Response to Letter: Intrapatient Comparison of the Hepatobiliary Phase of Gd-BOPTA and Gd-EOB-DTPA in the Differentiation of HCA From FNH.

LEVEL OF EVIDENCE: 5 TECHNICAL EFFICACY STAGE: 3 J. Magn. Reson. Imaging 2020;52:1281-1282.

Performance of Contrast-Enhanced Sonography Versus MRI With a Liver-Specific Contrast Agent for Diagnosis of Hepatocellular Adenoma and Focal Nodular Hyperplasia.

OBJECTIVE. The purpose of this article is to compare contrast-enhanced sonography (CEUS) with sulfur hexafluoride with MRI with the liver-specific contrast agent gadobenate dimeglumine in the diagnosis of hepatocellular adenoma (HCA) and focal nodular hyperplasia (FNH) in a cohort of consecutive patients. MATERIALS AND METHODS. Patients referred to a tertiary center for hepatobiliary disease who had suspected HCA or FNH on MRI performed with an extracellular gadolinium-based contrast agent underwent a prospective workup including CEUS and MRI with a liver-specific contrast agent. Diagnosis was definite when the findings of CEUS and MRI with a liver-specific contrast agent were concordant; histopathologic examination (HPE) was performed for cases with discordant findings. Descriptive statistics and the association between categoric variables were presented as numbers and percentages and were assessed using the Fisher exact test. The primary analysis was patient based. Sensitivity, specificity, and AUC and predictive values for the diagnosis of HCA and FNH were calculated separately for CEUS and MRI with a liver-specific contrast agent. RESULTS. A total of 181 patients were selected for the first analysis. Findings from CEUS and MRI with a liver-specific contrast agent were concordant for 132 patients (73%) and discordant for 49 (27%). HPE was performed for 26 of the 49 patients with discordant findings (53%), with findings indeterminate for two of these patients, the findings of MRI with a liver-specific contrast agent correct for 21 of the remaining 24 patients (87.5%), and the findings of CEUS correct for three of these 24 patients (12.5%) (p < 0.05). For further analysis, 156 patients with concordant findings or HPE-proven cases were included. For CEUS, the sensitivity and specificity for the diagnosis of HCA and FNH were 85% and 87%, respectively; the ROC AUC value was 0.856; and the positive predictive value and negative predictive value were 79% and 90%, respectively. For MRI with a liver-specific contrast agent, the sensitivity and specificity were 95% each, the ROC AUC value was 0.949, and the positive predictive value and negative predictive value were 92% and 97%, respectively, for the diagnosis of HCA and FNH. CONCLUSION. The findings of CEUS and MRI with a liver-specific contrast agent showed fair agreement for the diagnosis of HCA and FNH. MRI with a liver-specific contrast agent is diagnostically correct significantly more often than CEUS in cases with discordant findings that are HPE proven.

Intrapatient Comparison of the Hepatobiliary Phase of Gd-BOPTA and Gd-EOB-DTPA in the Differentiation of Hepatocellular Adenoma From Focal Nodular Hyperplasia.

