Limitations and pitfalls of Couinaud's segmentation of the liver in transaxial Imaging

Quantitative differences in volumetric calculations for radiation dosimetry in segmental Y90 treatment planning using hybrid angiography-CT compared with anatomic segmentation

OBJECTIVE

To compare treatment volumes reconstructed from hybrid Angio-CT catheter-directed infusion imaging and Couinaud anatomic model as well as the implied differences in Y-90 radiation dosimetry.

METHODS

Patients who underwent transarterial radioembolization (TARE) using Y-90 glass microspheres with pretreatment CT or MRI imaging as well as intraprocedural angiography-CT (Angio-CT) were analysed. Treatment volumes were delineated using both tumoural angiosomes (derived from Angio-CT) and Couinaud anatomic landmarks. Segmental and lobar treatment volumes were calculated via semi-automated contouring software. Volume and dose differences were compared by the two-tailed Student t test or Wilcoxon signed-rank test. Factors affecting volume and dose differences were assessed via simple and/or multiple variable linear regression analysis.

RESULTS

From September 2018 to March 2021, 44 patients underwent 45 lobar treatments and 38 patients received 56 segmental treatments. All target liver lobes and all tumours were completely included within the field-of-view by Angio-CT. Tumour sizes ranged between 1.1 and 19.5 cm in diameter. Segmental volumes and treatment doses were significantly different between the Couinaud and Angio-CT volumetry methods (316 vs 404 mL, P < .0001 and 253 vs 212 Gy, P < .01, respectively). Watershed tumours were significantly correlated with underestimated volumes by the Couinaud anatomic model (P < .001). There was a significant linear relationship between tumour diameter and percent volume difference (R2 = 0.44, P < .0001). The Couinaud model overestimated volumes for large tumours that exhibited central hypovascularity/necrosis and for superselected peripheral tumours.

CONCLUSIONS

Angio-CT may confer advantages over the Couinaud anatomic model and enable more accurate, personalized dosimetry for TARE.

ADVANCES IN KNOWLEDGE

Angio-CT may confer advantages over traditional cross-sectional and cone-beam CT imaging for selective internal radiation therapy planning.

Reproducibility and reliability of computed tomography volumetry in estimation of the right-lobe graft weight in adult-to-adult living donor liver transplantation: Cantlie's line vs portal vein territorialization

BACKGROUND/PURPOSE

In living-donor liver transplantation (LDLT), liver volume assessment is a mandatory step in determining donor appropriateness. This study aimed to compare reliability and reproducibility between two major methods to define virtual hepatectomy plane, based on Cantlie's line (CTV-Cantlie) and portal vein territorialization (CTV-PVT) for right-lobe graft weight estimation in LDLT.

METHODS

A total of 188 donors who underwent preoperative CT scans were included. The liver was divided into right and left lobes using CTV-Cantlie and CTV-PTV measurements by two readers. Intraclass correlation coefficient (ICC) was used to determine interreader variability of hepatic weight measured using each CTV method. Intraoperative graft weight (IOW) was used as reference standard of right-lobe graft weight. Pearson correlation test was performed to determine correlation coefficients between presumed graft weight by each CTV method and IOW.

RESULTS

Intraclass correlation coefficients for total liver weight were roughly equivalent between the two CTV methods (CTV-Cantlie: 0.965 [95% CI, 0.954-0.974], CTV-PVT: 0.977 [0.970-0.983]). However, ICCs of right-and left-lobe weights between two readers were higher with CTV-PVT (0.997 and 0.850) than with CTV-Cantlie (0.829 and 0.668). The IOW was 716.0 ± 162.0 g. Correlation coefficients between presumed graft weight by CTV-Cantlie or CTV-PVT and IOW were 0.722 and 0.807, respectively (both P < .001).

CONCLUSIONS

For estimation of the right-lobe graft weight in LDLT, CTV-PVT may provide higher reliability and reproducibility than CTV-Cantlie.

