Histogram Analysis of Apparent Diffusion Coefficient Maps Provides Genotypic and Pretreatment Phenotypic Information in Pediatric and Young Adult Rhabdomyosarcoma.

INTRODUCTION: We evaluate the role of apparent diffusion coefficient (ADC) histogram metrics in stratifying pediatric and young adult rhabdomyosarcomas. METHODS: We retrospectively evaluated baseline diffusion-weighted imaging (DWI) from 38 patients with rhabdomyosarcomas (Not otherwise specified: 2; Embryonal: 21; Spindle Cell: 2; Alveolar: 13, mean ± std dev age: 8.1 ± 7.76 years). The diffusion images were obtained on a wide range of 1.5 T and 3 T scanners at multiple sites. FOXO1 fusion status was available for 35 patients, nine of whom harbored the fusion. 13 patients were TNM stage 1, eight had stage 2 disease, nine were stage 3, and eight had stage 4 disease. 23 patients belonged to Clinical Group III and seven to Group IV, while two and five were CG I and II, respectively. Nine patients were classified as low risk, while 21 and five were classified as intermediate and high risk respectively. Histogram parameters of the apparent diffusion coefficient (ADC) map from the entire tumor were obtained based on manual tumor contouring. A two-tailed Mann-Whitney U test was used for all two-group, and the Kruskal-Wallis's test was used for multiple-group comparisons. Bootstrapped receiver operating characteristic (ROC) curves and areas under the curve (AUC) were generated for the statistically significant histogram parameters to differentiate genotypic and phenotypic parameters. RESULTS: Alveolar rhabdomyosarcomas had a statistically significant lower 10th Percentile (586.54 ± 164.52, mean ± std dev, values are in ×10-6mm2/s) than embryonal rhabdomyosarcomas (966.51 ± 481.33) with an AUC of 0.85 (95%CI. 0.73-0.95) for differentiating the two. The 10th percentile was also significantly different between FOXO1 fusion-positive (553.87 ± 187.64) and negative (898.07 ± 449.38) rhabdomyosarcomas with an AUC of 0.83 (95% CI 0.71-0.94). Alveolar rhabdomyosarcomas also had statistically significant lower Mean, Median, and Root Mean Squared ADC histogram values than embryonal rhabdomyosarcomas. Four, five, and seven of the 18 histogram parameters evaluated demonstrated a statistically significant increase with higher TNM stage, clinical group, assignment, and pretreatment risk stratification, respectively. For example, Entropy had an AUC of 0.8 (95% CI. 0.67-0.92) for differentiating TNM stage 1 from ≥ stage 2 and 0.9 (95% CI. 0.8-0.98) for differentiating low from intermediate or high-risk stratification. CONCLUSION: Our findings demonstrate the potential of ADC histogram metrics to predict clinically relevant variables for rhabdomyosarcoma, including FOXO1 fusion status, histopathology, Clinical Group, TNM staging, and risk stratification.

Arterial spin-labeled (ASL) perfusion in children with Sturge-Weber syndrome: a retrospective cross-sectional study.

PURPOSE: Sturge-Weber syndrome (SWS) is a developmental disorder with venous hypertension and associated tissue responses including pial angiomatosis, cortical calcifications, and cerebral atrophy. Arterial spin-labeled (ASL) perfusion is an advanced MR sequence which can assess perfusion, without the need for contrast. We systematically evaluated the potential benefits of using ASL in Sturge-Weber syndrome, to determine the extent of intracranial perfusion abnormality and stage of disease, relevant for prognostication and surgical planning. METHODS: Two pediatric neuroradiologists retrospectively evaluated ASL perfusion imaging of 31 children with confirmed SWS and recorded the presence of hyper-perfusion, hypo-perfusion, or normal perfusion. The presence and distribution of ASL abnormality were compared against the presence and side of atrophy/calcification and pial angiomatosis on standard MR sequences. RESULTS: Thirty-one children (52% female, median age 16.7 months) with SWS had ASL imaging. Seven (23%) had hyper-perfusion, 15 (48%) had hypo-perfusion, and 9 (29%) had no perfusion abnormalities. ASL perfusion abnormality matched the location of SWS findings on conventional imaging in 86% (19/22). ASL demonstrated statistically significant increased perfusion in the early stage of the disease and decreased perfusion when there was atrophy. The parietal lobe was involved in 86% of cases. CONCLUSION: ASL perfusion imaging is an advanced technique which may contribute to earlier diagnosis and more accurate prognostication of Sturge-Weber syndrome, helping guide management and potential surgical planning.

