Ultra-High-Resolution T2-Weighted PROPELLER MRI of the Rectum With Deep Learning Reconstruction: Assessment of Image Quality and Diagnostic Performance.

OBJECTIVE: The aim of this study was to evaluate the impact of ultra-high-resolution acquisition and deep learning reconstruction (DLR) on the image quality and diagnostic performance of T2-weighted periodically rotated overlapping parallel lines with enhanced reconstruction (PROPELLER) imaging of the rectum. MATERIALS AND METHODS: This prospective study included 34 patients who underwent magnetic resonance imaging (MRI) for initial staging or restaging of rectal tumors. The following 4 types of oblique axial PROPELLER images perpendicular to the tumor were obtained: a standard 3-mm slice thickness with conventional reconstruction (3-CR) and DLR (3-DLR), and 1.2-mm slice thickness with CR (1.2-CR) and DLR (1.2-DLR). Three radiologists independently evaluated the image quality and tumor extent by using a 5-point scoring system. Diagnostic accuracy was evaluated in 22 patients with rectal cancer who underwent surgery after MRI without additional neoadjuvant therapy (median interval between MRI and surgery, 22 days). The signal-to-noise ratio and tissue contrast were measured on the 4 types of PROPELLER imaging. RESULTS: 1.2-DLR imaging showed the best sharpness, overall image quality, and rectal and lesion conspicuity for all readers (P < 0.01). Of the assigned scores for tumor extent, extramural venous invasion (EMVI) scores showed moderate agreement across the 4 types of PROPELLER sequences in all readers (intraclass correlation coefficient, 0.60-0.71). Compared with 3-CR imaging, the number of cases with MRI-detected extramural tumor spread was significantly higher with 1.2-DLR imaging (19.0 ± 2.9 vs 23.3 ± 0.9, P = 0.03), and the number of cases with MRI-detected EMVI was significantly increased with 1.2-CR, 3-DLR, and 1.2-DLR imaging (8.0 ± 0.0 vs 9.7 ± 0.5, 11.0 ± 2.2, and 12.3 ± 1.7, respectively; P = 0.02). For the diagnosis of histopathologic extramural tumor spread, 3-CR and 1.2-CR had significantly higher specificity than 3-DLR and 1.2-DLR imaging (0.75 and 0.78 vs 0.64 and 0.58, respectively; P = 0.02), and only 1.2-CR had significantly higher accuracy than 3-CR imaging (0.83 vs 0.79, P = 0.01). The accuracy of MRI-detected EMVI with reference to pathological EMVI was significantly lower for 3-CR and 3-DLR compared with 1.2-CR (0.77 and 0.74 vs 0.85, respectively; P < 0.01), and was not significantly different between 1.2-CR and 1.2-DLR (0.85 vs 0.80). Using any pathological venous invasion as the reference standard, the accuracy of MRI-detected EMVI was significantly the highest with 1.2-DLR, followed by 1.2-CR, 3-CR, and 3-DLR (0.71 vs 0.67 vs 0.59 vs 0.56, respectively; P < 0.01). The signal-to-noise ratio was significantly highest with 3-DLR imaging (P < 0.05). There were no significant differences in tumor-to-muscle contrast between the 4 types of PROPELLER imaging. CONCLUSIONS: Ultra-high-resolution PROPELLER T2-weighted imaging of the rectum combined with DLR improved image quality, increased the number of cases with MRI-detected extramural tumor spread and EMVI, but did not improve diagnostic accuracy with respect to pathology in rectal cancer, possibly because of false-positive MRI findings or false-negative pathologic findings.

BERT-based Transfer Learning in Sentence-level Anatomic Classification of Free-Text Radiology Reports.