Human liver territories: Think beyond the 8-segments scheme

Worldwide, compartmentalization of the human liver into portal venous territories today follows the eight-segments scheme credited to Couinaud. However, there are increasing reports of anatomical, radiological and surgical observations that contradict this concept. This paper presents a viewpoint that enhances understanding of these inconsistencies and can serve as a basis for customized liver interventions. Clin. Anat. 30:974-977, 2017. © 2017 Wiley Periodicals, Inc.

Resection plane-dependent error in computed tomography volumetry of the right hepatic lobe in living liver donors

Background/Aims

Computed tomography (CT) hepatic volumetry is currently accepted as the most reliable method for preoperative estimation of graft weight in living donor liver transplantation (LDLT). However, several factors can cause inaccuracies in CT volumetry compared to real graft weight. The purpose of this study was to determine the frequency and degree of resection plane-dependent error in CT volumetry of the right hepatic lobe in LDLT.

Methods

Forty-six living liver donors underwent CT before donor surgery and on postoperative day 7. Prospective CT volumetry (V_P) was measured via the assumptive hepatectomy plane. Retrospective liver volume (V_R) was measured using the actual plane by comparing preoperative and postoperative CT. Compared with intraoperatively measured weight (W), errors in percentage (%) V_P and V_R were evaluated. Plane-dependent error in V_P was defined as the absolute difference between V_P and V_R. % plane-dependent error was defined as follows: |V_P–V_R|/W∙100.

Results

Mean V_P, V_R, and W were 761.9 mL, 755.0 mL, and 696.9 g. Mean and % errors in V_P were 73.3 mL and 10.7%. Mean error and % error in V_R were 64.4 mL and 9.3%. Mean plane-dependent error in V_P was 32.4 mL. Mean % plane-dependent error was 4.7%. Plane-dependent error in V_P exceeded 10% of W in approximately 10% of the subjects in our study.

Conclusions

There was approximately 5% plane-dependent error in liver V_P on CT volumetry. Plane-dependent error in V_P exceeded 10% of W in approximately 10% of LDLT donors in our study. This error should be considered, especially when CT volumetry is performed by a less experienced operator who is not well acquainted with the donor hepatectomy plane.

The error analysis of Lobular and segmental division of right liver by volume measurement

The aim of this study is to explore the inconsistencies between right liver volume as measured by imaging and the actual anatomical appearance of the right lobe. Five healthy donated livers were studied. The liver slices were obtained with hepatic segments multicolor-infused through the portal vein. In the slices, the lobes were divided by two methods: radiological landmarks and real anatomical boundaries. The areas of the right anterior lobe (RAL) and right posterior lobe (RPL) on each slice were measured using Photoshop CS5 and AutoCAD, and the volumes of the two lobes were calculated. There was no statistically significant difference between the volumes of the RAL or RPL as measured by the radiological landmarks (RL) and anatomical boundaries (AB) methods. However, the curves of the square error value of the RAL and RPL measured using CT showed that the three lowest points were at the cranial, intermediate, and caudal levels. The U- or V-shaped curves of the square error rate of the RAL and RPL revealed that the lowest value is at the intermediate level and the highest at the cranial and caudal levels. On CT images, less accurate landmarks were used to divide the RAL and RPL at the cranial and caudal layers. The measured volumes of hepatic segments VIII and VI would be less than their true values, and the measured volumes of hepatic segments VII and V would be greater than their true values, according to radiological landmarks. Clin. Anat. 30:585-590, 2017. © 2017 Wiley Periodicals, Inc.

Patient specific anatomy: the new area of anatomy based on computer science illustrated on liver

BACKGROUND

Over the past century, medical imaging has brought a new revolution: internal anatomy of a patient could be seen without any invasive technique. This revolution has highlighted the two main limits of current anatomy: the anatomical description is physician dependent, and the average anatomy is more and more frequently insufficient to describe anatomical variations. These drawbacks can sometimes be so important that they create mistakes but they can be overcome through the use of 3D patient-specific surgical anatomy.

METHODS

In this article, we propose to illustrate such improvement of standard anatomy on liver. We first propose a general scheme allowing to easily compare the four main liver anatomical descriptions by Takasaki, Goldsmith and Woodburne, Bismuth and Couinaud. From this general scheme we propose four rules to apply in order to correct these initial anatomical definitions. Application of these rules allows to correct usual vascular topological mistakes of standard anatomy. We finally validate such correction on a database of 20 clinical cases compared to the 111 clinical cases of a Couinaud article.