Role of MRI radiomics for the prediction of MYCN amplification in neuroblastomas.

OBJECTIVES: We evaluate MR radiomics and develop machine learning-based classifiers to predict MYCN amplification in neuroblastomas. METHODS: A total of 120 patients with neuroblastomas and baseline MR imaging examination available were identified of whom 74 (mean age ± standard deviation [SD] of 6 years and 2 months ± 4 years and 9 months; 43 females and 31 males, 14 MYCN amplified) underwent imaging at our institution. This was therefore used to develop radiomics models. The model was tested in a cohort of children with the same diagnosis but imaged elsewhere (n = 46, mean age ± SD: 5 years 11 months ± 3 years 9 months, 26 females and 14 MYCN amplified). Whole tumour volumes of interest were adopted to extract first-order histogram and second-order radiomics features. Interclass correlation coefficient and maximum relevance and minimum redundancy algorithm were applied for feature selection. Logistic regression, support vector machine, and random forest were employed as the classifiers. Receiver operating characteristic (ROC) analysis was performed to evaluate the diagnostic accuracy of the classifiers on the external test set. RESULTS: The logistic regression model and the random forest both showed an AUC of 0.75. The support vector machine classifier obtained an AUC of 0.78 on the test set with a sensitivity of 64% and a specificity of 72%. CONCLUSION: The study provides preliminary retrospective evidence demonstrating the feasibility of MRI radiomics in predicting MYCN amplification in neuroblastomas. Future studies are needed to explore the correlation between other imaging features and genetic markers and to develop multiclass predictive models. KEY POINTS: • MYCN amplification in neuroblastomas is an important determinant of disease prognosis. • Radiomics analysis of pre-treatment MR examinations can be used to predict MYCN amplification in neuroblastomas. • Radiomics machine learning models showed good generalisability to external test set, demonstrating reproducibility of the computational models.

Automating Angle Measurements on Foot Radiographs in Young Children: Feasibility and Performance of a Convolutional Neural Network Model.

Measurement of angles on foot radiographs is an important step in the evaluation of malalignment. The objective is to develop a CNN model to measure angles on radiographs, using radiologists' measurements as the reference standard. This IRB-approved retrospective study included 450 radiographs from 216 patients (< 3 years of age). Angles were automatically measured by means of image segmentation followed by angle calculation, according to Simon's approach for measuring pediatric foot angles. A multiclass U-Net model with a ResNet-34 backbone was used for segmentation. Two pediatric radiologists independently measured anteroposterior and lateral talocalcaneal and talo-1st metatarsal angles using the test dataset and recorded the time used for each study. Intraclass correlation coefficients (ICC) were used to compare angle and paired Wilcoxon signed-rank test to compare time between radiologists and the CNN model. There was high spatial overlap between manual and CNN-based automatic segmentations with dice coefficients ranging between 0.81 (lateral 1st metatarsal) and 0.94 (lateral calcaneus). Agreement was higher for angles on the lateral view when compared to the AP view, between radiologists (ICC: 0.93-0.95, 0.85-0.92, respectively) and between radiologists' mean and CNN calculated (ICC: 0.71-0.73, 0.41-0.52, respectively). Automated angle calculation was significantly faster when compared to radiologists' manual measurements (3 ± 2 vs 114 ± 24 s, respectively; P < 0.001). A CNN model can selectively segment immature ossification centers and automatically calculate angles with a high spatial overlap and moderate to substantial agreement when compared to manual methods, and 39 times faster.

A Patch-Based Deep Learning Approach for Detecting Rib Fractures on Frontal Radiographs in Young Children.