Purpose: To assess whether transfer learning with a bidirectional encoder representations from transformers (BERT) model, pretrained on a clinical corpus, can perform sentence-level anatomic classification of free-text radiology reports, even for anatomic classes with few positive examples. Materials and Methods: This retrospective study included radiology reports of patients who underwent whole-body PET/CT imaging from December 2005 to December 2020. Each sentence in these reports (6272 sentences) was labeled by two annotators according to body part ("brain," "head & neck," "chest," "abdomen," "limbs," "spine," or "others"). The BERT-based transfer learning approach was compared with two baseline machine learning approaches: bidirectional long short-term memory (BiLSTM) and the count-based method. Area under the precision-recall curve (AUPRC) and area under the receiver operating characteristic curve (AUC) were computed for each approach, and AUCs were compared using the DeLong test. Results: The BERT-based approach achieved a macro-averaged AUPRC of 0.88 for classification, outperforming the baselines. AUC results for BERT were significantly higher than those of BiLSTM for all classes and those of the count-based method for the "brain," "chest," "abdomen," and "others" classes (P values < .025). AUPRC results for BERT were superior to those of baselines even for classes with few labeled training data (brain: BERT, 0.95, BiLSTM, 0.11, count based, 0.41; limbs: BERT, 0.74, BiLSTM, 0.28, count based, 0.46; spine: BERT, 0.82, BiLSTM, 0.53, count based, 0.69). Conclusion: The BERT-based transfer learning approach outperformed the BiLSTM and count-based approaches in sentence-level anatomic classification of free-text radiology reports, even for anatomic classes with few labeled training data.Keywords: Anatomy, Comparative Studies, Technology Assessment, Transfer Learning Supplemental material is available for this article. © RSNA, 2023.

Assessment of Characteristics of Imaging Biomarkers for Quantifying Anterior Cingulate Cortex Changes: A Twin Study of Middle- to Advanced-Aged Populations in East Asia.

Background and Objectives: Our aim was to assess genetic and environmental effects on surface morphological parameters for quantifying anterior cingulate cortex (ACC) changes in middle- to advanced-age East Asians using twin analysis. Materials and Methods: Normal twins over 39 years old comprising 37 monozygotic pairs and 17 dizygotic pairs underwent 3-dimensional (3D) T1-weighted imaging of the brain at 3T. Freesurfer-derived ACC parameters including thickness, standard deviation of thickness (STDthickness), volume, surface area, and sulcal morphological parameters (folding, mean, and Gaussian curvatures) were calculated from 3D T1-weighted volume images. Twin analysis with a model involving phenotype variance components of additive genetic effects (A), common environmental effects (C), and unique environmental effects (E) was performed to assess the magnitude of each genetic and environmental influence on parameters. Results: Most parameters fit best with an AE model. Both thickness (A: left 0.73/right 0.71) and surface area (A: left 0.63/right 0.71) were highly heritable. STDthickness was low to moderately heritable (A: left 0.48/right 0.29). Volume was moderately heritable (A: left 0.37). Folding was low to moderately heritable (A: left 0.44/right 0.28). Mean curvature (A: left 0.37/right 0.65) and Gaussian curvature (A: right 0.79) were moderately to highly heritable. Right volume and left Gaussian curvature fit best with a CE model, indicating a relatively weak contribution of genetic factors to these parameters. Conclusions: When assessing ACC changes in middle- to advanced-age East Asians, one must keep in mind that thickness and surface area appear to be strongly affected by genetic factors, whereas sulcal morphological parameters tend to involve environmental factors.

Assessment of Myometrial Invasion in Premenopausal Grade 1 Endometrial Carcinoma: Is Magnetic Resonance Imaging a Reliable Tool in Selecting Patients for Fertility-Preserving Therapy?

OBJECTIVES: The aim of this study was to evaluate the diagnostic ability of magnetic resonance imaging (MRI) in premenopausal women with G1 endometrial carcinoma. METHODS: Twenty-six patients underwent T2W, diffusion weighted, and dynamic contrast-enhanced 3-T MRI. The degree of myometrial invasion was pathologically classified into no invasion, shallow (3 mm or less), and more. Two radiologists assessed myometrial invasion on MRI. Diagnostic accuracy, sensitivity, specificity, positive and negative predictive values, AUC, and interobserver agreement were analyzed. RESULTS: For assessing myometrial invasion, mean accuracy, sensitivity, specificity, positive predictive values, negative predictive values, and AUC, respectively, were as follows: 63%, 42%, 85%, 79%, 47%, and 0.75. Mean interobserver agreement was fair (k = 0.36). Shallow invasions were underestimated as no invasion on MRI in all 6 cases. CONCLUSIONS: Magnetic resonance imaging produced false-negative result on half of patients. The misjudgments tended to happen in patients with shallow invasion.