BACKGROUND: Current imaging guidelines do not specify the preferred hepatobiliary contrast agent when differentiating hepatocellular adenoma (HCA) from focal nodular hyperplasia (FNH) on MRI. PURPOSE: To analyze intrapatient differences in the hepatobiliary phase (HBP) after use of both gadobenate dimeglumine (Gd-BOPTA) and gadoxetic acid (Gd-EOB-DTPA)-enhanced MRI to differentiate HCA from FNH. STUDY TYPE: Retrospective. POPULATION: Patients who underwent both Gd-BOPTA and Gd-EOB-DTPA-enhanced MRI, including 33 patients with 82 lesions (67 HCA; 15 FNH), with a step-down reference standard of pathology, 20% regression, identical appearance to earlier biopsied lesions, and stringent imaging findings. FIELD STRENGTH/SEQUENCE: 1.5T and 3T HBP of Gd-BOPTA and Gd-EOB-DTPA-enhanced MRI, precontrast fat-suppressed T1 -weighted sequence. ASSESSMENT: Signal intensities relative to the surrounding liver in the HBP were assessed by two observers. STATISTICAL TESTS: Sensitivity and specificity of HCA diagnosis were calculated for both contrast agents. Interobserver agreement was evaluated using Cohen's kappa; differences in degree of certainty for scoring a lesion were calculated by means of the Wilcoxon signed rank test. Differences in signal intensity between Gd-BOPTA and Gd-EOB-DTPA were calculated using McNemar's test. RESULTS: Almost perfect agreement was found between observers for scored signal intensities with both contrast agents. In 30 of the 82 lesions (37%) a difference was observed between contrast agents in the HBP, with Gd-EOB-DTPA proving correct in all but one of the discordant lesions. When distinguishing HCA from FNH, Gd-BOPTA showed a sensitivity of 46% (31/67) and a specificity of 87% (13/15), while the sensitivity and specificity of Gd-EOB-DTPA was 85% (57/67) and 100% (15/15), respectively. A risk of misclassifying HCA as FNH typically occurs for Gd-BOPTA when lesions are intrinsically hyperintense (P < 0.005). DATA CONCLUSION: The HBP of Gd-EOB-DTPA shows superior accuracy in ruling out HCA in comparison with Gd-BOPTA, especially when the lesion is intrinsically hyperintense on T1 -weighted imaging. LEVEL OF EVIDENCE: 3 Technical Efficacy: Stage 3 J. Magn. Reson. Imaging 2019;49:700-710.

Primary Cystic Lesions of the Retrorectal Space: MRI Evaluation and Clinical Assessment.

OBJECTIVE: The purpose of this study was to assess the a priori chance that primary cystic lesions of the retrorectal space are malignant and to investigate MRI characteristics that indicate malignancy. MATERIALS AND METHODS: Patients referred to a center for colorectal surgery were recruited from 2000 to 2014. Lesions were proven by clinical assessment and histopathology. MRI was performed at 1.5 T with examinations evaluated by two radiologists. Interobserver agreement was assessed (Cohen kappa) and differences between malignant and benign lesions calculated (Fisher exact test). RESULTS: Twenty-eight patients (22 women, six men; age range, 18-70 years) with 31 lesions were included. Lesions were categorized as tailgut cysts (n = 16, 52%), teratomas (n = 9, 29%), lesions of colorectal origin (n = 4, 13%), or neurogenic lesions (n = 2, 6%). Five patients (18%) had malignant lesions. Colorectal lesions had the highest percentage of malignancy (3/4, 75%). A solid tissue component was found in all five (100%) malignant lesions and two (8%) of the benign lesions, which were both teratomas (p < 0.05). Sensitivity and specificity for malignancy according to the presence of a solid tissue component was 100% (5/5) and 92% (24/26). For unilocularity, multilocularity, debris, septa, and wall thickening, differences were not significant. Interobserver agreement was excellent (κ = 1) for all characteristics except debris (κ = 0.795). CONCLUSION: The majority of retrorectal cystic lesions are benign. The presence of a solid tissue component should raise suspicion for malignancy.

Natural Language Processing in Radiology: A Systematic Review.

Radiological reporting has generated large quantities of digital content within the electronic health record, which is potentially a valuable source of information for improving clinical care and supporting research. Although radiology reports are stored for communication and documentation of diagnostic imaging, harnessing their potential requires efficient and automated information extraction: they exist mainly as free-text clinical narrative, from which it is a major challenge to obtain structured data. Natural language processing (NLP) provides techniques that aid the conversion of text into a structured representation, and thus enables computers to derive meaning from human (ie, natural language) input. Used on radiology reports, NLP techniques enable automatic identification and extraction of information. By exploring the various purposes for their use, this review examines how radiology benefits from NLP. A systematic literature search identified 67 relevant publications describing NLP methods that support practical applications in radiology. This review takes a close look at the individual studies in terms of tasks (ie, the extracted information), the NLP methodology and tools used, and their application purpose and performance results. Additionally, limitations, future challenges, and requirements for advancing NLP in radiology will be discussed.