Chest radiography is the modality of choice for the identification of rib fractures in young children and there is value for the development of computer-aided rib fracture detection in this age group. However, the automated identification of rib fractures on chest radiographs can be challenging due to the need for high spatial resolution in deep learning frameworks. A patch-based deep learning algorithm was developed to automatically detect rib fractures on frontal chest radiographs in children under 2 years old. A total of 845 chest radiographs of children 0-2 years old (median: 4 months old) were manually segmented for rib fractures by radiologists and served as the ground-truth labels. Image analysis utilized a patch-based sliding-window technique, to meet the high-resolution requirements for fracture detection. Standard transfer learning techniques used ResNet-50 and ResNet-18 architectures. Area-under-curve for precision-recall (AUC-PR) and receiver-operating-characteristic (AUC-ROC), along with patch and whole-image classification metrics, were reported. On the test patches, the ResNet-50 model showed AUC-PR and AUC-ROC of 0.25 and 0.77, respectively, and the ResNet-18 showed an AUC-PR of 0.32 and AUC-ROC of 0.76. On the whole-radiograph level, the ResNet-50 had an AUC-ROC of 0.74 with 88% sensitivity and 43% specificity in identifying rib fractures, and the ResNet-18 had an AUC-ROC of 0.75 with 75% sensitivity and 60% specificity in identifying rib fractures. This work demonstrates the utility of patch-based analysis for detection of rib fractures in children under 2 years old. Future work with large cohorts of multi-institutional data will improve the generalizability of these findings to patients with suspicion of child abuse.

Deep learning-based prediction of rib fracture presence in frontal radiographs of children under two years of age: a proof-of-concept study.

OBJECTIVE: In this proof-of-concept study, we aimed to develop deep-learning-based classifiers to identify rib fractures on frontal chest radiographs in children under 2 years of age. METHODS: This retrospective study included 1311 frontal chest radiographs (radiographs with rib fractures, n = 653) from 1231 unique patients (median age: 4 m). Patients with more than one radiograph were included only in the training set. A binary classification was performed to identify the presence or absence of rib fractures using transfer learning and Resnet-50 and DenseNet-121 architectures. The area under the receiver operating characteristic curve (AUC-ROC) was reported. Gradient-weighted class activation mapping was used to highlight the region most relevant to the deep learning models' predictions. RESULTS: On the validation set, the ResNet-50 and DenseNet-121 models obtained an AUC-ROC of 0.89 and 0.88, respectively. On the test set, the ResNet-50 model demonstrated an AUC-ROC of 0.84 with a sensitivity of 81% and specificity of 70%. The DenseNet-50 model obtained an AUC of 0.82 with 72% sensitivity and 79% specificity. CONCLUSION: In this proof-of-concept study, a deep learning-based approach enabled the automatic detection of rib fractures in chest radiographs of young children with performances comparable to pediatric radiologists. Further evaluation of this approach on large multi-institutional data sets is needed to assess the generalizability of our results. ADVANCES IN KNOWLEDGE: In this proof-of-concept study, a deep learning-based approach performed well in identifying chest radiographs with rib fractures. These findings provide further impetus to develop deep learning algorithms for identifying rib fractures in children, especially those with suspected physical abuse or non-accidental trauma.

Artificial Intelligence Curriculum Needs Assessment for a Pediatric Radiology Fellowship Program: What, How, and Why?

RATIONALE AND OBJECTIVES: Artificial intelligence (AI) holds enormous potential for improvements in patient care, efficiency, and innovation in pediatric radiology practice. Although there is a pressing need for a radiology-specific training curriculum and formalized AI teaching, few resources are available. The purpose of our study was to perform a needs assessment for the development of an AI curriculum during pediatric radiology training and continuing education. MATERIALS AND METHODS: A focus group study using a semistructured moderator-guided interview was conducted with radiology trainees' and attending radiologists' perceptions of AI, perceived competence in interpretation of AI literature, and perceived expectations from radiology AI educational programs. The focus group was audio-recorded, transcribed, and thematic analysis was performed. RESULTS: The focus group was held virtually with seven participants. The following themes we identified: (1) AI knowledge, (2) previous training, (3) learning preferences, (4) AI expectations, and (5) AI concerns. The participants had no previous formal training in AI and variability in perceived needs and interests. Most preferred a case-based approach to teaching AI. They expressed incomplete understanding of AI hindered its clinical applicability and reiterated a need for improved training in the interpretation and application of AI literature in their practice. CONCLUSION: We found heterogeneity in perspectives about AI; thus, a curriculum must account for the wide range of these interests and needs. Teaching the interpretation of AI research methods, literature critique, and quality control through implementation of specific scenarios could engage a variety of trainees from different backgrounds and interest levels while ensuring a baseline level of competency in AI.