Application of non-Gaussian water diffusional kurtosis imaging in the assessment of uterine tumors: A preliminary study.

OBJECTIVES: To evaluate the interobserver reliability and value of diffusional kurtosis imaging (DKI) in the assessment of uterine tumors compared with those of conventional diffusion-weighted imaging (DWI). METHODS: This retrospective study was approved by our institutional review board, which waived the requirement for informed consent. Fifty-eight women (mean age: 55.0 ± 13.6 years; range: 30-89 years) with suspected malignant uterine tumors underwent 3-T magnetic resonance imaging using DKI and DWI. Twelve had coexisting leiomyoma. Two observers analyzed region-of-interest measurements of diffusivity (D), kurtosis (K), and the apparent diffusion coefficient (ADC) of uterine lesions and healthy adjacent tissues. Interobserver agreement was evaluated using the intra-class correlation coefficient (ICC). The mean values were compared using one-way analysis of variance with a post-hoc Tukey's honestly significant difference test. The diagnostic accuracy of D and ADC in differentiating malignant tumors from benign leiomyomas was analyzed using receiver operating characteristic (ROC) analysis. RESULTS: The ICCs between the two observers in evaluating D, K, and the ADC of the malignant tumors were higher than 0.84, suggesting excellent interobserver agreements. The mean D (×10-3 mm2/s) of uterine cancers (1.05 ± 0.41 and 1.09 ± 0.40 for observers 1 and 2, respectively) were significantly lower than those of leiomyoma (1.40 ± 0.37 and 1.56 ± 0.33, respectively; P < 0.05), healthy myometrium (1.72 ± 0.27 and 1.69 ± 0.30, respectively; P < 0.001), and healthy endometrium (1.53 ± 0.35 and 1.42 ± 0.37, respectively; P < 0.005). There was no significant difference in the area under the ROC curve between D and ADC. The mean K of uterine cancers (0.88 ± 0.28 and 0.90 ± 0.23, respectively) were higher than those of myometrium (0.72 ± 0.10 and 0.73 ± 0.10, respectively; P < 0.001), healthy endometrium (0.65 ± 0.13 and 0.60 ± 0.18, respectively; P < 0.001), and leiomyoma (0.76 ± 0.14 and 0.77 ± 0.16, respectively; not significant, P > 0.1). CONCLUSIONS: Interobserver agreements in evaluating D, K, and ADC were moderate to excellent. D performed equally to conventional DWI in differentiating between benign and malignant uterine lesions. The mean K of malignant uterine lesions was significantly higher than that of non-tumorous myometrium or endometrium.

Assessment of Mediastinal Tumors Using SUVmax and Volumetric Parameters on FDG-PET/CT.

OBJECTIVES: This study aimed to evaluate the role of pretreatment SUVmax and volumetric FDG positron emission tomography (PET) parameters in the differentiation between benign and malignant mediastinal tumors. In addition, we investigated whether pretreatment SUVmax and volumetric FDG-PET parameters could distinguish thymomas from thymic carcinomas, and low-risk from high-risk thymomas. METHODS: This study was conducted on 52 patients with mediastinal tumors undergoing FDG-PET/CT. Histological examination indicated that 29 mediastinal tumors were benign, and 23 cases were malignant. To obtain quantitative PET/CT parameters, we determined the maximum standardized uptake value (SUVmax), volumetric parameters, metabolic tumor volume (MTV), and total lesion glycolysis (TLG) for primary tumors using SUVmax cut-off value of 2.5. SUVmax, MTV and TLG of benign and malignant tumors were compared using the Mann-Whitney U test. Moreover, receiver-operating curve (ROC) analysis was applied to identify the cut-off values of SUVmax, MTV and TLG for the accurate differentiation of benign and malignant tumors. SUVmax, MTV and TLG were compared between thymomas and thymic carcinomas, as well as low-risk and high-risk thymomas. RESULTS: Mean SUVmax, MTV and TLG of malignant mediastinal tumors were significantly higher compared to benign tumors (P<0.001). Sensitivity, specificity, accuracy, positive predictive value, and negative predictive value of SUVmax were 78.2%, 86.2%, 82.6%, 81.8%, and 83.3%, respectively. These values were estimated at 82.6%, 96.6%, 90.4%, 95%, and 87.5% for MTV and TLG, respectively. Additionally, optimal cut-off values for the differentiation of benign and malignant mediastinal tumors were determined at 4.2 and 22.3 mL and 79.7 g for SUVmax, MTV and TLG, respectively. Mean SUVmax, MTV and TLG of thymic carcinomas were significantly higher compared to thymomas (P<0.01), while no significant differences were observed in the mean quantitative parameters between low-risk and high-risk thymomas. CONCLUSION: Although SUVmax, MTV and TLG could not distinguish between low-risk and high-risk thymomas, these parameters might be able to differentiate benign tumors from malignant mediastinal tumors noninvasively. These parameters could be used to distinguish between thymomas and thymic carcinomas as well. Therefore, FDG-PET/CT parameters seem to be accurate indices for the detection of malignant mediastinal tumors.

