ACTA Radiologica (Test)

Acta Radiologica, Ahead of Print. <br/>BackgroundPulmonary nodules are an early imaging indication of lung cancer, and early detection of pulmonary nodules can improve the prognosis of lung cancer. As one of the applications of machine learning, the convolutional neural network (CNN) applied to computed tomography (CT) imaging data improves the accuracy of diagnosis, but the results could be more consistent.PurposeTo evaluate the diagnostic performance of CNN in assisting in detecting pulmonary nodules in CT images.Material and MethodsPubMed, Cochrane Library, Web of Science, Elsevier, CNKI and Wanfang databases were systematically retrieved before 30 April 2023. Two reviewers searched and checked the full text of articles that might meet the criteria. The reference criteria are joint diagnoses by experienced physicians. The pooled sensitivity, specificity and the area under the summary receiver operating characteristic curve (AUC) were calculated by a random-effects model. Meta-regression analysis was performed to explore potential sources of heterogeneity.ResultsTwenty-six studies were included in this meta-analysis, involving 2,391,702 regions of interest, comprising segmented images with a few wide pixels. The combined sensitivity and specificity values of the CNN model in detecting pulmonary nodules were 0.93 and 0.95, respectively. The pooled diagnostic odds ratio was 291. The AUC was 0.98. There was heterogeneity in sensitivity and specificity among the studies. The results suggested that data sources, pretreatment methods, reconstruction slice thickness, population source and locality might contribute to the heterogeneity of these eligible studies.ConclusionThe CNN model can be a valuable diagnostic tool with high accuracy in detecting pulmonary nodules.

Acta Radiologica, Ahead of Print. <br/>BackgroundVascular malformations are rare diseases that should be treated in dedicated vascular anomaly centers (VAC). There is only a small amount of data on the diagnostic and therapeutic handling of these patients, before they are referred to a VAC.PurposeTo demonstrate the disease-specific patient characteristics in a German VAC, which are required to determine diagnostic and therapeutic steps.Material and MethodsIn a retrospective study, all patients who were treated in the VAC from April 2014 until August 2021 were identified. In total, 593 patients were included in this study.ResultsAlmost all patients had previously consulted a physician (591/593, 99.7%). A mean of two different physicians had been consulted (range 0–10). Patients with more complex, syndromal vascular malformations had significantly more previous appointments (P = 0.0018). In only 44% (261/593) of patients, the referral diagnosis was made correctly. Most patients had been previously treated for their vascular anomaly: pharmacotherapy (n = 130; 21.9%), compression garments (n = 141; 23.8%), surgical resection (n = 80; 17.3%) and sclerotherapy (n = 68; 11.5%). Fifty-two patients who had been falsely diagnosed had also received therapy prior to their referral to the VAC (8.8%). Most patients received an ultrasound examination in the VAC (n = 464; 78.2%). Most frequently, compression therapy was prescribed (n = 256; 43.2%), followed by sclerotherapy (n = 175, 29.5%) and pharmacotherapy (n = 55; 9.3%).ConclusionPatients suffering from vascular anomalies often go through a complicated scheduling with numerous outpatient appointments and have a high risk of misdiagnosis and mistreatment prolonging the medical condition. Therefore, patients with vascular anomalies should be treated in a dedicated vascular anomaly center.

Acta Radiologica, Ahead of Print. <br/>BackgroundComputed tomography angiography (CTA) is a reliable, non-invasive screening method for diagnosing panvascular disease. By using low contrast agent volume, CTA imaging enables one-stop multi-organ scanning, thereby minimizing the potential risk of contrast-induced nephropathy in patients with impaired renal function.PurposeTo evaluate the feasibility of one-stop CTA following a heart rate (HR)-based protocol using a low volume of contrast medium (CM) for examination of the coronary, carotid and cerebrovascular arteries.Material and MethodsSixty patients undergoing coronary carotid, and cerebrovascular CTA after a single injection of CM were recruited and randomly divided into two groups. Group A (n = 30) underwent CTA following a traditional protocol. The timing of the scans in Group B (n = 30) was determined according to the patient's HR.ResultsThe CT values for the thoracic aorta (432.2 ± 104.28 HU), anterior cerebral artery (303.96 ± 99.29 HU), and right coronary artery (366.70 ± 85.10 HU) in Group A did not differ significantly from those in Group B (445.80 ± 106.13, 293.73 ± 75.25 and 344.13 ± 111.04 HU, respectively). The qualities of most of the scanned images for both groups were scored as 3 or 4 (on a five-point scale). The radiation dose and the volume of CM were significantly higher in Group A (303.05 ± 110.95 mGy) (100 mL) than in Group B (239.46 ± 101.12 mGy) (50 mL).ConclusionThe radiation dose and volume of CM were significantly reduced in CTA following the HR-based protocol. The personalized administration of CM also simplified the scanning process.