Rate of Change of Liver Iron Content by MR Imaging Methods: A Comparison Study.

Objective: Magnetic resonance imaging (MRI) can accurately quantify liver iron concentration (LIC), eliminating the need for an invasive liver biopsy. Currently, the most widely used relaxometry methods for iron quantification are R2 and R2*, which are based on T2 and T2* acquisition sequences, respectively. We compared the rate of change of LIC as measured by the R2-based, FDA-approved commercially available third-party software with the rate of change of LIC measured by in-house analysis using R2*-relaxometry-based MR imaging in patients undergoing follow-up MRI scans for liver iron estimation. Methods: We retrospectively included patients who had undergone serial MRIs for liver iron estimation. The MR studies were performed on a 1.5T scanner; standard multi-slice, multi-echo T2- and T2*-based sequences were acquired, and LIC was estimated. The comparison between the rate of change of LIC by R2 and R2* values was performed via correlation coefficients and Bland−Altman difference plots. Results: One hundred and eighty-nine MR abdomen studies for liver iron evaluation from 81 patients (male: 38; female: 43) were included in the study. Fifty-nine patients had two serial scans, eighteen patients had three serial scans, three patients had four serial scans, and one patient had five serial scans. The average time interval between the first and last scans for each patient was 13.3 months. The average rates of change of LIC via R2 and R2* methods were −0.0043 ± 0.0214 and −0.0047 ± 0.012 mg/g per month, respectively. There was no significant difference in the rate of change of LIC observed between the two methods. Linearity between the rate of change of LIC measured by R2 (LIC R2) and R2* (LIC R2*) was strong, showing a correlation coefficient of r = 0.72, p < 0.01. A Bland−Altman plot between the rate of change of the two methods showed that the majority of the plotted variables were between two standard deviations. Conclusion: There was no significant difference in the rate of change of LIC detected between the R2 method and the R2* method that uses a gradient echo (GRE) sequence acquired with breath-hold. Since R2* is relatively faster and less prone to motion artifacts, R2*-derived LIC is recommended for iron homeostasis follow-up in patients with liver iron overload.

Contrast-enhanced multidetector computed tomography features and histogram analysis can differentiate ameloblastomas from central giant cell granulomas.

BACKGROUND: No qualitative or quantitative analysis of contrast-enhanced computed tomography (CT) images has been reported for the differentiation between ameloblastomas and central giant cell granulomas (CGCGs). AIM: To describe differentiating multidetector CT (MDCT) features in CGCGs and ameloblastomas and to compare differences in enhancement of these lesions qualitatively and using histogram analysis. METHODS: MDCT of CGCGs and ameloblastomas was retrospectively reviewed to evaluate qualitative imaging descriptors. Histogram analysis was used to compare the extent of enhancement of the soft tissue. Fisher's exact tests and Mann-Whitney U test were used for statistical analysis (P < 0.05). RESULTS: Twelve CGCGs and 33 ameloblastomas were reviewed. Ameloblastomas had a predilection for the posterior mandible with none of the CGCGs involving the angle. CGCGs were multilocular (58.3%), with a mixed lytic sclerotic appearance (75%). Soft tissue component was present in 91% of CGCGs, which showed hyperenhancement (compared to surrounding muscles) in 50% of cases, while the remaining showed isoenhancement. Matrix mineralization was present in 83.3% of cases. Ameloblastomas presented as a unilocular (66.7%), lytic (60.6%) masses with solid components present in 81.8% of cases. However, the solid component showed isoenhancement in 63%. No matrix mineralization was present in 69.7% of cases. Quantitatively, the enhancement of soft tissue in CGCG was significantly higher than in ameloblastoma on histogram analysis (P < 0.05), with a minimum enhancement of > 49.05 HU in the tumour providing 100% sensitivity and 85% specificity in identifying a CGCG. CONCLUSION: A multilocular, lytic sclerotic lesion with significant hyperenhancement in soft tissue, which spares the angle of the mandible and has matrix mineralization, should indicate prospective diagnosis of CGCG.

