Improving detection performance of hepatocellular carcinoma and interobserver agreement for liver imaging reporting and data system on CT using deep learning reconstruction.

PURPOSE: This study aimed to compare the hepatocellular carcinoma (HCC) detection performance, interobserver agreement for Liver Imaging Reporting and Data System (LI-RADS) categories, and image quality between deep learning reconstruction (DLR) and conventional hybrid iterative reconstruction (Hybrid IR) in CT. METHODS: This retrospective study included patients who underwent abdominal dynamic contrast-enhanced CT between October 2021 and March 2022. Arterial, portal, and delayed phase images were reconstructed using DLR and Hybrid IR. Two blinded readers independently read the image sets with detecting HCCs, scoring LI-RADS, and evaluating image quality. RESULTS: A total of 26 patients with HCC (mean age, 73 years ± 12.3) and 23 patients without HCC (mean age, 66 years ± 14.7) were included. The figures of merit (FOM) for the jackknife alternative free-response receiver operating characteristic analysis in detecting HCC averaged for the readers were 0.925 (reader 1, 0.937; reader 2, 0.913) in DLR and 0.878 (reader 1, 0.904; reader 2, 0.851) in Hybrid IR, and the FOM in DLR were significantly higher than that in Hybrid IR (p = 0.038). The interobserver agreement (Cohen's weighted kappa statistics) for LI-RADS categories was moderate for DLR (0.595; 95% CI, 0.585-0.605) and significantly superior to Hybrid IR (0.568; 95% CI, 0.553-0.582). According to both readers, DLR was significantly superior to Hybrid IR in terms of image quality (p ≤ 0.021). CONCLUSION: DLR improved HCC detection, interobserver agreement for LI-RADS categories, and image quality in evaluations of HCC compared to Hybrid IR in abdominal dynamic contrast-enhanced CT.

The automatic diagnosis artificial intelligence system for preoperative magnetic resonance imaging of uterine sarcoma.

OBJECTIVE: Magnetic resonance imaging (MRI) is efficient for the diagnosis of preoperative uterine sarcoma; however, misdiagnoses may occur. In this study, we developed a new artificial intelligence (AI) system to overcome the limitations of requiring specialists to manually process datasets and a large amount of computer resources. METHODS: The AI system comprises a tumor image filter, which extracts MRI slices containing tumors, and sarcoma evaluator, which diagnoses uterine sarcomas. We used 15 types of MRI patient sequences to train deep neural network (DNN) models used by tumor filter and sarcoma evaluator with 8 cross-validation sets. We implemented tumor filter and sarcoma evaluator using ensemble prediction technique with 9 DNN models. Ten tumor filters and sarcoma evaluator sets were developed to evaluate fluctuation accuracy. Finally, AutoDiag-AI was used to evaluate the new validation dataset, including 8 cases of sarcomas and 24 leiomyomas. RESULTS: Tumor image filter and sarcoma evaluator accuracies were 92.68% and 90.50%, respectively. AutoDiag-AI with the original dataset accuracy was 89.32%, with 90.47% sensitivity and 88.95% specificity, whereas AutoDiag-AI with the new validation dataset accuracy was 92.44%, with 92.25% sensitivity and 92.50% specificity. CONCLUSION: Our newly established AI system automatically extracts tumor sites from MRI images and diagnoses them as uterine sarcomas without human intervention. Its accuracy is comparable to that of a radiologist. With further validation, the system could be applied for diagnosis of other diseases. Further improvement of the system's accuracy may enable its clinical application in the future.

Detection of left ventricular dysfunction on axial non-contrast chest CT.

PURPOSE: To evaluate the performance of axial non-contrast CT images in detecting systolic and diastolic left ventricular (LV) dysfunction. METHOD: This single-center retrospective study assessed 178 participants who underwent transthoracic ultrasonography and had non-contrast chest CT data within three months of ultrasonography. The patients were divided into LV systolic dysfunction (<52% ejection fraction in men and <54% in women), LV diastolic dysfunction (at least three of the following four criteria were met: average E/e' ratio >14; septal e' <7 cm/s or lateral e' <10 cm/s; peak tricuspid regurgitation velocity >2.8 m/s; or left atrial maximum volume index >34 ml/m2), and normal LV function groups. CT parameters were evaluated as predictive factors for LV dysfunction. These parameters were: I, maximum minor axis diameter of the LV lumen; II, I plus myocardial wall thickness; III, maximum left atrium anteroposterior diameter; IV, maximum transverse cardiac diameter; V, myocardial wall thickness; I-IV divided by maximum medial thoracic diameter; and I-IV divided by anteroposterior thoracic diameter. All parameters were measured on axial images: diameters were maximized. RESULTS: LV systolic dysfunction was indicated when parameter IV exceeded 131.2 mm with sensitivity and specificity of 71.8% and 77.0%, respectively. Moreover, LV diastolic dysfunction was indicated when parameter III divided by anteroposterior thoracic diameter exceeded 0.165 with sensitivity and specificity of 77.8% and 83.0%, respectively. CONCLUSIONS: Axial non-contrast chest CT is useful for detecting LV dysfunction. Even CT scans for other purposes provide LV function information that may lead to appropriate examination.