Cost-Effectiveness Analysis of Percutaneous Sclerotherapy for Venous Malformations.

PURPOSE: To assess cost-effectiveness of sclerotherapy for venous malformations (VMs) to improve patient quality of life (QOL). MATERIALS AND METHODS: This prospective study enrolled 28 patients with symptomatic VMs who underwent sclerotherapy. EuroQol-5 Dimension (EQ-5D) and Short-Form 36 (SF-36) Health Survey were used to measure health-related QOL. Questionnaires were collected before and 1, 3, 6, and 12 months after sclerotherapy. Quality-adjusted life years (QALYs) were calculated using EQ-5D score as a measure of health utility. Medical costs obtained from the hospital accounting system and other costs of staff, drugs, materials, and angiographic equipment were calculated for each procedure. Cost-effectiveness was analyzed using incremental cost-effectiveness ratio (ICER) as the medical cost/gain of QALYs. RESULTS: Median EQ-5D scores improved from 0.768 (range, 0.705-1) to 1 (range, 0.768-1) after 6 months (P = .023) and 1 (range, 0.768-1) after 12 months (P = .063). The gain of QALYs at 12 months was 0.043. The mean medical cost was ¥281,228 ($2,337). The pain group (baseline bodily pain scale of SF-36 score < 70) showed greater improvement in median EQ-5D score, from 0.705 (range, 0.661-0.768) to 0.768 (range, 0.705-1) after 6 months (P = .041) and 0.768 (range, 0.768-1) after 12 months (P = .049). ICER at 12 months was ¥6,600,483 ($54,840) in the overall group and decreased to ¥3,998,113 ($33,218) in the pain group, < ¥6,000,000 ($49,850), threshold for acceptance of a public health benefit in Japan, even accounting for 50% increase in costs. CONCLUSIONS: Sclerotherapy was cost-effective for improving QOL for symptomatic VMs, especially for patients with moderate to severe pain.

The assessment of collateral communication after hybrid repair for Crawford extent II thoraco-abdominal aortic aneurysms.

OBJECTIVES: The repair of extensive thoraco-abdominal aortic aneurysms (TAAAs) is invasive and carries a high risk for spinal cord injury (SCI). The aim of this study was to assess the early results and collateral circulation to the spinal cord after hybrid repair for Crawford extent II aortic aneurysms. METHODS: Between 1997 and 2013, we performed 128 thoracic endovascular aortic repair (TEVAR) procedures for TAAAs. This study reviews 12 patients who underwent hybrid TEVAR for a Crawford extent II aortic aneurysm (mean age: 56 years, 6 men, chronic dissection: 10). Aortic arch repair was performed to create a proximal landing zone and visceral debranching bypass was performed to create a distal landing zone at separate stages prior to TEVAR. Subsequently, a stent graft was deployed to cover the residual downstream aorta. TEVAR was generally performed the day after the final debranching procedure. Cerebrospinal fluid drainage was performed, and the mean blood pressure was maintained at >90 mmHg in all cases. RESULTS: The median operation time for TEVAR was 94 min (range: 71-421 min) and the mean blood loss was 300 ml (range: 130-1350 ml). No SCI or in-hospital death was observed after TEVAR. Multidetector computed tomography identified three arteries (subclavian artery, external iliac artery and internal iliac artery) providing collateral circulation to spinal segmental arteries (SAs). In all cases, mid-thoracic SAs (Th5-8) and low lumbar SAs (L2-5) were fed by the subclavian artery and the internal iliac artery, respectively. Additionally, low thoracic to high lumbar SAs (Th9-L1) communicated with the subclavian artery via the lateral thoracic wall and/or the external iliac artery via the abdominal wall. CONCLUSIONS: We achieved satisfactory early and mid-term outcomes with hybrid repair for Crawford extent II TAAAs. Furthermore, collateral circulation to SAs was maintained during and after TEVAR regardless of the extent of the aortic repair.