The role of apparent diffusion coefficient histogram metrics for differentiating pediatric medulloblastoma histological variants and molecular groups.

BACKGROUND: Medulloblastoma, a high-grade embryonal tumor, is the most common primary brain malignancy in the pediatric population. Molecular medulloblastoma groups have documented clinically and biologically relevant characteristics. Several authors have attempted to differentiate medulloblastoma molecular groups and histology variants using diffusion-weighted imaging (DWI) and apparent diffusion coefficient (ADC) maps. However, literature on the use of ADC histogram analysis in medulloblastomas is still scarce. OBJECTIVE: This study presents data from a sizable group of pediatric patients with medulloblastoma from a single institution to determine the performance of ADC histogram metrics for differentiating medulloblastoma variants and groups based on both histological and molecular features. MATERIALS AND METHODS: In this retrospective study, we evaluated the distribution of absolute and normalized ADC values of medulloblastomas. Tumors were manually segmented and diffusivity metrics calculated on a pixel-by-pixel basis. We calculated a variety of first-order histogram metrics from the ADC maps, including entropy, minimum, 10th percentile, 90th percentile, maximum, mean, median, skewness and kurtosis, to differentiate molecular and histological variants. ADC values of the tumors were also normalized to the bilateral cerebellar cortex and thalami. We used the Kruskal-Wallis and Mann-Whitney U tests to evaluate differences between the groups. We carried out receiver operating characteristic (ROC) curve analysis to evaluate the areas under the curves and to determine the cut-off values for differentiating tumor groups. RESULTS: We found 65 children with confirmed histopathological diagnosis of medulloblastoma. Mean age was 8.3 ± 5.8 years, and 60% (n = 39) were male. One child was excluded because histopathological variant could not be determined. In terms of medulloblastoma variants, tumors were classified as classic (n = 47), desmoplastic/nodular (n = 9), large/cell anaplastic (n = 6) or as having extensive nodularity (n = 2). Seven other children were excluded from the study because of incomplete imaging or equivocal molecular diagnosis. Regarding medulloblastoma molecular groups, there were: wingless (WNT) group (n = 7), sonic hedgehog (SHH) group (n = 14) and non-WNT/non-SHH (n = 36). Our results showed significant differences among the molecular groups in terms of the median (P = 0.002), mean (P = 0.003) and 90th percentile (P = 0.002) ADC histogram metrics. No significant differences among the various medulloblastoma histological variants were found. CONCLUSION: ADC histogram analysis can be implemented as a complementary tool in the preoperative evaluation of medulloblastoma in children. This technique can provide valuable information for differentiating among medulloblastoma molecular groups. ADC histogram metrics can help predict medulloblastoma molecular classification preoperatively.

Application of Apparent Diffusion Coefficient Histogram Metrics for Differentiation of Pediatric Posterior Fossa Tumors : A Large Retrospective Study and Brief Review of Literature.

