Feasibility of three-dimensional virtual surgical planning in living liver donors

Two-step artificial intelligence algorithm for liver segmentation automates anatomic virtual hepatectomy

BACKGROUND

Anatomic virtual hepatectomy with precise liver segmentation for hemilivers, sectors, or Couinaud's segments using conventional three-dimensional simulation is not automated and artificial intelligence (AI)-based algorithms have not yet been applied.

METHODS

Computed tomography data of 174 living-donor candidates for liver transplantation (training data) were used for developing a new two-step AI algorithm to automate liver segmentation that was validated in another 51 donors (validation data). The Pure-AI (no human intervention) and ground truth (GT, full human intervention) data groups were compared.

RESULTS

In the Pure-AI group, the median Dice coefficients of the right and left hemilivers were highly similar, 0.95 and 0.92, respectively; sectors, posterior to lateral: 0.86-0.92, and Couinaud's segments 1-8: 0.71-0.89. Labeling of the first-order branch as hemiliver, right or left portal vein perfectly matched; 92.8% of the second-order (sectors); 91.6% of third-order (segments) matched between the Pure-AI and GT data.

CONCLUSIONS

The two-step AI algorithm for liver segmentation automates anatomic virtual hepatectomy. The AI-based algorithm correctly divided all hemilivers, and more than 90% of the sectors and segments.

High-resolution MR imaging with gadoxetate disodium for the comprehensive evaluation of potential living liver donors

OBJECTIVE

Several major transplantation centers have used composite multimodality evaluation for the preoperative evaluation of potential living liver donors. This approach can be time-consuming and, although rare, can cause complications. We aimed to demonstrate the clinical feasibility of our comprehensive preoperative MR protocol for the preoperative assessment of living liver donor candidates instead of composite multimodality evaluation.

MATERIALS AND METHODS

Thirty-five consecutive living liver donor candidates underwent multiphasic liver CT and comprehensive donor protocol MR examinations for preoperative evaluation in a single large-volume liver transplantation (LT) center. Three blinded abdominal radiologists reviewed the CT and MR images for vascular and biliary variations. The strength of agreement between CT and MR angiography was assessed using the kappa index. The detection rate of biliary anatomical variations was calculated. The sensitivity and specificity for detecting significant steatosis (>5%) were calculated. The estimated total volume and right lobe volumes measured by MR volumetry were compared with the corresponding CT volumetry measurements using the intraclass correlation coefficient (ICC).

RESULTS

Among the 35 patients, 26 underwent LT. The measurement of agreement showed a moderate to substantial agreement between CT and MR angiography interpretations (kappa values, 0.47-0.79; p < 0.001). Combining T2-weighted and T1-weighted MR cholangiography techniques detected all biliary anatomical variations in 9 of the 26 patients. MR-proton density fat fraction showed a sensitivity of 100% (3/3) and a specificity of 91.3% (21/23) for detecting pathologically determined steatosis (>5%). MR volumetry reached an excellent agreement with CT volumetry (reviewers 1 and 2: ICC, 0.92; 95% CI, 0.84-0.96).

CONCLUSION

Our one-stop comprehensive liver donor MR imaging protocol can provide complete information regarding hepatic vascular and biliary anatomies, hepatic parenchymal quality, and liver volume for living liver donor candidates and can replace composite multimodality evaluation.

Survey on Liver Tumour Resection Planning System: Steps, Techniques, and Parameters

Preoperative planning for liver surgical treatments is an essential planning tool that aids in reducing the risks of surgical resection. Based on the computed tomography (CT) images, the resection can be planned before the actual tumour resection surgery. The computer-aided system provides an overview of the spatial relationships of the liver organ and its internal structures, tumours, and vasculature. It also allows for an accurate calculation of the remaining liver volume after resection. The aim of this paper was to review the main stages of the computer-aided system that helps to evaluate the risk of resection during liver cancer surgical treatments. The computer-aided system assists with surgical planning by enabling physicians to get volumetric measurements and visualise the liver, tumours, and surrounding vasculature. In this paper, it is concluded that for accurate planning of tumour resections, the liver organ and its internal structures should be segmented to understand the clear spatial relationship between them, thus allowing for a safer resection. This paper presents the main proposed segmentation techniques for each stage in the computer-aided system, namely the liver organ, tumours, and vessels. From the reviewed methods, it has been found that instead of relying on a single specific technique, a combination of a group of techniques would give more accurate segmentation results. The extracted masks from the segmentation algorithms are fused together to give the surgeons the 3D visualisation tool to study the spatial relationships of the liver and to calculate the required resection planning parameters.

