Comparison of MRI- and CT-based semiautomated liver segmentation: a validation study

Liver volumetry and liver-regenerative interventions: history, rationale, and emerging tools

BACKGROUND

Postoperative hepatic insufficiency (PHI) is the most feared complication after hepatectomy. Volume of the future liver remnant (FLR) is one objectively measurable indicator to identify patients at risk of PHI. In this review, we summarized the development and rationale for the use of liver volumetry and liver-regenerative interventions and highlighted emerging tools that could yield new advancements in liver volumetry.

METHODS

A review of MEDLINE/PubMed, Embase, and Cochrane Library databases was conducted to identify literature related to liver volumetry. The references of relevant articles were reviewed to identify additional publications.

RESULTS

Liver volumetry based on radiologic imaging was developed in the 1980s to identify patients at risk of PHI and later used in the 1990s to evaluate grafts for living donor living transplantation. The field evolved in the 2000s by the introduction of standardized FLR based on the hepatic metabolic demands and in the 2010s by the introduction of the degree of hypertrophy and kinetic growth rate as measures of the FLR regenerative and functional capacity. Several liver-regenerative interventions, most notably portal vein embolization, are used to increase resectability and reduce the risk of PHI. In parallel with the increase in automation and machine assistance to physicians, many semi- and fully automated tools are being developed to facilitate liver volumetry.

CONCLUSION

Liver volumetry is the most reliable tool to detect patients at risk of PHI. Advances in imaging analysis technologies, newly developed functional measures, and liver-regenerative interventions have been improving our ability to perform safe hepatectomy.

Automated liver volumetry and hepatic steatosis quantification with magnetic resonance imaging proton density fat fraction

BACKGROUND

Preoperative imaging evaluation of liver volume and hepatic steatosis for the donor affects transplantation outcomes. However, computed tomography (CT) for liver volumetry and magnetic resonance spectroscopy (MRS) for hepatic steatosis are time consuming. Therefore, we investigated the correlation of automated 3D-multi-echo-Dixon sequence magnetic resonance imaging (ME-Dixon MRI) and its derived proton density fat fraction (MRI-PDFF) with CT liver volumetry and MRS hepatic steatosis measurements in living liver donors.

METHODS

This retrospective cross-sectional study was conducted from December 2017 to November 2022. We enrolled donors who received a dynamic CT scan and an MRI exam within 2 days. First, the CT volumetry was processed semiautomatically using commercial software, and ME-Dixon MRI volumetry was automatically measured using an embedded sequence. Next, the signal intensity of MRI-PDFF volumetric data was correlated with MRS as the gold standard.

RESULTS

We included the 165 living donors. The total liver volume of ME-Dixon MRI was significantly correlated with CT (r = 0.913, p < 0.001). The fat percentage measured using MRI-PDFF revealed a strong correlation between automatic segmental volume and MRS (r = 0.705, p < 0.001). Furthermore, the hepatic steatosis group (MRS ≥5%) had a strong correlation than the non-hepatic steatosis group (MRS <5%) in both volumetric (r = 0.906 vs. r = 0.887) and fat fraction analysis (r = 0.779 vs. r = 0.338).

CONCLUSION

Automated ME-Dixon MRI liver volumetry and MRI-PDFF were strongly correlated with CT liver volumetry and MRS hepatic steatosis measurements, especially in donors with hepatic steatosis.

The Natural History of Observed SDHx -Related Head and Neck Paragangliomas Using Three-Dimensional Volumetric Tumor Analysis

OBJECTIVE

Characterize the natural history and clinical behavior of head and neck paragangliomas (HNPGLs) in subjects with succinate dehydrogenase ( SDHx ) pathogenic variants using volumetric tumor measurements.

STUDY DESIGN

Cohort study.

SETTING

Tertiary academic referral center.

PATIENTS

Subjects with SDHx HNPGLs under observation for at least 6 months with 2 or more magnetic resonance imaging or computed tomography scans.

INTERVENTIONS

Diagnostic interventions include next-generation sequencing, magnetic resonance imaging, and computed tomography. Therapeutic interventions include microsurgical resection or stereotactic radiosurgery.

MAIN OUTCOME MEASURES

Radiographic progression was defined as a 20% or greater increase in volume. Cranial nerve (CN) functional outcomes were assessed using clinical documentation.