RESULTS

Out of the 20 images of the database, we note a revealing difference in 14 cases (70%) on at least one important branch of the portal network. Only six cases (30%) do not present a revealing difference between both labellings. We also show that the right portal fissure location on our 20 cases defined between segment V and VI of our anatomical definition is well correlated with the real position described by Couinaud on 111 cases, knowing that the theoretical position was only found in 46 cases out of 111, i.e., 41.44% of cases with the non-corrected Couinaud definition.

CONCLUSIONS

We have proposed a new anatomical segmentation of the liver based on four main rules to apply in order to correct topological errors of the four main standard segmentations. Our validation clearly illustrates that this new definition corrects the large amount of mistakes created by the current standard definitions, increased by physician interpretation that can vary from one case to another.

Horizontal liver cleft: a rare anatomic variant

BACKGROUND

Improvements in hepatobiliary surgical techniques, with increased usage of segmental and subsegmental resection, make accurate preoperative radiological assessment delineation of the liver segments ever more crucial. Conventionally, this is done by drawing imaginary straight planes along the portal and hepatic veins. We herein report a rare case of a horizontal cleft between the superior and inferior liver segments seen on CT.

CASE REPORT

A 74-year-old female patient with a known medical history of ovarian cancer with peritoneal metastasis and retroperitoneal lymphadenopathy was referred to our department for CT to assess disease response after treatment. On contrast-enhanced CT, apart from the ovarian cancer, the liver had a smooth, well-defined horizontally orientated cleft that broadly divided the organ into 2 halves. The cleft contained the right and left main portal veins, and consequently had a curved down-sloping configuration accommodating the curved course of these veins. This liver cleft was present from an earlier CT study performed 3 years ago, and there was no history of preceding liver surgery.

CONCLUSIONS

To the best of our knowledge, this is the first report of the anomaly of a horizontal liver cleft, which may be attributed to early cessation of the embryological formation of the liver. This liver cleft also illustrates the difficulties in liver segmentation using Couinaud's classification.

Anatomy of the liver: an outline with three levels of complexity--a further step towards tailored territorial liver resections

The vascular anatomy of the liver can be described at three different levels of complexity according to the use that the description has to serve. The first--conventional--level corresponds to the traditional 8-segments scheme of Couinaud and serves as a common language between clinicians from different specialties to describe the location of focal hepatic lesions. The second--surgical--level, to be applied to anatomical liver resections and transplantations, takes into account the real branching of the major portal pedicles and of the hepatic veins. Radiological and surgical techniques exist nowadays to make full use of this anatomy, but this requires accepting that the Couinaud scheme is a simplification, and looking at the vascular architecture with an unprejudiced eye. The third--academic--level of complexity concerns the anatomist, and the need to offer a systematization that resolves the apparent contradictions between anatomical literature, radiological imaging, and surgical practice. Based on the real number of second-order portal branches that, although variable averages 20, we submit a system called the "1-2-20 concept", and suggest that it fits best the number of actual--as opposed to idealized--anatomical liver segments.

Concepts for Liver Segment Classification: Neither Old Ones nor New Ones, but a Comprehensive One

Concepts dealing with the subdivision of the human liver into independent vascular and biliary territories are applied routinely in radiological, surgical, and gastroenterological practice. Despite Couinaud's widely used eight-segments scheme, opinions on the issue differ considerably between authors. The aim of this article is to illustrate the scientific basis for understanding and harmonizing inconsistencies between seemingly contradictory observations. Possible clinical implications are addressed.