Assessment of combination of contrast-enhanced magnetic resonance imaging and positron emission tomography/computed tomography for evaluation of ovarian masses.

OBJECTIVES: The objectives of this study were to correlate fluorodeoxyglucose uptake in ovarian masses on positron emission tomography/computed tomography (PET/CT) with pathological grades of malignancy and subtypes and to determine the appropriate approach for combining PET/CT and contrast-enhanced magnetic resonance imaging (CE-MRI) to characterize ovarian masses. MATERIALS AND METHODS: A retrospective study was conducted including 127 patients who underwent surgical resection of an ovarian mass (30 benign, 31 borderline, 66 malignant). Maximum standardized uptake values (SUVmax) obtained with PET/CT were compared between pathological grades of malignancy and subtypes. Two radiologists each independently conducted a blind evaluation of CE-MRI for all lesions and classified them by the grade of malignancy as determinate (benign, borderline, or malignant) or indeterminate and by subtype as mucinous or nonmucinous. The appropriate approach for combining CE-MRI and PET/CT was determined by comparing the combined diagnostic ability with that of CE-MRI alone. RESULTS: The SUVmax of malignant tumors was significantly higher than that of benign and borderline lesions (mean, 7.8, 1.7, 2.4; P < 0.05). Among malignant tumors, SUVmax was significantly lower in mucinous adenocarcinomas compared with nonmucinous malignant tumors (mean, 3.3, 8.4; P < 0.05) and lower in clear cell adenocarcinomas compared with other subtypes of nonmucinous malignant tumors (mean, 6.0, 9.4; P < 0.05). The SUVmax cutoff that best differentiated malignant lesions from benign/borderline lesions was 2.4 for mucinous and 4.0 for nonmucinous tumors. These cutoffs correctly classified lesions as malignant or not in 88.2% of cases (112/127). When PET/CT was combined with CE-MRI, the readers correctly classified 85% (34/40) and 86.5% (32/37) of indeterminate lesions on CE-MRI. However, PET/CT was not useful for classifying determinate lesions on CE-MRI, particularly because PET/CT correctly classified only 70.1% (12/17) of clear cell adenocarcinomas, whereas CE-MRI alone correctly classified 94.1% (1617). Thus, compared with CE-MRI alone, the diagnostic accuracy of CE-MRI + PET/CT when PET/CT was added only for indeterminate lesions on CE-MRI was significantly higher for both readers for differentiating between benign and borderline/malignant (P < 0.05), as well as between benign/borderline and malignant (P < 0.01). CONCLUSION: Fluorodeoxyglucose uptake in ovarian masses correlates with pathological subtypes as well as the grade of malignancy. Furthermore, the combination of CE-MRI and PET/CT is a highly accurate method for characterizing ovarian masses because PET/CT can be used as a complement to classify indeterminate lesions as malignant or not based on appropriate cutoff SUVmax for mucinous and nonmucinous tumors.

Automated assessment of malignant degree of small peripheral adenocarcinomas using volumetric CT data: correlation with pathologic prognostic factors.

