MRI-based radiomic feature analysis of end-stage liver disease for severity stratification

Abstract

Purpose We aimed to develop a predictive model of disease severity for cirrhosis using MRI-derived radiomic features of the liver and spleen and compared it to the existing disease severity metrics of MELD score and clinical decompensation. The MELD score is compiled solely by blood parameters, and so far, it was not investigated if extracted image-based features have the potential to reflect severity to potentially complement the calculated score. Methods This was a retrospective study of eligible patients with cirrhosis ($$n=90$$ n = 90 ) who underwent a contrast-enhanced MR screening protocol for hepatocellular carcinoma (HCC) screening at a tertiary academic center from 2015 to 2018. Radiomic feature analyses were used to train four prediction models for assessing the patient’s condition at time of scan: MELD score, MELD score $$\ge $$ ≥ 9 (median score of the cohort), MELD score $$\ge $$ ≥ 15 (the inflection between the risk and benefit of transplant), and clinical decompensation. Liver and spleen segmentations were used for feature extraction, followed by cross-validated random forest classification. Results Radiomic features of the liver and spleen were most predictive of clinical decompensation (AUC 0.84), which the MELD score could predict with an AUC of 0.78. Using liver or spleen features alone had slightly lower discrimination ability (AUC of 0.82 for liver and AUC of 0.78 for spleen features only), although this was not statistically significant on our cohort. When radiomic prediction models were trained to predict continuous MELD scores, there was poor correlation. When stratifying risk by splitting our cohort at the median MELD 9 or at MELD 15, our models achieved AUCs of 0.78 or 0.66, respectively. Conclusions We demonstrated that MRI-based radiomic features of the liver and spleen have the potential to predict the severity of liver cirrhosis, using decompensation or MELD status as imperfect surrogate measures for disease severity.

Combining Hepatic and Splenic CT Radiomic Features Improves Radiomic Analysis Performance for Liver Fibrosis Staging

“…In a recent study by Nitsch et al, magnetic resonance imaging (MRI)-based radiomic features and random forest classifier were used to predict the severity of liver cirrhosis, with clinical decompensation and model for end-stage liver disease (MELD) scores as reference standard. Nitsch et al concluded that adding splenic MRI-based radiomic features can increase the AUC for predicting a higher median MELD score or clinical decompensation [ 24 ]. However, their study lacked an objective reference standard and was only focused on patients with known cirrhosis.…”

Quantitative Analysis of Liver Disease Using MRI-Based Radiomic Features of the Liver and Spleen

“…Feature analysis was performed on the contrast-enhanced imaging acquisition [ 17 ]. Due to the homogenous imaging conditions described above, we did not apply any kind of pre-processing such as re-sampling or other normalization.…”

“…Due to the homogenous imaging conditions described above, we did not apply any kind of pre-processing such as re-sampling or other normalization. Liver and spleen images were segmented using a U-net-based network architecture successively trained on expert segmentations [ 17 , 19 ]. The PyRadiomics library (version 2.0.1) [ 20 ] was used to extract 1288 features each from the liver and spleen, resulting in a total of 2577 imaging features (including a size ratio).…”

“…These two organs were chosen as they manifest findings of cirrhosis and portal hypertension that can be easily segmented radiographically. We recently used this approach to test the ability of MR radiomic features against different measures of cirrhosis severity [ 17 ]. We now first aim to demonstrate the ability of these MR radiomic features of the liver and spleen to detect the presence of cirrhosis when comparing MR imaging in patients with cirrhosis to those without cirrhosis.…”

Can Persistent Homology Features Capture More Intrinsic Information about Tumors from 18F-Fluorodeoxyglucose Positron Emission Tomography/Computed Tomography Images of Head and Neck Cancer Patients?

“…Owing to the association of tumor genetic heterogeneity in HN cancer with patients’ prognoses, it was concluded that higher heterogeneity is related to worse outcomes [ 6 , 7 , 8 ]. This genetic heterogeneity can lead to imaging heterogeneity, which can be quantified by radiomics [ 9 ] on 18 F-fluorodeoxyglucose (FDG) positron emission tomography (PET)/computed tomography (CT) images [ 10 , 11 ]. The feasibility of conventional PET/CT features for risk assessment in patients with HN cancer has been extensively investigated [ 12 , 13 ].…”