Iron overload is a systemic disorder and is either primary (genetic) or secondary (exogenous iron administration). Primary iron overload is most commonly associated with hereditary hemochromatosis and secondary iron overload with ineffective erythropoiesis (predominantly caused by β-thalassemia major and sickle cell disease) that requires long-term transfusion therapy, leading to transfusional hemosiderosis. Iron overload may lead to liver cirrhosis and hepatocellular carcinoma, in addition to cardiac and endocrine complications. The liver is one of the main iron storage organs and the first to show iron overload. Therefore, detection and quantification of liver iron overload are critical to initiate treatment and prevent complications. Liver biopsy was the historical reference standard for detection and quantification of liver iron content. Magnetic resonance (MR) imaging is now commonly used for liver iron quantification, including assessment of distribution, detection, grading, and monitoring of treatment response in iron overload. Several MR imaging techniques have been developed for iron quantification, each with advantages and limitations. The liver-to-muscle signal intensity ratio technique is simple and widely available; however, it assumes that the reference tissue is normal. Transverse magnetization (also known as R2) relaxometry is validated but is prone to respiratory motion artifacts due to a long acquisition time, is presently available only for 1.5-T imaging, and requires additional cost and delay for off-line analysis. The R2* technique has fast acquisition time, demonstrates a wide range of liver iron content, and is available for 1.5-T and 3.0-T imaging but requires additional postprocessing software. Quantitative susceptibility mapping has the highest sensitivity for detecting iron deposition; however, it is still investigational, and the correlation with liver iron content is not yet established. RSNA, 2018.

Purpose To evaluate the performance of major features, ancillary features, and categories of Liver Imaging Reporting and Data System (LI-RADS) version 2014 at magnetic resonance (MR) imaging for the diagnosis of hepatocellular carcinoma (HCC). Materials and Methods This retrospective institutional review board-approved study included patients with liver MR imaging and at least one pathologically proved lesion. Between 2004 and 2016, 102 patients (275 observations including 113 HCCs) met inclusion criteria. Two radiologists independently assessed major and ancillary imaging features for each liver observation and assigned a LI-RADS category. Per-lesion estimates of diagnostic performance of major features, ancillary features, and LI-RADS categories were assessed by using generalized estimating equation models. Results Major features (arterial phase hyperenhancement, washout, capsule, and threshold growth) had a sensitivity of 88.5%, 60.6%, 32.9%, and 41.6%, and a specificity of 18.6%, 84.8%, 98.8%, and 83.2% for HCC, respectively. Ancillary features (mild-moderate T2 hyperintensity, restricted diffusion, mosaic architecture, intralesional fat, lesional fat sparing, blood products, and subthreshold growth) had a sensitivity of 62.2%, 54.8%, 9.9%, 30.9%, 23.1%, 2.8%, and 48.3%, and a specificity of 79.4%, 90.6%, 99.4%, 94.2%, 83.1%, 99.3%, and 91.4% for HCC, respectively. The LR-5 or LR-5 V categories had a per-lesion sensitivity of 50.8% and a specificity of 95.8% for HCC, respectively. The LR-4, LR-5, or LR-5 V categories (determined by using major features only vs combination of major and ancillary features) had a per-lesion sensitivity of 75.9% and 87.9% and a per-lesion specificity of 87.5% and 86.2%, respectively. Conclusion The use of ancillary features in combination with major features increases the sensitivity while preserving a high specificity for the diagnosis of HCC.

Interest for deep learning in radiology has increased tremendously in the past decade due to the high achievable performance for various computer vision tasks such as detection, segmentation, classification, monitoring, and prediction. This article provides step-by-step practical guidance for conducting a project that involves deep learning in radiology, from defining specifications, to deployment and scaling. Specifically, the objectives of this article are to provide an overview of clinical use cases of deep learning, describe the composition of multi-disciplinary team, and summarize current approaches to patient, data, model, and hardware selection. Key ideas will be illustrated by examples from a prototypical project on imaging of colorectal liver metastasis. This article illustrates the workflow for liver lesion detection, segmentation, classification, monitoring, and prediction of tumor recurrence and patient survival. Challenges are discussed, including ethical considerations, cohorting, data collection, anonymization, and availability of expert annotations. The practical guidance may be adapted to any project that requires automated medical image analysis.

No abstract

No abstract