Preoperative portal vein embolization: a comparison of ethanol and coils versus ethanol alone.

PURPOSE: To compare the efficacy and safety of preoperative portal vein embolization (PVE) with ethanol and coils versus ethanol alone. MATERIAL AND METHODS: Between April 2014 and May 2019, 45 patients underwent right preoperative PVE with ethanol and coils (n = 19; EthCo group) or ethanol alone (n = 26; Eth group). RESULTS: The change in % future liver remnant (FLR) was not significantly different between the EthCo and Eth groups (11.2 ± 4.3% versus 11.3 ± 4.1%, p = .98). Less ethanol was used in the EthCo group (9.7 ± 3.5 mL versus 11.9 ± 4.4 mL, p = .02). Recanalization was observed in eight patients only in the Eth group (p < .01). There were no differences in the pre-/post-PVE laboratory data between the two groups, except for post-PVE albumin. The volume of ethanol used was positively correlated with the post-PVE total bilirubin (p = .01), aspartate aminotransferase (AST) (p < .01) and alanine aminotransferase (ALT) (p < .01) levels. CONCLUSION: The efficacy of PVE did not differ between the EthCo and Eth groups. The use of ethanol and coils was associated with less recanalization and liver damage compared with ethanol alone.

Rib fracture detection in computed tomography images using deep convolutional neural networks.

ABSTRACT: To evaluate the rib fracture detection performance in computed tomography (CT) images using a software based on a deep convolutional neural network (DCNN) and compare it with the rib fracture diagnostic performance of doctors.We included CT images from 39 patients with thoracic injuries who underwent CT scans. In these images, 256 rib fractures were detected by two radiologists. This result was defined as the gold standard. The performances of rib fracture detection by the software and two interns were compared via the McNemar test and the jackknife alternative free-response receiver operating characteristic (JAFROC) analysis.The sensitivity of the DCNN software was significantly higher than those of both Intern A (0.645 vs 0.313; P < .001) and Intern B (0.645 vs 0.258; P < .001). Based on the JAFROC analysis, the differences in the figure-of-merits between the results obtained via the DCNN software and those by Interns A and B were 0.057 (95% confidence interval: -0.081, 0.195) and 0.071 (-0.082, 0.224), respectively. As the non-inferiority margin was set to -0.10, the DCNN software is non-inferior to the rib fracture detection performed by both interns.In the detection of rib fractures, detection by the DCNN software could be an alternative to the interpretation performed by doctors who do not have intensive training experience in image interpretation.

Metal artifacts reduction in computed tomography: A phantom study to compare the effectiveness of metal artifact reduction algorithm, model-based iterative reconstruction, and virtual monochromatic imaging.

The purpose of this study was to compare the effectiveness of a metal artifact reduction algorithm (MAR), model-based iterative reconstruction (MBIR), and virtual monochromatic imaging (VMI) for reducing metal artifacts in CT imaging.A phantom study was performed for quantitatively evaluating the dark bands and fine streak artifacts generated by unilateral hip prostheses. Images were obtained by conventional scanning at 120 kilovolt peak, and reconstructed using filtered back projection, MAR, and MBIR. Furthermore, virtual monochromatic images (VMIs) at 70 kilo-electron volts (keV) and 140 keV with/without use of MAR were obtained by dual-energy CT. The extents and mean CT values of the dark bands and the differences in the standard deviations and location parameters of the fine streak artifacts evaluated by the Gumbel method in the images obtained by each of the methods were statistically compared by analyses of variance.Significant reduction of the extent of the dark bands was observed in the images reconstructed using MAR than in those not reconstructed using MAR (all, P < .01). Images obtained by VMI at 70 keV and 140 keV with use of MAR showed significantly increased mean CT values of the dark bands as compared to those obtained by reconstructions without use of MAR (all, <.01). Significant reduction of the difference in the standard deviations used to evaluate fine streak artifacts was observed in each of the image sets obtained with VMI at 140 keV with/without MAR and conventional CT with MBIR as compared to the images obtained using other methods (all, P < .05), except between VMI at 140 keV without MAR and conventional CT with MAR. The location parameter to evaluate fine streak artifacts was significantly reduced in CT images obtained using MBIR and in images obtained by VMI at 140 keV with/without MAR as compared to those obtained using other reconstruction methods (all, P < .01).In our present study, MAR appeared to be the most effective reconstruction method for reducing dark bands in CT images, and MBIR and VMI at 140 keV appeared to the most effective for reducing streak artifacts.