Consensus recommendations of three-dimensional visualization for diagnosis and management of liver diseases

Three-dimensional (3D) visualization involves feature extraction and 3D reconstruction of CT images using a computer processing technology. It is a tool for displaying, describing, and interpreting 3D anatomy and morphological features of organs, thus providing intuitive, stereoscopic, and accurate methods for clinical decision-making. It has played an increasingly significant role in the diagnosis and management of liver diseases. Over the last decade, it has been proven safe and effective to use 3D simulation software for pre-hepatectomy assessment, virtual hepatectomy, and measurement of liver volumes in blood flow areas of the portal vein; meanwhile, the use of 3D models in combination with hydrodynamic analysis has become a novel non-invasive method for diagnosis and detection of portal hypertension. We herein describe the progress of research on 3D visualization, its workflow, current situation, challenges, opportunities, and its capacity to improve clinical decision-making, emphasizing its utility for patients with liver diseases. Current advances in modern imaging technologies have promised a further increase in diagnostic efficacy of liver diseases. For example, complex internal anatomy of the liver and detailed morphological features of liver lesions can be reflected from CT-based 3D models. A meta-analysis reported that the application of 3D visualization technology in the diagnosis and management of primary hepatocellular carcinoma has significant or extremely significant differences over the control group in terms of intraoperative blood loss, postoperative complications, recovery of postoperative liver function, operation time, hospitalization time, and tumor recurrence on short-term follow-up. However, the acquisition of high-quality CT images and the use of these images for 3D visualization processing lack a unified standard, quality control system, and homogeneity, which might hinder the evaluation of application efficacy in different clinical centers, causing enormous inconvenience to clinical practice and scientific research. Therefore, rigorous operating guidelines and quality control systems need to be established for 3D visualization of liver to develop it to become a mature technology. Herein, we provide recommendations for the research on diagnosis and management of 3D visualization in liver diseases to meet this urgent need in this research field.

Comparison of semi-automatic and deep learning-based automatic methods for liver segmentation in living liver transplant donors

PURPOSE

To compare the accuracy and repeatability of emerging machine learning based (i.e. deep) automatic segmentation algorithms with those of well-established semi-automatic (interactive) methods for determining liver volume in living liver transplant donors at computerized tomography (CT) imaging.

METHODS

A total of 12 (6 semi-, 6 full-automatic) methods are evaluated. The semi-automatic segmentation algorithms are based on both traditional iterative models including watershed, fast marching, region growing, active contours and modern techniques including robust statistical segmenter and super-pixels. These methods entail some sort of interaction mechanism such as placing initialization seeds on images or determining a parameter range. The automatic methods are based on deep learning and they include three framework templates (DeepMedic, NiftyNet and U-Net) the first two of which are applied with default parameter sets and the last two involve adapted novel model designs. For 20 living donors (6 training and 12 test datasets), a group of imaging scientists and radiologists created ground truths by performing manual segmentations on contrast material-enhanced CT images. Each segmentation is evaluated using five metrics (i.e. volume overlap and relative volume errors, average/RMS/maximum symmetrical surface distances). The results are mapped to a scoring system and a final grade is calculated by taking their average. Accuracy and repeatability were evaluated using slice by slice comparisons and volumetric analysis. Diversity and complementarity are observed through heatmaps. Majority voting and Simultaneous Truth and Performance Level Estimation (STAPLE) algorithms are utilized to obtain the fusion of the individual results.

RESULTS

The top four methods are determined to be automatic deep models having 79.63, 79.46 and 77.15 and 74.50 scores. Intra-user score is determined as 95.14. Overall, deep automatic segmentation outperformed interactive techniques on all metrics. The mean volume of liver of ground truth is found to be 1409.93 mL ± 271.28 mL, while it is calculated as 1342.21 mL ± 231.24 mL using automatic and 1201.26 mL ± 258.13 mL using interactive methods, showing higher accuracy and less variation on behalf of automatic methods. The qualitative analysis of segmentation results showed significant diversity and complementarity enabling the idea of using ensembles to obtain superior results. The fusion of automatic methods reached 83.87 with majority voting and 86.20 using STAPLE that are only slightly less than fusion of all methods that achieved 86.70 (majority voting) and 88.74 (STAPLE).

CONCLUSION

Use of the new deep learning based automatic segmentation algorithms substantially increases the accuracy and repeatability for segmentation and volumetric measurements of liver. Fusion of automatic methods based on ensemble approaches exhibits best results almost without any additional time cost due to potential parallel execution of multiple models.

Automatic atlas-based liver segmental anatomy identification for hepatic surgical planning

PURPOSE

For the liver to remain viable, the resection during hepatectomy procedure should proceed along the major vessels; hence, the resection planes of the anatomic segments are defined, which mark the peripheries of the self-contained segments inside the liver. Liver anatomic segments identification represents an essential step in the preoperative planning for liver surgical resection treatment.