Aortic calcifications on routine preoperative chest X-ray and perioperative stroke during cardiac surgery: a nested matched case-control study.

OBJECTIVES: Perioperative stroke in cardiac surgery is most often embolic in origin. Preoperative chest X-ray (CXR) is routinely used amongst others to screen for relevant aortic calcification, a potential source of embolic stroke. We performed a nested matched case-control study to examine the relationship between aortic calcifications on CXR and the occurrence of embolic stroke. METHODS: Among all consecutive patients undergoing cardiac surgery in our hospital between January 2014 and July 2017, we selected all patients with perioperative embolic stroke (cases). Controls, all patients without perioperative stroke, were matched on age, sex and type of surgery. All preoperative CXRs were scored for aortic calcifications (none, mild, severe) in the ascending aorta, arch, aortic knob and descending aorta. RESULTS: Out of the 3038 eligible patients, 27 cases were detected and 78 controls were selected. In the stroke group, mild-to-severe calcifications were found least often in the ascending aorta [9% of patients, 95% confidence interval (CI) 1-29%] and most frequently in the aortic knob (63% of patients, 95% CI 44-78%). The distribution of aortic calcification was comparable in cases versus controls. CONCLUSIONS: Calcification burden was comparable between the cases and their matched controls. In our study population, CXR findings on aortic calcification were not related to the risk of embolic stroke. In these patients, the correlation between findings on the preoperative CXR and the risk of stroke might, therefore, be too weak to use the CXR directly for the assessment of the safety of manipulating the ascending aorta during surgery.

Towards patient-related information needs.

The quality of health care depends, among other factors, on the quality of a physician's domain knowledge. Since it is impossible to keep up with all new findings and developments, physicians usually have gaps in their domain knowledge. To handle exceptional cases, access to the full range of medical literature is required. The specific literature needed for appropriate treatment of the patient is described by a physician's information need. Physicians are often unaware of their information needs. To support them, this paper presents a first step towards automatically formulating patient-related information needs. We start investigating how we can model a physician's information needs in general. Then we propose an approach to instantiate the model into a representation of a physician's information needs using the patient data as stored in a medical record. Our experiments show that this approach is feasible. Since the number of formulated patient-related information needs is rather high, it has to be reduced. To reduce the number of formulated information needs we propose the use of additional knowledge. Four types of knowledge are discussed, viz. (a) knowledge about temporal aspects, (b) domain knowledge, (c) knowledge about a physician's specialism, and (d) a user model. Future research has to clarify which type of knowledge (or combination thereof) is most appropriate for our purpose. It is expected that the resultant set of information needs will have a manageable size and contributes to the quality of health care.

Avoiding literature overload in the medical domain.

The retrieval of patient-related literature is hampered by the large size of medical literature. Various computer systems have been developed to assist physicians during information retrieval. However, in general, physicians lack the time and skills required to employ such systems effectively. Our goal is to investigate to what extent a physician can be provided with patient-related literature without spending extra time and without acquiring additional skills. In previous research we developed a method to formulate a physician's patient-related information needs automatically, without requiring any interaction between the physician and the system. The formulated information needs can be used as a starting point for literature retrieval. As a result we found that the number of information needs formulated per physician was quite high and had to be reduced to avoid a literature overload. In this paper we present four types of knowledge that may be used to accomplish a reduction in the number of information needs. The usefulness of each of these knowledge types depends heavily on the specific cause underlying the multitude of information needs. To determine the nature of the cause, we performed an experimental analysis. The results of the analysis led us to conclude that the knowledge types can be ordered according to their appropriateness as follows: (1) knowledge concerning temporal aspects, (2) knowledge concerning a physician's specialism, (3) domain knowledge, and (4) a user model. Further research has to be performed, in particular on precisely assessing the performance of each type of knowledge within our domain.