A comparison of hepatic segmental anatomy as revealed by cross-sections and MPR CT imaging

To compare the areas of human liver horizontal sections with computed tomography (CT) images and to evaluate whether the subsegments determined by CT are consistent with the actual anatomy. Six human cadaver livers were made into horizontal slices with multislice spiral CT three-dimensional (3D) reconstruction was used during infusion process. Each liver segment was displayed using different color, and 3D images of the portal and hepatic vein were reconstructed. Each segmental area was measured on CT-reconstructed images, which were compared with the actual area on the sections of the same liver. The measurements were performed at four key levels namely: (1) the three hepatic veins, (2) the left, and (3) the right branch of portal vein (PV), and (4) caudal to the bifurcation of the PV. By dividing the sum of these areas by the total area of the liver, the authors got the percentage of the incorrectly determined subsegmental areas. In addition to these percentage values, the maximum distances of the radiologically determined intersegmental boundaries from the true anatomic boundaries were measured. On the four key levels, an average of 28.64 ± 10.26% of the hepatic area of CT images was attributed to an incorrect segment. The mean-maximum error between artificial segments on images and actual anatomical segments was 3.81 ± 1.37 cm. The correlation between radiological segmenting method and actual anatomy was poor. The hepatic segments being divided strictly according to the branching point of the PV could be more informative during liver segmental resection.

Recent advances in liver imaging

Liver surgery remains a difficult challenge in which preoperative data analysis and strategy definition may play a significant role in the success of the procedure. Medical image processing led to a major improvement of patient care by guiding the surgical gesture. From this initial data, new technologies of virtual reality and augmented reality can increase the potential of such images. The 3D modeling of the liver of patients from their CT scan or MRI thus allows an improved surgical planning. Simulation allows the procedure to be simulated preoperatively and offers the opportunity to train the surgical gesture before carrying it out. These three preoperative steps can be used intraoperatively thanks to the development of augmented reality, which consists of superimposing the preoperative 3D modeling of the patient onto the real intraoperative view of the patient and his/her organs. Augmented reality provides surgeons with a transparent view of the patient. This facilitated the intraoperative identification of the vascular anatomy and the control of the segmental arteries and veins in liver surgery, thus preventing intraoperative bleeding. It can also offer guidance due to the virtual improvement of their real surgical tools, which are tracked in real-time during the procedure. During the surgical procedure, augmented reality, therefore, offers surgeons a transparent view of their patient, which will lead to the automation of the most complex maneuvers. The new ways of processing and analyzing liver images have dramatically changed the approach to liver surgery.

Liver segments: an anatomical rationale for explaining inconsistencies with Couinaud's eight-segment concept

BACKGROUND AND PURPOSE

An increasing number of surgical and radiological observations call Couinaud's concept of eight liver segments into question and such inconsistencies are commonly explained with anatomical variations. This paper was intended to demonstrate that, beyond variability, another anatomical principle may allow to understand supposedly differing concepts on liver segmentation.

MATERIALS AND METHODS

The study was performed on 25 portal vein casts scanned by helical CT. The branches of the right and left portal vein and their corresponding territories were determined both anatomically and mathematically (MEVIS LiverAnalyzer, MEVISLab).

RESULTS

The number of branches coming-off the right and left portal vein was never 8, but many more (mean number 20, range 9-44). Different combinations of these branches and their respective territories, carried out in this study, yielded larger entities and supposedly contradictory subdivisions (including Couinaud's eight segments), without calling upon anatomical variability.

CONCLUSIONS

We suggest the human liver to be considered as corresponding to 1 portal venous territory at the level of the portal vein, to 2 territories at the level of the right and left branch of the portal vein, and to 20 at the level of the rami of the right and left branch. This "1-2-20-concept" is a rationale for reconciling apparent discrepancies with the eight-segment concept. On a pragmatic level, in cases in which imaging or surgical observations do not fit with Couinaud's scheme, we propose clinicians not to autonomically conclude to the presence of an anatomical variation, but to become aware of the presence of an average of 20 (and not 8) second-order portal venous territories within the human liver.

Reclassification of segmental anatomy of liver

Radiologists have become increasingly interested in segmental anatomy of human liver, especially in view of the need for an accurate preoperative localization of focal hepatic lesions and preoperative appraisal of living donor liver transplantation. The method proposed by Couinaud in 1964 to divide the liver into 8 surgically relevant segments longitudinally along the hepatic veins and transversely through the right and left portal pedicles, has been widely accepted. Although this method may be used in some patients, from a morphologic point of view, it is questionable because of the variation in segmental anatomy of human liver. In this review, reclassification of segmental anatomy of liver is described.

