Two-point Dixon and six-point Dixon magnetic resonance techniques in the detection, quantification and grading of hepatic steatosis.

BACKGROUND: Hepatic steatosis is a very common problem worldwide. AIM: To assess the performance of two- and six-point Dixon magnetic resonance (MR) techniques in the detection, quantification and grading of hepatic steatosis. METHODS: A single-center retrospective study was performed in 62 patients with suspected parenchymal liver disease. MR sequences included two-point Dixon, six-point Dixon, MR spectroscopy (MRS) and MR elastography. Fat fraction (FF) estimates on the Dixon techniques were compared to the MRS-proton density FF (PDFF). Statistical tests used included Pearson's correlation and receiver operating characteristic. RESULTS: FF estimates on the Dixon techniques showed excellent correlation (≥ 0.95) with MRS-PDFF, and excellent accuracy [area under the receiver operating characteristic (AUROC) ≥ 0.95] in: (1) Detecting steatosis; and (2) Grading severe steatosis, (P < 0.001). In iron overload, two-point Dixon was not evaluable due to confounding T2* effects. FF estimates on six-point Dixon vs MRS-PDFF showed a moderate correlation (0.82) in iron overload vs an excellent correlation (0.97) without iron overload, (P < 0.03). The accuracy of six-point Dixon in grading mild steatosis improved (AUROC: 0.59 to 0.99) when iron overload cases were excluded. The excellent correlation (> 0.9) between the Dixon techniques vs MRS-PDFF did not change in the presence of liver fibrosis (P < 0.01). CONCLUSION: Dixon techniques performed satisfactorily for the evaluation of hepatic steatosis but with exceptions.

Deep-learning-based ensemble method for fully automated detection of renal masses on magnetic resonance images.

Purpose: Accurate detection of small renal masses (SRM) is a fundamental step for automated classification of benign and malignant or indolent and aggressive renal tumors. Magnetic resonance image (MRI) may outperform computed tomography (CT) for SRM subtype differentiation due to improved tissue characterization, but is less explored compared to CT. The objective of this study is to autonomously detect SRM on contrast-enhanced magnetic resonance images (CE-MRI). Approach: In this paper, we described a novel, fully automated methodology for accurate detection and localization of SRM on CE-MRI. We first determine the kidney boundaries using a U-Net convolutional neural network. We then search for SRM within the localized kidney regions using a mixture-of-experts ensemble model based on the U-Net architecture. Our dataset contained CE-MRI scans of 118 patients with different solid kidney tumor subtypes including renal cell carcinomas, oncocytomas, and fat-poor renal angiomyolipoma. We evaluated the proposed model on the entire CE-MRI dataset using 5-fold cross validation. Results: The developed algorithm reported a Dice similarity coefficient of 91.20 ± 5.41 % (mean ± standard deviation) for kidney segmentation from 118 volumes consisting of 25,025 slices. Our proposed ensemble model for SRM detection yielded a recall and precision of 86.2% and 83.3% on the entire CE-MRI dataset, respectively. Conclusions: We described a deep-learning-based method for fully automated SRM detection using CE-MR images, which has not been studied previously. The results are clinically important as SRM localization is a pre-step for fully automated diagnosis of SRM subtypes.

Transfer learning-based approach for automated kidney segmentation on multiparametric MRI sequences.