RESULTS

A total of 19 subjects with 32 tumors met the inclusion criteria. Median radiographic follow-up was 2.2 years, and the median volumetric growth rate was 0.47 cm 3 /yr. Kaplan-Meier estimated rates of survival free of radiographic progression for all SDHx tumors at 1, 2, and 3 years were 69, 50, and 22%, respectively. No tumors developed new CN palsies during the period of observation.

CONCLUSIONS

Over intermediate-term follow-up, observation of treatment-naive SDHx -related HNPGLs did not result in new cranial neuropathy. Although indefinite observation is only appropriate for select cases, these data support an interval of observation to characterize growth rate in asymptomatic to minimally symptomatic patients, who are at high risk of treatment-related morbidity. Given the early age at diagnosis and high risk of bilateral multifocal phenotypes in SDHx HNPGL mutation carriers, these data may aid in optimizing patient tumor control and CN functional preservation. Further studies are necessary to determine whether pretreatment growth rate is correlated with clinical outcomes.

Assessment of Anatomical Uniqueness of Maxillary Sinuses through 3D-3D Superimposition: An Additional Help to Personal Identification

Paranasal sinuses represent one of the most individualizing structures of the human body and some of them have been already analyzed for possible applications to personal identification, such as the frontal and sphenoid sinuses. This study explores the application of 3D-3D superimposition to maxillary sinuses in personal identification. One hundred head CT-scans of adult subjects (equally divided among males and females) were extracted from a hospital database. Maxillary sinuses were segmented twice from each subject through ITK-SNAP software and the correspondent 3D models were automatically superimposed to obtain 100 matches (when they belonged to the same person) and 100 mismatches (when they were extracted from different individuals), both from the right and left side. Average RMS (root mean square) point-to-point distance was then calculated for all the superimpositions; differences according to sex, side, and group (matches and mismatches) were assessed through three-way ANOVA test (p < 0.017). On average, RMS values were lower in matches (0.26 ± 0.19 mm in males, 0.24 ± 0.18 mm in females) than in mismatches (2.44 ± 0.87 mm in males, 2.20 ± 0.73 mm in females) with a significant difference (p < 0.001). No significant differences were found according to sex or side (p > 0.017). The study verified the potential of maxillary sinuses as reliable anatomical structures for personal identification in the forensic context.

Radiomics in Abdominopelvic Solid-Organ Oncologic Imaging: Current Status

Radiomics is the process of extraction of high-throughput quantitative imaging features from medical images. These features represent noninvasive quantitative biomarkers that go beyond the traditional imaging features visible to the human eye. This article first reviews the steps of the radiomics pipeline, including image acquisition, ROI selection and image segmentation, image preprocessing, feature extraction, feature selection, and model development and application. Current evidence for the application of radiomics in abdominopelvic solid-organ cancers is then reviewed. Applications including diagnosis, subtype determination, treatment response assessment, and outcome prediction are explored within the context of hepatobiliary and pancreatic cancer, renal cell carcinoma, prostate cancer, gynecologic cancer, and adrenal masses. This literature review focuses on the strongest available evidence, including systematic reviews, meta-analyses, and large multicenter studies. Limitations of the available literature are highlighted, including marked heterogeneity in radiomics methodology, frequent use of small sample sizes with high risk of overfitting, and lack of prospective design, external validation, and standardized radiomics workflow. Thus, although studies have laid a foundation that supports continued investigation into radiomics models, stronger evidence is needed before clinical adoption.

Liver volume based prediction model for patients with hepatitis B virus-related acute-on-chronic liver failure

BACKGROUND

Hepatitis B virus-related acute-on-chronic liver failure (HBV-ACLF) is a life-threatening disease with high short-term mortality. Early and accurate prognosis is significant for clinical decisions, in which liver volume (LV) imparts important information. However, LV has not been considered in current prognostic models for HBV-ACLF.

METHODS

Three hundred and twenty-three patients were recruited to the deriving cohort, while 163 were enrolled to validation cohort. The primary end-point was death within 28 days since admission. Estimated liver volume (ELV) was calculated by the formula based on healthy population. Logistic regression was used to develop a prediction model. Accuracy of models were evaluated by receiver operating characteristic (ROC) curves.