PURPOSE: This study aimed to evaluate the application of apparent diffusion coefficient (ADC) histogram analysis to differentiate posterior fossa tumors (PFTs) in children. METHODS: A total of 175 pediatric patients with PFT, including 75 pilocytic astrocytomas (PA), 59 medulloblastomas, 16 ependymomas, and 13 atypical teratoid rhabdoid tumors (ATRT), were analyzed. Tumors were visually assessed using DWI trace and conventional MRI images and manually segmented and post-processed using parametric software (pMRI). Furthermore, tumor ADC values were normalized to the thalamus and cerebellar cortex. The following histogram metrics were obtained: entropy, minimum, 10th, and 90th percentiles, maximum, mean, median, skewness, and kurtosis to distinguish the different types of tumors. Kruskal Wallis and Mann-Whitney U tests were used to evaluate the differences. Finally, receiver operating characteristic (ROC) curves were utilized to determine the optimal cut-off values for differentiating the various PFTs. RESULTS: Most ADC histogram metrics showed significant differences between PFTs (p < 0.001) except for entropy, skewness, and kurtosis. There were significant pairwise differences in ADC metrics for PA versus medulloblastoma, PA versus ependymoma, PA versus ATRT, medulloblastoma versus ependymoma, and ependymoma versus ATRT (all p < 0.05). Our results showed no significant differences between medulloblastoma and ATRT. Normalized ADC data showed similar results to the absolute ADC value analysis. ROC curve analysis for normalized ADCmedian values to thalamus showed 94.9% sensitivity (95% CI: 85-100%) and 93.3% specificity (95% CI: 87-100%) for differentiating medulloblastoma from ependymoma. CONCLUSION: ADC histogram metrics can be applied to differentiate most types of posterior fossa tumors in children.

Whole-tumour apparent diffusion coefficient (ADC) histogram analysis to identify MYCN-amplification in neuroblastomas: preliminary results.

OBJECTIVES: To determine the role of apparent diffusion coefficient (ADC) histogram analysis in the identification of MYCN-amplification status in neuroblastomas. METHODS: We retrospectively evaluated imaging records from 62 patients with neuroblastomas (median age: 15 months (interquartile range (IQR): 7-24 months); 38 females) who underwent magnetic resonance imaging at our institution before the initiation of any therapy or biopsy. Fourteen patients had MYCN-amplified (MYCNA) neuroblastoma. Histogram parameters of ADC maps from the entire tumour was obtained from the baseline images and the normalised images. The Mann-Whitney U test was used to compare the absolute and normalised histogram parameters amongst neuroblastomas with and without MYCN-amplification. Receiver operating characteristic (ROC) curves and area under the curves (AUC) were generated for the statistically significant histogram parameters. Cut-offs obtained from the ROC curves were evaluated on an external validation set (n-15, MYCNA-6, F-7, age 24 months (10-60)). A logistic regression model was trained to predict MYCNA by combining statistically significant histogram parameters and was evaluated on the validation set. RESULTS: MYCN-amplified neuroblastomas had statistically significant higher maximum ADC and lower minimum ADC than non-amplified neuroblastomas. They also demonstrated higher entropy, variance, energy, and lower uniformity than non-amplified neoplasms (p > 0.05). Energy, entropy, and maximum ADC had AUC of 0.85, 0.79, and 0.82, respectively. CONCLUSIONS: Whole tumour ADC histogram analysis of neuroblastomas can differentiate between tumours with and without MYCN-amplification. These parameters can help identify areas for targeted biopsies or can be used to predict subtypes of these high-risk tumours before biopsy results are available. KEY POINTS: • MYCN-amplification significantly affects treatment decisions in neuroblastomas. • MYCN-amplified neuroblastomas had significantly different ADC histogram metrics as compared to tumours without amplification. • ADC histogram metrics can be used to predict MYCN-amplification status based on imaging.

Respiratory triggered diffusion-weighted imaging with a single diffusion sensitising gradient to reduce image acquisition time - A feasibility study in the workup of hepatocellular carcinoma.