PURPOSE: To evaluate a custom-developed software for analyzing malignant degrees of small peripheral adenocarcinomas on volumetric CT data compared to pathological prognostic factors. MATERIALS AND METHODS: Forty-six adenocarcinomas with a diameter of 2cm or less from 46 patients were included. The custom-developed software can calculate the volumetric rates of solid parts to whole nodules even though solid parts show a punctate distribution, and automatically classify nodules into the following six types according to the volumetric rates of solid parts: type 1, pure ground-glass opacity (GGO); type 2, semiconsolidation; type 3, small solid part with a GGO halo; type 4, mixed type with an area that consisted of GGO and solid parts which have air-bronchogram or show a punctate distribution; type 5, large solid part with a GGO halo; and type 6, pure solid type. The boundary between solid portion and GGO on CT was decided using two threshold selection methods for segmenting gray-scale images. A radiologist also examined two-dimensional rates of solid parts to total opacity (2D%solid) which was already confirmed with previous reports. RESULTS: There were good agreements between the classification determined by the software and radiologists (weighted kappa=0.778-0.804). Multivariate logistic regression analyses showed that both 2D%solid and computer-automated classification were significantly useful in estimating lymphatic invasion (p=0.0007, 0.0027), vascular invasion (p=0.003, 0.012), and pleural invasion (p=0.021, 0.025). CONCLUSION: Using our custom-developed software, it is feasible to predict the pathological prognostic factors of small peripheral adenocarcinomas.

Preoperative assessment of thymic veins on multidetector row CT: optimization of contrast material volume.

PURPOSE: Preoperative localization of the thymic veins is considered important to prevent intraoperative severe bleeding prior to video-assisted thoracoscopic thymectomy. The purpose of this study was to determine the optimal dose of contrast material for preoperative CT imaging for the detection of thymic veins on the basis of patient weight. MATERIALS AND METHODS: The records of 31 patients who underwent thymectomy were examined retrospectively. All patients were scanned using an eight-channel multidetector-row computed tomography (CT) scanner at 1.25 mm collimation and a 0.625-mm reconstruction interval. CT scans were obtained after injection of 300 mg I/ml nonionic contrast material at a rate of 2 ml/s. A 90-ml contrast bolus was used for the first 16 consecutive patients (group I), and a 150-ml bolus was used for the following 15 patients (group II). The scan delay was 60 s and 90 s in groups I and II respectively. Two independent radiologists who were blinded to the surgical results evaluated the number of thymic veins observed on preoperative CT, which was later correlated with the actual number of thymic veins clipped during surgery. The responses were analyzed with respect to contrast amount by single bolus and per kilogram of body weight. RESULTS: Thymic veins were correctly detected in 9 of 16 (56%) patients in group I and 14 of 15 (93%) patients in group II. Thymic vein detection was significantly better in patients who received the >or=2.0 ml/kg contrast medium compared to those who received the 1.00-1.99 ml/kg medium (P < 0.05). CONCLUSION: An intravenous contrast material volume of 2 ml/kg (300 mg I/ml) is appropriate for the identification of thymic veins on prethymectomy CT.

Coronal multiplanar reconstruction view from isotropic voxel data sets obtained with multidetector-row CT: assessment of detection and size of mediastinal and hilar lymph nodes.

OBJECTIVE: To assess the detection and size of mediastinal and hilar lymph nodes by multiplanar reconstruction (MPR) view from isotropic voxel data sets obtained with multidetector-row computed tomography (MDCT). MATERIALS AND METHODS: Thin-section CT of 27 patients with mediastinal or hilar lymph node swelling was obtained with a 25.6-cm FOV, 512 x 512 matrix, and two protocols: A) 0.5-mm collimation, 0.3-mm interval, and B) 2-mm collimation, 1-mm interval. MPR views with a 0.5-mm slice thickness were obtained from these two data sets. Postcontrast axial CT used 5-mm collimation (set C). Two observers evaluated the presence and cranio-caudal length of swollen lymph nodes. Two other board-certified chest radiologists evaluated all three sets and established a gold standard by consensus. RESULTS: The accuracy of detection was 76%, 73%, and 68% for sets A, B, and C, respectively. There was a significant difference between sets A and C (McNemar's test: p<0.05) but not between sets A and B or B and C (p>0.05). The cranio-caudal length of lymph nodes was significantly correlated with the gold standard only in set A (Pearson's correlation coefficient: r=0.53, p<0.05). CONCLUSION: Non-contrast enhanced coronal MPR views constructed from isotropic voxel data sets may be substituted for axial enhanced CT for the evaluation of mediastinal and hilar lymph nodes.