RESULTS

The ratio of LV to ELV (LV/ELV%) was significantly lower in non-survivors, and LV/ELV% ≤82% indicated poor prognosis. LV/ELV%, Age, prothrombin time (PT), the grade of hepatic encephalopathy (HE), ln-transformed total bilirubin (lnTBil), and log-transformed HBV DNA (Log₁₀ HBV DNA) were identified as independent predictors to develop an LV-based model, LEAP-HBV. The mean area under the ROC (AUC) of LEAP-HBV was 0.906 (95% CI, 0.904-0.908), higher than other non-LV-based models.

CONCLUSION

Liver volume was an independent predictor, and LEAP-HBV, a prediction model based on LV, was developed for the short-term mortality in HBV-ACLF. This study was registered on ClinicalTrails (NCT03977857).

Polycystic liver: automatic segmentation using deep learning on CT is faster and as accurate compared to manual segmentation

OBJECTIVE

This study aimed to develop and investigate the performance of a deep learning model based on a convolutional neural network (CNN) for the automatic segmentation of polycystic livers at CT imaging.

METHOD

This retrospective study used CT images of polycystic livers. To develop the CNN, supervised training and validation phases were performed using 190 CT series. To assess performance, the test phase was performed using 41 CT series. Manual segmentation by an expert radiologist (Rad1a) served as reference for all comparisons. Intra-observer variability was determined by the same reader after 12 weeks (Rad1b), and inter-observer variability by a second reader (Rad2). The Dice similarity coefficient (DSC) evaluated overlap between segmentations. CNN performance was assessed using the concordance correlation coefficient (CCC) and the two-by-two difference between the CCCs; their confidence interval was estimated with bootstrap and Bland-Altman analyses. Liver segmentation time was automatically recorded for each method.

RESULTS

A total of 231 series from 129 CT examinations on 88 consecutive patients were collected. For the CNN, the DSC was 0.95 ± 0.03 and volume analyses yielded a CCC of 0.995 compared with reference. No statistical difference was observed in the CCC between CNN automatic segmentation and manual segmentations performed to evaluate inter-observer and intra-observer variability. While manual segmentation required 22.4 ± 10.4 min, central and graphics processing units took an average of 5.0 ± 2.1 s and 2.0 ± 1.4 s, respectively.

CONCLUSION

Compared with manual segmentation, automated segmentation of polycystic livers using a deep learning method achieved much faster segmentation with similar performance.

KEY POINTS

• Automatic volumetry of polycystic livers using artificial intelligence method allows much faster segmentation than expert manual segmentation with similar performance. • No statistical difference was observed between automatic segmentation, inter-observer variability, or intra-observer variability.

Development of an Endovascular Model of Pelvic Hemorrhage Using Volumetric Computed Tomography Validation

PURPOSE

Uncontrolled pelvic hemorrhage from trauma is associated with mortality rates above 30%. The ability of an intervention to reduce blood loss from pelvic trauma is paramount to its success. The objective of this study was to determine if computed tomography volumetric analysis could be used to quantify blood loss in a porcine endovascular pelvic hemorrhage model.

MATERIALS AND METHODS

Yorkshire swine under general anesthesia underwent balloon dilation and rupture of the profunda femoris artery, which was confirmed by digital subtraction angiography. Computed tomography angiography and postprocessing segmentation were performed to quantify pelvic hemorrhage volume at 5 and 30 minutes after injury. Continuous hemodynamic and iliofemoral flow data were obtained. Baseline and postinjury hemoglobin, hematocrit and lactate were collected.

RESULTS

Of 6 animals enrolled, 5 survived the 30-minute post-injury period. One animal died at 15 minutes. Median volume of pelvic hemorrhage was 141±106 cm³ at 5 minutes and 302±79 cm³ at 30 minutes with a 114% median increase in hematoma volume over 25 minutes (p=0.040). There was a significant decrease in mean arterial pressure (107 to 71 mm Hg, p=0.030) and iliofemoral flow (561 to 122 mL/min, p=0.014) at 30 minutes postinjury, but no significant changes in hemoglobin, hematocrit, or heart rate.

CONCLUSION

Computed tomography volumetric analysis can be used to quantify rate and volume of blood loss in a porcine endovascular pelvic hemorrhage model. Future studies can incorporate this approach when evaluating the effect of hemorrhage control interventions associated with pelvic fractures.