Purpose: Multiparametric magnetic resonance imaging (mp-MRI) is being investigated for kidney cancer because of better soft tissue contrast ability. The necessity of manual labels makes the development of supervised kidney segmentation algorithms challenging for each mp-MRI protocol. Here, we developed a transfer learning-based approach to improve kidney segmentation on a small dataset of five other mp-MRI sequences. Approach: We proposed a fully automated two-dimensional (2D) attention U-Net model for kidney segmentation on T1 weighted-nephrographic phase contrast enhanced (CE)-MRI (T1W-NG) dataset ( N = 108 ). The pretrained weights of T1W-NG kidney segmentation model transferred to five other distinct mp-MRI sequences model (T2W, T1W-in-phase (T1W-IP), T1W-out-of-phase (T1W-OP), T1W precontrast (T1W-PRE), and T1W-corticomedullary-CE (T1W-CM), N = 50 ) and fine-tuned by unfreezing the layers. The individual model performances were evaluated with and without transfer-learning fivefold cross-validation on average Dice similarity coefficient (DSC), absolute volume difference, Hausdorff distance (HD), and center-of-mass distance (CD) between algorithm generated and manually segmented kidneys. Results: The developed 2D attention U-Net model for T1W-NG produced kidney segmentation DSC of 89.34 ± 5.31 % . Compared with randomly initialized weight models, the transfer learning-based models of five mp-MRI sequences showed average increase of 2.96% in DSC of kidney segmentation ( p = 0.001 to 0.006). Specifically, the transfer-learning approach increased average DSC on T2W from 87.19% to 89.90%, T1W-IP from 83.64% to 85.42%, T1W-OP from 79.35% to 83.66%, T1W-PRE from 82.05% to 85.94%, and T1W-CM from 85.65% to 87.64%. Conclusions: We demonstrate that a pretrained model for automated kidney segmentation of one mp-MRI sequence improved automated kidney segmentation on five other additional sequences.

Progression Rates of LR-2 and LR-3 Observations on MRI to Higher LI-RADS Categories in Patients at High Risk of Hepatocellular Carcinoma: A Retrospective Study.

BACKGROUND. Reported rates of hepatocellular carcinoma (HCC) for LR-2 and LR-3 observations are generally greater than those expected on the basis of clinical experience, possibly reflecting some studies' requirement for pathologic reference. OBJECTIVE. The purpose of this study was to determine rates of progression to higher LI-RADS categories of LR-2 and LR-3 observations in patients at high risk of HCC. METHODS. This retrospective study included 91 patients (64 men, 27 women; mean age, 62 years) at high risk of HCC who had clinically reported LR-2 (n = 55) or LR-3 (n = 36) observations on MRI who also underwent follow-up CT or MRI at least 12 months after the observation was made. A study coordinator annotated the location of a single LR-2 or LR-3 observation per patient on the basis of the clinical reports. Using LI-RADS version 2018 criteria, two radiologists independently assigned LI-RADS categories on the follow-up examinations. Progression rates from LR-2 or LR-3 to higher categories were determined. A post hoc consensus review was performed of observations that progressed to LR-4 or LR-5. Subgroup analyses were performed with respect to presence of prior HCC (n = 34) or a separate baseline LR-5 observation (n = 12). RESULTS. For LR-2 observations, the rate of progression to LR-4 was 0.0% (95% CI, 0.0-6.7%) and to LR-5 was 3.6% (95% CI, 0.4-13.1%) for both readers. For LR-3 observations, the rate of progression to LR-4 was 22.2% (95% CI, 9.6-43.8%) and to LR-5 was 11.1% (95% CI, 3.0-28.4%) for both readers. Fourteen observations progressed to LR-4 or LR-5 for both readers. Post hoc analysis revealed no instances of progression from LR-2 to LR-4; two, from LR-2 to LR-5; eight, from LR-3 to LR-4; and four, from LR-3 to LR-5. The progression rate from LR-3 to LR-5 was higher (p < .001) among patients with (100.0%) than those without (3.0%) a separate baseline LR-5 observation for both readers. The progression rate from LR-2 to LR-5 was not associated with a separate baseline LR-5 observation for either reader (p = .30). Progression rates were not different (p > .05) between patients with versus those without prior HCC. CONCLUSION. On the basis of progression to LR-4 or LR-5, LR-2 and LR-3 observations had lower progression rates than reported in studies incorporating pathology results in determining progression. CLINICAL IMPACT. The findings refine understanding of the clinical significance of LR-2 and LR-3 observations.