PURPOSE: We evaluated respiratory triggered unidirectional single-shot echo-planar imaging (u-SSEPI) as a time-saving measure in diffusion imaging of the upper abdomen. Specifically, we compared the ADC values obtained from unidirectional DWI (u-SSEPI) and routine DWI (4t-SSEPI) and also the diagnostic accuracies of unidirectional and routine DWI sequences in the identification of focal liver lesions in the setting of chronic liver disease (CLD). MATERIALS AND METHODS: This prospective, IRB approved study, included 48 patients of CLD who underwent-DCE-MRI on a 1.5 T scanner for hepatocellular carcinoma (HCC) workup. In addition to 4t-SSEPI, u-SSEPI was acquired with the diffusion sensitising gradient being applied in only one direction; keeping all other parameters same as 4t-SSEPI. Two blinded radiologists evaluated the DWI studies for image quality and detection of liver lesions. A composite gold standard was established using DCE-MRI, follow-up imaging and patient clinical details. The apparent diffusion coefficients (ADCs) of the liver, spleen and the lesions were compared between the two sequences. ROC analysis evaluated the diagnostic accuracy of ADC from both the sequences in identifying HCC. RESULTS: Eighty-eight lesions were identified using the composite gold standard. u-SSEPI resulted in 3 times faster image acquisition. No statistically significant differences were demonstrated between the unidirectional and routine DWI sequences for image quality parameters and lesion detection rates. Lesion wise comparison of the ADC values from both the sequences was not statistically different (p = 0.8) with a coefficient of variation = 12-14 %. The Bland- Altman plots and the Passing-Bablock regression analysis demonstrated a systematic and proportional bias between the ADC values obtained. The AUC of the ROC curve, however, was 0.63-observer1; 0.62-oobserver2 for routine DWI and 0.65; 0.62 for unidirectional DWI when ADC was used to identify HCC (the AUCs were not statistically different (p =  0.6-0.8)). CONCLUSION: No significant differences were demonstrated in the diagnostic accuracies of unidirectional and routine DWI in the diagnosis of HCC. Unidirectional diffusion may be further evaluated in other organs where diffusion is isotropic, especially in respiratory triggered sequences where the imaging time dividend is significant.

Two-dimensional (2D) morphologic measurements can quantify the severity of liver disease in children with autosomal recessive polycystic kidney disease (ARPKD).

PURPOSE: To evaluate the correlation of 2D shape-based features with magnetic resonance elastography (MRE)-derived liver stiffness and portal hypertension (pHTN) in children with ARPKD-associated congenital hepatic fibrosis. METHODS: In a prospective IRB-approved study, 14 children with ARPKD (mean age ± SD = 13.8 ± 5.8 years) and 14 healthy controls (mean age ± SD = 13.7 ± 3.9 years) underwent liver MRE. A 2D region of interest (ROI) outlining the left liver lobe at the level of the abdominal aorta was drawn on sagittal T2-weighted images. Eight shape features (perimeter, major axis length, maximum diameter, perimeter to surface ratio (PSR), elongation, sphericity, minor axis length, and mesh surface) describing the 2D-ROI were calculated. Spearman's correlation was calculated between shape features and MRE-derived liver stiffness (kPa) (n = 28). Shape features were compared between participants with ARPKD with pHTN (splenomegaly and thrombocytopenia), (n = 4) and without pHTN (n = 8) using the Mann Whitney U test. Receiver operating characteristic (ROC) curves were generated to examine the diagnostic accuracy of shape features in identifying cases with liver stiffness > 2.9 kPa. RESULTS: In ARPKD participants and healthy controls, all eight shape features, except elongation, showed moderate to strong correlation with liver stiffness (kPa); the perimeter surface ratio had the strongest correlation (rho = - 0.75, p < 0.001). In ROC analysis, a cut-off of PSR ≤ 0.057 mm-1 gave 100% (95% CI: 59.0-100.0) sensitivity and 100% (95% CI: 83.9-100.0) specificity in identifying ARPKD participants with liver stiffness > 2.9 kPa, with an area under the ROC curve (AUC) of 1.0 (95% CI: 0.88-1.00). Individuals with pHTN had a lower median PSR (mean ± SD = 0.05 ± 0.01) than those without (0.07 ± 0.01; p = 0.027) with an AUC of 0.91 (95% CI: 0.60-0.99) in differentiating the participants with and without pHTN. CONCLUSION: Shape-based features of the left liver lobe show potential as non-invasive biomarkers of liver fibrosis and portal hypertension in children with ARPKD.

Texture analysis of routine T2 weighted fat-saturated images can identify head and neck paragangliomas - A pilot study.