Portal vein embolization with ethylene-vinyl alcohol copolymer for contralateral lobe hypertrophy before liver resection: safety, feasibility and initial experience

Background

To report our preliminary experience with preoperative portal vein embolization (PVE) using liquid ethylene vinyl alcohol (EVOH) copolymer.

Methods

Retrospectively review of patients with primary or secondary liver malignancies scheduled for extensive hepatectomy after the induction of future liver remnant (FLR) hypertrophy by right or left PVE with EVOH as the only embolic agent between 2014 and 2018 at two academic centers. Cross-sectional imaging liver volumetry data obtained before and 3-6 weeks after PVE were used to assess the FLR volume (FLRV) increase, degree of FLR hypertrophy and the FLR kinetic growth rate (KGR).

Results

Twenty-six patients (17 males; mean age, 58.7±11 years; range, 32-79 years) were included. The technical and clinical success rate was 100%. PVE produced adequate FLR hypertrophy in all patients. Embolization occurred in all targeted portal branches and in no non-target vessels. The %FLRV increased by 52.9%±32.5% and the degree of FLR hypertrophy was 16.7%±6.8%. The KGR was 4.4%±2.0% per week. Four patients experience minor complications after PVE which resolved with symptomatic treatment. The resection rate was 84.5%. One patient died during surgery for reasons unrelated to PVE.

Conclusions

Preoperative PVE with EVOH copolymer is feasible, safe, and effective in inducing FLR hypertrophy.

Deep Learning Algorithm for Automated Segmentation and Volume Measurement of the Liver and Spleen Using Portal Venous Phase Computed Tomography Images

Objective

Measurement of the liver and spleen volumes has clinical implications. Although computed tomography (CT) volumetry is considered to be the most reliable noninvasive method for liver and spleen volume measurement, it has limited application in clinical practice due to its time-consuming segmentation process. We aimed to develop and validate a deep learning algorithm (DLA) for fully automated liver and spleen segmentation using portal venous phase CT images in various liver conditions.

Materials and Methods

A DLA for liver and spleen segmentation was trained using a development dataset of portal venous CT images from 813 patients. Performance of the DLA was evaluated in two separate test datasets: dataset-1 which included 150 CT examinations in patients with various liver conditions (i.e., healthy liver, fatty liver, chronic liver disease, cirrhosis, and post-hepatectomy) and dataset-2 which included 50 pairs of CT examinations performed at ours and other institutions. The performance of the DLA was evaluated using the dice similarity score (DSS) for segmentation and Bland-Altman 95% limits of agreement (LOA) for measurement of the volumetric indices, which was compared with that of ground truth manual segmentation.

Results

In test dataset-1, the DLA achieved a mean DSS of 0.973 and 0.974 for liver and spleen segmentation, respectively, with no significant difference in DSS across different liver conditions (p = 0.60 and 0.26 for the liver and spleen, respectively). For the measurement of volumetric indices, the Bland-Altman 95% LOA was −0.17 ± 3.07% for liver volume and −0.56 ± 3.78% for spleen volume. In test dataset-2, DLA performance using CT images obtained at outside institutions and our institution was comparable for liver (DSS, 0.982 vs. 0.983; p = 0.28) and spleen (DSS, 0.969 vs. 0.968; p = 0.41) segmentation.

Conclusion

The DLA enabled highly accurate segmentation and volume measurement of the liver and spleen using portal venous phase CT images of patients with various liver conditions.

Validation of a fully automated liver segmentation algorithm using multi-scale deep reinforcement learning and comparison versus manual segmentation

PURPOSE

To evaluate the performance of an artificial intelligence (AI) based software solution tested on liver volumetric analyses and to compare the results to the manual contour segmentation.

MATERIALS AND METHODS

We retrospectively obtained 462 multiphasic CT datasets with six series for each patient: three different contrast phases and two slice thickness reconstructions (1.5/5 mm), totaling 2772 series. AI-based liver volumes were determined using multi-scale deep-reinforcement learning for 3D body markers detection and 3D structure segmentation. The algorithm was trained for liver volumetry on approximately 5000 datasets. We computed the absolute error of each automatically- and manually-derived volume relative to the mean manual volume. The mean processing time/dataset and method was recorded. Variations of liver volumes were compared using univariate generalized linear model analyses. A subgroup of 60 datasets was manually segmented by three radiologists, with a further subgroup of 20 segmented three times by each, to compare the automatically-derived results with the ground-truth.

