfusion and surgical procedures, for example partial hepatectomy. Microcirculation morphological data enable gaining more insight into structural-related characteristics (porosity, permeability tensor), which allow capturing and modeling the microcirculation behavior especially relevant when studying microvascular alterations caused by diseases such as cirrhosis.

Concluding remarks

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Concluding remarks
  8. Acknowledgements
  9. Conflict of interest
  10. References

In conclusion, combining vascular corrosion casting with state-of-the-art high-resolution micro-CT scanning provides novel and unique data on the human hepatic vasculature from the macrocirculation down to the microcirculation. A new multiscale approach allows visualizing the complex liver vasculature at different levels in 3D, and performing a detailed analysis of the topology and geometrical features. These data are useful to create numerical models of the hepatic blood circulation. This approach could also be applied to other organs, such as kidneys.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Concluding remarks
  8. Acknowledgements
  9. Conflict of interest
  10. References

This research was supported by the Agency for Innovation by Science and Technology in Flanders (IWT), Belgium. Diethard Monbaliu holds a chair of the Centrale Afdeling voor Fractionering (CAF), Vilvoorde, Belgium.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Concluding remarks
  8. Acknowledgements
  9. Conflict of interest
  10. References