Acta Radiologica, Ahead of Print. <br/>BackgroundThe 2021 World Health Organization (WHO) classification considers a histological low grade glioma with specific molecular characteristics as molecular glioblastoma (mGBM). Accurate identification of mGBM will aid in risk stratification of glioma patients.PurposeTo explore the value of machine learning models based on magnetic resonance imaging (MRI) radiomics features in predicting mGBM.Material and MethodsIn total, 166 patients histologically diagnosed as low-grade diffuse glioma (WHO II and III) were included in the study. Fifty-three cases were reclassified as mGBM based on molecular status. Four dimensionality reduction methods including distance correlation (DC), gradient boosted decision tree (GBDT), least absolute shrinkage and selection operator (LASSO) and minimal redundancy maximal relevance (MRMR) were used to select the optimal signatures. Six machine learning algorithms including support vector machine (SVM), linear discriminant analysis (LDA), neural network (NN), logistic regression (LR), K-nearest neighbour (KNN) and decision tree (DT) were used to develop the classifiers. The relative SD was used to evaluate the stability of the models, and the area under the curve values in the independent test group were used to evaluate their performances.ResultsNN_DC was determined as the optimal classifier due to the highest area under the curve of 0.891 in the test group. The classification accuracy, sensitivity, specificity, positive predictive value and negative predictive value of NN_DC were 0.915, 0.842, 0.950, 0.889 and 0.927, respectively.ConclusionMachine learning models can predict mGBM non-invasively, which may help to develop personalized treatment strategies for neurosurgeons and provide an effective tool for accurate stratification in clinical trials.

Acta Radiologica, Ahead of Print. <br/>BackgroundAs guidelines endorse the use of computed tomography (CT) for examining coronary artery disease (CAD), it is important to compare the advantages and disadvantages of the novel photon counting detector CT (PCD-CT) technology with the established energy integrating detector CT (EID-CT).PurposeTo compare the image quality of coronary computed tomography angiography (CCTA) and the Agatston scores (AS) derived from EID-CT and PCD-CT.Material and MethodsIn this prospective observational study, 28 patients underwent clinical calcium score and CCTA scans on an EID-CT and a PCD-CT scanner. CCTA images were qualitatively analyzed by five observers using visual grading characteristics. The correlation and agreement of the AS were assessed using Spearman's rank correlation and Bland–Altman plots.ResultsThis qualitative analyses demonstrated a high fraction of “good” or “excellent” ratings for the image criteria in both CT systems. The sharpness of the distal lumen and image quality regarding motion artifacts were rated significantly higher for EID-CT (P < 0.05). However, the sharpness of coronary calcification was rated significantly higher for PCD-CT (P < 0.05). Spearman's rank correlation and Bland–Altman plots showed good correlation (P = 0.95) and agreement regarding the AS between EID-CT and PCD-CT.ConclusionBoth CT systems exhibited high CCTA image quality. The sharpness of calcifications was rated significantly higher for PCD-CT. A good correlation was observed between the AS derived from the two systems.

Acta Radiologica, Ahead of Print. <br/>BackgroundNon-invasive determination of H3 K27 alteration of pediatric brainstem glioma (pedBSG) remains a clinical challenge.PurposeTo predict H3 K27-altered pedBSG using amide proton transfer-weighted (APTw) imaging.Material and MethodsThis retrospective study included patients with pedBSG who underwent APTw imaging and had the H3 K27 alteration status determined by immunohistochemical staining. The presence or absence of foci of markedly increased APTw signal in the lesion was visually assessed. Quantitative APTw histogram parameters within the entire solid portion of tumors were extracted and compared between H3 K27-altered and wild-type groups using Student's t-test. The ability of APTw for differential diagnosis was evaluated using logistic regression.ResultsSixty pedBSG patients included 48 patients with H3 K27-altered tumor (aged 2–48 years) and 12 patients with wild-type tumor (aged 3–53 years). Visual assessment showed that the foci of markedly increased APTw signal intensity were more common in the H3 K27-altered group than in wild-type group (60% vs. 16%, P = 0.007). Histogram parameters of APTw signal intensity in the H3 K27-altered group were significantly higher than those in the wild-type group (median, 2.74% vs. 2.22%, P = 0.02). The maximum (area under the receiver operating characteristic curve [AUC] = 0.72, P = 0.01) showed the highest diagnostic performance among histogram analysis. A combination of age, median and maximum APTw signal intensity could predict H3 K27 alteration with a sensitivity of 81%, specificity of 75% and AUC of 0.80.ConclusionAPTw imaging may serve as an imaging biomarker for H3 K27 alteration of pedBSGs.