Reappraisal of Right Portal Segmental Ramification Based on 3-dimensional Volume Rendering of Computed Tomography During Arterial Portography

OBJECTIVE

To investigate and describe the segmental ramification patterns of the right portal vein (RPV) according to the Couinaud system.

MATERIALS AND METHODS

Between February 2004 and June 2005, 127 patients with hepatic tumors underwent computed tomography during arterial portography with a 16-slice multidetector computed tomography. The final analysis included 90 patients without RPV thrombosis or obvious vascular distortion. The ramification patterns of RPV were verified by 3-dimensional portograms using volume-rendering technique.

RESULTS

Seventy-five patients (83.3%) had bifurcation of the main portal vein, 12 (13.3%) had trifurcation, and 3 (3.3%) had the right posterior portal vein (RPPV) arising from main portal vein. A total of 5 segmental types and 3 subsegmental subgroups of RPV ramification patterns were clarified: type I, the classic ramification pattern with right anterior portal vein (RAPV) branching to S8/S5 and RPPV branching to S7/S6 (63; 70%); II, two separate segmental branches to S7 and S6 without a definite main stem of RPPV (18; 20%); III, "whisk-like" ramification pattern of RPV (2; 2.2%); IV, RAPV branching to S8 alone and RPPV to S5, S6, and S7, consecutively (5; 5.6%); and V, RPV first branching to S8/S5 and then to S7/S6 after a common path (2; 2.2%); subgroup a with dorsocranially directed branches arising from P8 and supplying S8 posterior to the right hepatic vein (28; 31.1%); subgroup b with RPPV branching to the dorsal part of S5 (11; 12.2%); and subgroup a + b, combination of the aforementioned 2 subgroups (45; 50%). In most patients, RAPV had dorsocranially directed branches posterior to the right hepatic vein (73; 81.1%), and RPPV gave off branches to the dorsal part of S5 (56; 62.2%).

CONCLUSIONS

Recognition of these ramification patterns could be helpful for more accurate anatomical resection of right hemiliver and preoperative planning, although some variants are present.

Liver segmentation in living liver transplant donors: comparison of semiautomatic and manual methods

PURPOSE

To compare the accuracy and repeatability of a semiautomatic segmentation algorithm with those of manual segmentation for determining liver volume in living liver transplant donors at magnetic resonance (MR) imaging.

MATERIALS AND METHODS

The institutional review board approved this retrospective study and waived the requirement for informed consent. The semiautomatic segmentation algorithm is based on geometric deformable models and the level-set technique. It entails (a) placing initialization circle(s) on each image section, (b) running the algorithm, (c) inspecting and possibly manually modifying the contours obtained with the segmentation algorithm, and (d) placing lines to separate the liver segments. For 18 living donors (eight men and 10 women; mean age, 34 years; age range, 25-46 years), two observers each performed two semiautomatic and two manual segmentations on contrast material-enhanced T1-weighted MR images. Each measurement was timed. Actual graft weight was measured during surgery. The time needed for manual and that needed for semiautomatic segmentation were compared. Accuracy and repeatability were evaluated with the Bland-Altman method.

RESULTS

Mean interaction time was reduced from 25 minutes with manual segmentation to 5 minutes with semiautomatic segmentation. The mean total time for the semiautomatic process was 7 minutes 20 seconds. Differences between the actual volume and the estimated volume ranged from -223 to +123 mL for manual segmentation and from -214 to +86 mL for semiautomatic segmentation. The 95% limits of agreement for the ratio of actual graft volume to estimated graft volume were 0.686 and 1.601 for semiautomatic segmentation and 0.651 and 1.957 for manual segmentation. Semiautomatic segmentation improved estimation in 15 of 18 cases. Inter- and intraobserver repeatability was higher with semiautomatic segmentation.

CONCLUSION

Use of the semiautomatic segmentation algorithm substantially reduces the time needed for volumetric measurement of liver segments while improving both accuracy and repeatability.