METHOD

The method based on constructing atlases for the portal and the hepatic veins bifurcations, the atlas is used to localize the corresponding vein in each segmented vasculature using atlas matching. Point-based registration is used to deform the mesh of atlas to the vein branch. Three-dimensional distance map of the hepatic veins is constructed; the fast marching scheme is applied to extract the centerlines. The centerlines of the labeled major veins are extracted by defining the starting and the ending points of each labeled vein. Centerline is extracted by finding the shortest path between the two points. The extracted centerline is used to define the trajectories to plot the required planes between the anatomical segments.

RESULTS

The proposed approach is validated on the IRCAD database. Using visual inspection, the method succeeded to extract the major veins centerlines. Based on that, the anatomic segments are defined according to Couinaud segmental anatomy.

CONCLUSION

Automatic liver segmental anatomy identification assists the surgeons for liver analysis in a robust and reproducible way. The anatomic segments with other liver structures construct a 3D visualization tool that is used by the surgeons to study clearly the liver anatomy and the extension of the cancer inside the liver.

Computed tomography donor liver volumetry before liver transplantation in infants ≤10 kg: does the estimated graft diameter affect the outcome?

Aim of the study

Living donor liver transplantation (LDLT) is regularly performed in small-sized infants. Computed tomography (CT)-based donor liver volumetry is used to estimate the graft size. The aim of our study was to assess the results of CT liver volumetry and their impact on the clinical outcome after LDLT in extremely small-sized infants.

Patients and methods

In this study, we included all patients with a body weight of ≤10 kg who underwent living related liver transplantation at our centre between January 2004 and December 2014. In all cases of LDLT, a preoperative CT scan of the donor liver was performed, and the total liver and graft volumes were calculated. The graft shape was estimated by measuring the ventro-dorsal (thickness), cranio-caudal, and transversal (width) diameter of segment II/III. We assessed the impact of CT donor liver volumetry and other risk factors on the outcome, defined as patient and graft survival.

Results

In the study period, a total of 48 living related liver transplantations were performed at our centre in infants ≤10 kg [20 male (42%), 28 female (58%)]. The mean weight was 7.3 kg (range 4.4–10 kg). Among the recipients, 33 (69%) received primary abdominal closure and 15 (31%) had temporary abdominal closure. The patient and graft survival rates were 85% and 81%, respectively. In CT volumetry, the mean estimated graft volume was 255 mL (range 140–485 mL) and the actual measured mean graft weight was 307 g (range 127–463 g). The mean ventro-dorsal diameter of segment II/III was 6.9 cm (range 4.3–11.2 cm), the mean cranio-caudal diameter was 9 cm (range 5–14 cm), and the mean width was 10.5 cm (range 6–14.7 cm). The mean graft-body weight ratio (GBWR) was 4.38% (range 1.41–8.04%). A high graft weight, a GBWR >4%, and a large ventro-dorsal diameter of segment II/III were risk factors for poorer patient survival.

Conclusion

Preoperative assessment of the graft size is a crucial investigation before LDLT. For extremely small-sized recipients, not only the graft weight but also the graft shape seems to affect the outcome.

Dr. Liver: A preoperative planning system of liver graft volumetry for living donor liver transplantation

BACKGROUND AND OBJECTIVE

Manual tracing of the right and left liver lobes from computed tomography (CT) images for graft volumetry in preoperative surgery planning of living donor liver transplantation (LDLT) is common at most medical centers. This study aims to develop an automatic system with advanced image processing algorithms and user-friendly interfaces for liver graft volumetry and evaluate its accuracy and efficiency in comparison with a manual tracing method.

METHODS

The proposed system provides a sequential procedure consisting of (1) liver segmentation, (2) blood vessel segmentation, and (3) virtual liver resection for liver graft volumetry. Automatic segmentation algorithms using histogram analysis, hybrid level-set methods, and a customized region growing method were developed. User-friendly interfaces such as sequential and hierarchical user menus, context-sensitive on-screen hotkey menus, and real-time sound and visual feedback were implemented. Blood vessels were excluded from the liver for accurate liver graft volumetry. A large sphere-based interactive method was developed for dividing the liver into left and right lobes with a customized cutting plane. The proposed system was evaluated using 50 CT datasets in terms of graft weight estimation accuracy and task completion time through comparison to the manual tracing method. The accuracy of liver graft weight estimation was assessed by absolute difference (AD) and percentage of AD (%AD) between preoperatively estimated graft weight and intraoperatively measured graft weight. Intra- and inter-observer agreements of liver graft weight estimation were assessed by intraclass correlation coefficients (ICCs) using ten cases randomly selected.