Acta Radiologica, Ahead of Print. <br/>BackgroundVarious versions of artificial intelligence (AI) have been used as a diagnostic tool aid in the diagnosis of breast cancer. One of the most important problems in breast screening progmrams is interval breast cancer (IBC).PurposeTo compare the diagnostic performance of Transpara v1.6 and v1.7 in the detection of IBC.Material and MethodsReports of screening mammograms of a total 2,248,665 of women were evaluated retrospectively. Of 2,129,486 mammograms reported as Breast Imaging Reporting and Data System (BIRADS) 1 and 2, the IBC group consisted of 323 cases who were diagnosed as having cancer on mammography and were correlated with pathology in second mammogram taken >30 days after first mammogram. Four hundred and forty-one were defined as the control group because they did not change over 2 years. Cancer risk scores of both groups were determined from 1 to 10 with Tranpara v1.6 and v1.7. Diagnostic performances of both versions were evaluated by the receiver operating characteristic curve.ResultsCancer risk scores 1 and 10 in v1.7 increased compared to v1.6 (P < 0.001). In all cases, sensitivity for v1.6 was 56.6%, specificity was 90%, and, for v1.7, sensitivity was 65.9% and specificity was 90%, respectively. In all cases, area under the curve values were 0.812 for v1.6 and 0.856 for v1.7, which was higher in v1.7 (P < 0.001). Diagnostic performance of v1.7 was higher than v1.6 at the 7–12-month period (P < 0.001).ConclusionThe present study showed that Tranpara v1.7 has a higher specificity, sensitivity and diagnostic performance in IBC determination than v1.6. AI systems can be used in breast screening as a secondary or third reader in screening programs.

Acta Radiologica, Ahead of Print. <br/>BackgroundMagnetic resonance imaging (MRI) is effective in diagnosing deltoid ligament (DL) injury but its sensitivity in chronic cases is low. Additional diagnostic signs are required to reduce the risk of a false negative diagnosis.PurposeTo evaluate the added diagnostic value of bone marrow edema at the ligament insertion (BMELI) of DL to the MRI assessment of chronic DL injury.Material and MethodsOne hundred patients who consecutively came to our institution between November 2018 and December 2021 and underwent arthroscopic surgery for chronic ankle instability (CAI) were enrolled in the present study. Preoperative MR images were retrospectively reviewed by two orthopedic surgeons to evaluate the sensitivity, specificity and interobserver reliability of three MRI signs in diagnosing chronic DL injury, namely, abnormal ligamentous morphological characteristics (ALMC), BMELI and medial clear space (MCS).ResultsTaking arthroscopy as the reference standard, there were 34 patients with and 66 without DL injury. ALMC had 64.71% (22/34; 46.47–79.70) sensitivity and 83.33% (55/66; 71.71–91.00) specificity, BMELI had 70.59% (24/34; 52.33–84.29) sensitivity and 95.45% (63/66; 86.44–98.82) specificity and MCS had 26.47% (9/34; 13.51–44.65) sensitivity and 92.42% (61/66; 82.50–97.18) specificity. Compared with ALMC, BMELI had similar efficacy in superficial cases (P = 0.06) and greater efficacy in deep cases (P = 0.04). All three signs showed good interobserver agreement (kappa values all above 0.7).ConclusionBMELI can reliably indicate concomitant injury to the DL in CAI patients. Using BMELI as a sign of chronic DL injury when ALMC is unclear may reduce the risk of a false negative diagnosis.

Acta Radiologica, Ahead of Print. <br/>BackgroundTomographic 3D ultrasound (t3DUS) is a promise imaging technique for quantifying carotid plaque through measuring the degree of stenosis and plaque volume. Carotid plaque volume (CPV) could add benefit in predicting the potential risk of stroke.PurposeTo assess the reproducibility and accuracy of t3DUS for measuring CPV within the internal carotid artery in patients undergoing carotid endarterectomy.Material and Methodst3DUS was used to obtain CPV in vivo from 25 symptomatic patients prior to surgery. Ex vivo CPV from the carotid endarterectomy specimen was then measured using a validated saline displacement method as a reference standard. CPV for each patient was measured twice using both methods (total n = 50 per technique). Intraclass correlation coefficient (ICC) and Bland–Altman plot were used to establish bias and limit of agreement between CPV measurements.ResultsThere was an excellent agreement between t3DUS and reference test with respect to measuring CPV with an ICC value of 0.98 (95% confidence interval = 0.97–0.99, P < 0.001). Bias in measurements was 0.02 ± 0.11 cm3 (95% limit of agreement = −0.19 to 0.25). Intra-observer agreement of t3DUS CPV measurements was excellent with an ICC value of 0.95 (95% confidence interval = 0.92–0.97, P < 0.001). Bias in measurements was 0.004 ± 0.07 cm3 (95% limit of agreement = −0.14 to 0.15).Conclusiont3DUS is a reproducible imaging method and showed excellent agreement with the reference standard with respect to measuring CPV. These findings suggest that t3DUS has the potential to be a valuable non-invasive tool for assessing carotid plaque burden and predicting the risk of stroke.

Acta Radiologica, Ahead of Print. <br/>