RESULTS

The mean absolute error of the automatically-derived measurement was 44.3 mL (representing 2.37 % of the averaged liver volumes). The liver volume was neither dependent on the contrast phase (p = 0.697), nor on the slice thickness (p = 0.446). The mean processing time/dataset with the algorithm was 9.94 s (sec) compared to manual segmentation with 219.34 s. We found an excellent agreement between both approaches with an ICC value of 0.996.

CONCLUSION

The results of our study demonstrate that AI-powered fully automated liver volumetric analyses can be done with excellent accuracy, reproducibility, robustness, speed and agreement with the manual segmentation.

A robust multi-variability model based liver segmentation algorithm for CT-scan and MRI modalities

Developing methods to segment the liver in medical images, study and analyze it remains a significant challenge. The shape of the liver can vary considerably from one patient to another, and adjacent organs are visualized in medical images with similar intensities, making the boundaries of the liver ambiguous. Consequently, automatic or semi-automatic segmentation of liver is a difficult task. Moreover, scanning systems and magnetic resonance imaging have different settings and parameters. Thus the images obtained differ from one machine to another. In this article, we propose an automatic model-based segmentation that allows building a faithful 3-D representation of the liver, with a mean Dice value equal to 90.3% on CT and MRI datasets. We compare our algorithm with a semi-automatic method and with other approaches according to the state of the art. Our method works with different data sources, we use a large quantity of CT and MRI images from machines in various hospitals and multiple DICOM images available from public challenges. Finally, for evaluation of liver segmentation approaches in state of the art, robustness is not adequacy addressed with a precise definition. Another originality of this article is the introduction of a novel measure of robustness, which takes into account the liver variability at different scales.

Reducing inter-observer variability and interaction time of MR liver volumetry by combining automatic CNN-based liver segmentation and manual corrections

Purpose

To compare manual corrections of liver masks produced by a fully automatic segmentation method based on convolutional neural networks (CNN) with manual routine segmentations in MR images in terms of inter-observer variability and interaction time.

Methods

For testing, patient’s precise reference segmentations that fulfill the quality requirements for liver surgery were manually created. One radiologist and two radiology residents were asked to provide manual routine segmentations. We used our automatic segmentation method Liver-Net to produce liver masks for the test cases and asked a radiologist assistant and one further resident to correct the automatic results. All observers were asked to measure their interaction time. Both manual routine and corrected segmentations were compared with the reference annotations.

Results

The manual routine segmentations achieved a mean Dice index of 0.95 and a mean relative error (RVE) of 4.7%. The quality of liver masks produced by the Liver-Net was on average 0.95 Dice and 4.5% RVE. Liver masks resulting from manual corrections of automatically generated segmentations compared to routine results led to a significantly lower inter-observer variability (mean per case absolute RVE difference across observers 0.69%) when compared to manual routine ones (2.75%). The mean interaction time was 2 min for manual corrections and 10 min for manual routine segmentations.

Conclusions

The quality of automatic liver segmentations is on par with those from manual routines. Using automatic liver masks in the clinical workflow could lead to a reduction of segmentation time and a more consistent liver volume estimation across different observers.

Automatic segmentation methods for liver and hepatic vessels from CT and MRI volumes, applied to the Couinaud scheme

BACKGROUND

Proper segmentation of the liver from medical images is critical for computer-assisted diagnosis, therapy and surgical planning. Knowledge of its vascular structure allows division of the liver into eight functionally independent segments, each with its own vascular inflow, known as the Couinaud scheme. Couinaud's description is the most widely used classification, since it is well-suited for surgery and accurate for the localization of lesions. However, automatic segmentation of the liver and its vascular structure to construct the Couinaud scheme remains a challenging task.

METHODS

We present a complete framework to obtain Couinaud's classification in three main steps; first, we propose a model-based liver segmentation, then a vascular segmentation based on a skeleton process, and finally, the construction of the eight independent liver segments. Our algorithms are automatic and allow 3D visualizations.