PURPOSE: To evaluate the role of the first and second-order texture parameters obtained from T2-weighted fat-saturated DIXON images in differentiating paragangliomas from other neck masses, and to develop a statistical model to classify them. METHOD: We retrospectively evaluated 38 paragangliomas, 18 nerve-sheath tumours and 14 other miscellaneous neck lesions obtained from an IRB approved study conducted between January 2016 and June 2019; using a composite gold standard of histopathology, cytology and DOTANOC PET CT (A total of 70 lesions in 63 patients). Fat-suppressed T2weighted-DIXON axial images were used. First and second-order texture-parameters were calculated from the original and filtered images. Feature selection using F-statistics and collinearity analysis provided 14 texture parameters for further analysis. Mann-Whitney-U test was used to compare between the groups and p-values were adjusted for multiple comparisons. ROC curve analysis was used to obtain optimal cut-offs. RESULTS: A total of ten texture features were found to be significantly different between paragangliomas and non-paraganglioma lesions. Minimum from the histogram of grey levels was lower in paragangliomas with a cut off of ≤113.462 obtaining 62.9 % sensitivity and 77.27 % specificity in differentiating paragangliomas from non-paragangliomas. Logistic regression model was trained (n-49) using forward feature selection, which when evaluated on the validation set(n-21)- obtained an AUC of 0.855(95 %CI, 0.633 to 0.968) with a positive likelihood ratio of 4.545 (95 %CI, 1.298-15.923) in differentiating paragangliomas from non-paragangliomas. CONCLUSION: Texture analysis of a routine imaging sequence can identify paragangliomas with high accuracy. Further development of texture analysis would enable better imaging workflow, resource utilisation and imaging cost reductions.

Major and Minor Discordance in Dual-Energy X-Ray Absorptiometry Diagnosis of Osteoporosis - A Cross-Sectional, Population-Based, Observational Study in Indian Women.

OBJECTIVE: The study objective was to evaluate the presence of major and minor discordance in the diagnosis of osteoporosis in a population-based screening program of Indian women using hip and spine dual-energy X-ray absorptiometry (DEXA). METHODS: In this institutional review board-approved study, a population-based screening program was offered to women aged > 40 using a mobile van model. A total of 5708 women underwent DEXA between May 2012 and May 2016 as a population-based, opt-in screening program offered to women as an outreach program. Bone mineral density (BMD) was measured at the hip and spine, which was used to derive T-scores and to determine the prevalence of discordance. RESULTS: The densitometry scores were concordant in 42.50% of the cases, with abnormal bone mineral density, whereas in 54.15% of cases, there was minor discordance and major discordance in 3.35% of cases. Body mass index, weight, age, and postmenopausal status of the patient were important predictors of the presence of discordance. CONCLUSIONS: Clinicians and epidemiologists should be prepared for at least five out of every ten women screened to have discordance of the T scores at the two anatomical sites scanned. If there is discordance of BMD in underweight persons or in those with low body mass index, then causes other than physiological discordance should be considered, which may be further evaluated.

Interpretable Artificial Intelligence: Why and When.

OBJECTIVE. The purpose of this article is to discuss the problem of interpretability of artificial intelligence (AI) and highlight the need for continuing scientific discovery using AI algorithms to deal with medical big data. CONCLUSION. A plethora of AI algorithms are currently being used in medical research, but the opacity of these algorithms makes their clinical implementation a dilemma. Clinical decision making cannot be assigned to something that we do not understand. Therefore, AI research should not be limited to reporting accuracy and sensitivity but, rather, should try to explain the underlying reasons for the predictions, in an attempt to enrich biologic understanding and knowledge.

Solid variant ABC of long tubular bones: A diagnostic conundrum for the radiologist.

Solid variant of aneurysmal bone cyst (sABC) is an extremely rare, reactive and non-neoplastic osseous lesion. On imaging it presents as a diaphyseal aggressive, eccentrically placed lytic and expansile lesion. However, differentiating this entity from the other possible malignant differentials is confounded by the histopathology mimicking several commoner lesions. We describe the distinctive MRI features of sABC of long bones from a series of four cases and briefly review the literature. We hope this review will educate all radiologists about this rare entity increasing their diagnostic confidence while formulating differentials for similar appearing lesions.