RESULTS

The proposed system showed significantly higher accuracy and efficiency in liver graft weight estimation (AD = 21.0 ± 18.4 g; %AD = 3.1% ± 2.8%; percentage of %AD > 10% = none; task completion time = 7.3 ± 1.4 min) than the manual tracing method (AD = 70.5 ± 52.1 g; %AD = 10.2% ± 7.5%; percentage of %AD > 10% = 46%; task completion time = 37.9 ± 7.0 min). The proposed system showed slightly higher intra- and inter-observer agreements (ICC = 0.996 to 0.998) than the manual tracing method (ICC = 0.979 to 0.999).

CONCLUSIONS

The proposed system was proved accurate and efficient in liver graft volumetry for preoperative planning of LDLT.

Pre-treatment estimation of future remnant liver function using gadoxetic acid MRI in patients with HCC

BACKGROUND & AIMS

This study aimed to determine whether the predicted remnant liver function on dynamic hepatocyte-specific contrast media-enhanced magnetic resonance (DHCE-MR) imaging correlates with the results of the indocyanin green retention test (ICG R₁₅) after hepatic resection or radiofrequency ablation (RFA).

METHODS

This prospective multicenter study was approved by the Institutional Review Boards of each hospital. Informed consents were obtained from all. DHCE-MRI and ICG R₁₅ were performed in 57 patients scheduled to undergo hepatectomy or RFA for hepatocellular carcinoma, once before treatment and repeated on post-treatment day 3. In nine donors and three recipients, DHCE-MRI and ICG R₁₅ were performed only preoperatively. The predicted remnant liver function (HEFml) was estimated using the hepatic extraction fraction (HEF) multiplied by the remnant liver volume, and compared with post-treatment ICG R₁₅. Intra-individual heterogeneity of HEF was assessed using pooled coefficients of variation (CV) among hepatic segments. Finally, development of post-treatment hepatic failure was assessed according to the 50-50 criteria on post-treatment day 5.

RESULTS

Predicted remnant HEFml showed a negative correlation with post-treatment ICG R₁₅ (r=-0.45, p=0.001), whereas liver volume did not (p>0.05). There were significant correlations between pre-treatment HEFml and pre-treatment ICG R₁₅ (r=-0.33, p=0.006) and between post-treatment HEFml and post-treatment ICG R₁₅ (r=-0.54, p<0.001). Pooled CV among segmental HEFs was 12.6%. No patients showed post-treatment liver failure on post-treatment day 5.

CONCLUSIONS

DHCE-MRI using Gd-EOB-DTPA was able to provide both global and segmental liver function information, and post-treatment remnant liver function predicted on pre-treatment DHCE-MRI showed a significant negative correlation with post-treatment ICG R₁₅.

LAY SUMMARY

Post-treatment liver function could be predicted at pre-treatment DHCE-MRI. Liver function was heterogeneous among the liver segments. Liver anatomy, disease extent, and underlying liver function can be assessed in one DHCE-MRI examination.

CLINICAL TRIAL NUMBER

ClinicalTrials.gov number, NCT01490203.

Value of MR elastography for the preoperative estimation of liver regeneration capacity in patients with hepatocellular carcinoma

PURPOSE

To demonstrate the negative relationship between liver stiffness (LS) values measured at preoperative magnetic resonance elastography (MRE) and the regeneration capacity of the remnant liver after major hepatectomy, in patients with hepatocellular carcinoma (HCC).

MATERIALS AND METHODS

Thirty-eight patients with HCC (mean age, 57.1) who had undergone liver computed tomography (CT) and 1.5T MRE prior to right hepatectomy were included in this retrospective study. CT volumetric analysis of total functional liver (calculated by subtracting tumor volume from total liver volume), future liver remnant (FLR), and postoperative liver remnant (LR) were performed using a semiautomatic analysis program. The regeneration index was expressed as [(V_LR -V_FLR )/V_FLR ] × 100, where V_LR is the volume of the liver remnant and V_FLR is the volume of the FLR. The relationship between degree of LS measured at MRE and the regeneration index was assessed using the Spearman correlation test.

RESULTS

Average LS value at MRE increased along with hepatic fibrosis (HF) stage (r = 0.604, P < 0.001). At MRE, a cutoff value greater than 2.46 kPa yielded 90.0% sensitivity and 100% specificity in identifying HF stage 2 or greater (area under the curve [AUC], 0.95). Mean V_FLR and V_LR were 477.1 ± 147.5 mL and 911.9 ± 188.8 mL, respectively. The regeneration index of the liver remnant was 102.1 ± 58.2%. LS values at MRE and calculated regeneration index after right hepatectomy showed moderate negative correlation (r = -0.361, P = 0.026).

CONCLUSION

LS values measured at MRE may serve as a postoperative predictor of liver regeneration in patients with liver cirrhosis and HCC.

LEVEL OF EVIDENCE

4 Technical Efficacy: Stage 2 J. MAGN. RESON. IMAGING 2017;45:1627-1636.