RESULTS

We validate these algorithms on various databases with different imaging modalities (Magnetic Resonance Imaging (MRI) and Computed Tomography (CT)). Experimental results are presented on diseased livers, which pose complex challenges because both the overall organ shape and the vessels can be severely deformed. A mean DICE score of 0.915 is obtained for the liver segmentation, and an average accuracy of 0.98 for the vascular network. Finally, we present an evaluation of our method for performing the Couinaud segmentation thanks to medical reports with promising results.

CONCLUSIONS

We were able to automatically reconstruct 3-D volumes of the liver and its vessels on MRI and CT scans. Our goal is to develop an improved method to help radiologists with tumor localization.

Liver segmentation: indications, techniques and future directions

OBJECTIVES

Liver volumetry has emerged as an important tool in clinical practice. Liver volume is assessed primarily via organ segmentation of computed tomography (CT) and magnetic resonance imaging (MRI) images. The goal of this paper is to provide an accessible overview of liver segmentation targeted at radiologists and other healthcare professionals.

METHODS

Using images from CT and MRI, this paper reviews the indications for liver segmentation, technical approaches used in segmentation software and the developing roles of liver segmentation in clinical practice.

RESULTS

Liver segmentation for volumetric assessment is indicated prior to major hepatectomy, portal vein embolisation, associating liver partition and portal vein ligation for staged hepatectomy (ALPPS) and transplant. Segmentation software can be categorised according to amount of user input involved: manual, semi-automated and fully automated. Manual segmentation is considered the "gold standard" in clinical practice and research, but is tedious and time-consuming. Increasingly automated segmentation approaches are more robust, but may suffer from certain segmentation pitfalls. Emerging applications of segmentation include surgical planning and integration with MRI-based biomarkers.

CONCLUSIONS

Liver segmentation has multiple clinical applications and is expanding in scope. Clinicians can employ semi-automated or fully automated segmentation options to more efficiently integrate volumetry into clinical practice.

TEACHING POINTS

• Liver volume is assessed via organ segmentation on CT and MRI examinations. • Liver segmentation is used for volume assessment prior to major hepatic procedures. • Segmentation approaches may be categorised according to the amount of user input involved. • Emerging applications include surgical planning and integration with MRI-based biomarkers.

Assessment of treatment response in non-alcoholic steatohepatitis using advanced magnetic resonance imaging

BACKGROUND

Magnetic resonance imaging-derived measures of liver fat and volume are emerging as accurate, non-invasive imaging biomarkers in non-alcoholic steatohepatitis (NASH). Little is known about these measures in relation to histology longitudinally.

AIM

To examine any relationship between MRI-derived proton-density fat-fraction (PDFF), total liver volume (TLV), total liver fat index (TLFI), vs. histology in a NASH trial.

METHODS

This is a secondary analysis of a 24-week randomised, double-blind, placebo-controlled trial of 50 patients with biopsy-proven NASH randomised to oral ezetimibe 10 mg daily (n = 25) vs. placebo (n = 25). Baseline and post-treatment anthropometrics, biochemical profiling, MRI and biopsies were obtained.

RESULTS

Baseline mean PDFF correlated strongly with TLFI (Spearman's ρ = 0.94, n = 45, P < 0.0001) and had good correlation with TLV (ρ = 0.57, n = 45, P < 0.0001). Mean TLV correlated strongly with TLFI (ρ = 0.78, n = 45, P < 0.0001). After 24 weeks, PDFF remained strongly correlated with TLFI (ρ = 0.94, n = 45, P < 0.0001), maintaining good correlation with TLV (ρ = 0.51, n = 45, P = 0.0004). TLV remained strongly correlated with TLFI (ρ = 0.74, n = 45, P < 0.0001). Patients with Grade 1 vs. 3 steatosis had lower PDFF, TLV, and TLFI (P < 0.0001, P = 0.0003, P < 0.0001 respectively). Regression analysis of changes in MRI-PDFF vs. TLV indicates that 10% reduction in MRI-PDFF predicts 257 mL reduction in TLV.

CONCLUSIONS

The MRI-PDFF and TLV strongly correlated with TLFI. Decreases in steatosis were associated with an improvement in hepatomegaly. Lower values of these measures reflect lower histologic steatosis grades. MRI-derived measures of liver fat and volume may be used as dynamic and more responsive imaging biomarkers in a NASH trial, than histology.