Deep learning-based scan range optimization can reduce radiation exposure in coronary CT angiography.

OBJECTIVES: Cardiac computed tomography (CT) is essential in diagnosing coronary heart disease. However, a disadvantage is the associated radiation exposure to the patient which depends in part on the scan range. This study aimed to develop a deep neural network to optimize the delimitation of scan ranges in CT localizers to reduce the radiation dose. METHODS: On a retrospective training cohort of 1507 CT localizers randomly selected from calcium scoring and angiography scans and acquired between 2010 and 2017, optimized scan ranges were delimited by two radiologists in consensus. A neural network was trained to reproduce the scan ranges and was tested on two randomly selected and independent validation cohorts: an internal cohort of 233 CT localizers (January 2018-June 2020) and an external cohort from a nearby hospital of 298 CT localizers (July 2020-December 2020). Localizers where a bypass surgery was visible were excluded. The effective radiation dose to the patient was simulated using a Monte Carlo simulation. Scan ranges of radiographers, radiologists, and the network were compared using an equivalence test; likewise, the reduction in effective dose was tested using a superior test. RESULTS: The network replicated the radiologists' scan ranges with a Dice score of 96.5 ± 0.02 (p < 0.001, indicating equivalence). The generated scan ranges resulted in an effective dose reduction of 10.0% (p = 0.002) in the internal cohort and 12.6% (p < 0.001) in the external cohort compared to the scan ranges delimited by radiographers in clinical routine. CONCLUSIONS: Automatic delimitation of the scan range can result in a radiation dose reduction to the patient. CLINICAL RELEVANCE STATEMENT: Fully automated delimitation of the scan range using a deep neural network enables a significant reduction in radiation exposure during CT coronary angiography compared to manual examination planning. It can also reduce the workload of the radiographers. KEY POINTS: • Scan range delimitation for coronary computed tomography angiography could be performed with high accuracy by a deep neural network. • Automated scan ranges showed a high agreement of 96.5% with the scan ranges of radiologists. • Using a Monte Carlo simulation, automated scan ranges reduced the effective dose to the patient by up to 12.6% (0.9 mSv) compared to the scan ranges of radiographers in clinical routine.

Pediatric age estimation from radiographs of the knee using deep learning.

OBJECTIVES: Age estimation, especially in pediatric patients, is regularly used in different contexts ranging from forensic over medicolegal to clinical applications. A deep neural network has been developed to automatically estimate chronological age from knee radiographs in pediatric patients. METHODS: In this retrospective study, 3816 radiographs of the knee from pediatric patients from a German population (acquired between January 2008 and December 2018) were collected to train a neural network. The network was trained to predict chronological age from the knee radiographs and was evaluated on an independent validation cohort of 423 radiographs (acquired between January 2019 and December 2020) and on an external validation cohort of 197 radiographs. RESULTS: The model showed a mean absolute error of 0.86 ± 0.72 years and 0.9 ± 0.71 years on the internal and external validation cohorts, respectively. Separating age classes (< 14 years from ≥ 14 years and < 18 years from ≥ 18 years) showed AUCs between 0.94 and 0.98. CONCLUSIONS: The chronological age of pediatric patients can be estimated with good accuracy from radiographs of the knee using a deep neural network. KEY POINTS: • Radiographs of the knee can be used for age estimations in pediatric patients using a standard deep neural network. • The network showed a mean absolute error of 0.86 ± 0.72 years in an internal validation cohort and of 0.9 ± 0.71 years in an external validation cohort. • The network can be used to separate the age classes < 14 years from ≥ 14 years with an AUC of 0.97 and < 18 years from ≥ 18 years with an AUC of 0.94.

18F-FDG PET-MR enterography in predicting histological active disease using the Nancy index in ulcerative colitis: a randomized controlled trial.

PURPOSE: To evaluate the diagnostic performance of PET-MR enterography in detecting histological active inflammation in patients with ulcerative colitis and the impact of bowel purgation on diagnostic accuracies of PET-MR parameters. METHODS: Fifty patients were enrolled in this randomized controlled trial (clinicaltrials.gov [NCT03781284]). Forty patients were randomized in two study arms, in which bowel purgation was performed either before or after PET-MR enterography. All patients underwent ileocolonoscopy with mucosal biopsies after PET-MR within 24 h. Diagnostic performance of MR morphological parameters (MRmorph), diffusion-weighted imaging (DWI), and PET in detecting histological inflammation determined by the Nancy index was compared with each other and between study arms. Correlation between PET and histological inflammatory severity was calculated. RESULTS: In study arm without previous bowel purgation, SUVmax ratio of bowel segment (relative to SUVmax of the liver) facilitated the highest specificity and diagnostic accuracy compared with MRmorph and DWI. Bowel cleansing led to markedly increased metabolic activity of bowel segments, resulting in significantly reduced specificity of PET compared with study arm without purgation (0.808 vs. 0.966, p = 0.007, respectively). Inter-observer concordance for assessing MRmorph was clearly increased after bowel cleansing (Cohen's κ, 0.847 vs. 0.665; p = 0.013, respectively), though diagnostic performance of MRmorph was not significantly improved. Our findings suggested that the change of metabolic status was mainly associated with the grade of neutrophil infiltrate and less dependent on chronic infiltrate. CONCLUSION: PET-MR enterography was an excellent non-invasive diagnostic method in the assessment of histological active inflammation in ulcerative colitis without the need of previous bowel purgation. TRIAL REGISTRATION: https://clinicaltrials.gov/ct2/show/NCT03781284.

Non-invasive tumor decoding and phenotyping of cerebral gliomas utilizing multiparametric 18F-FET PET-MRI and MR Fingerprinting.

OBJECTIVES: The introduction of the 2016 WHO classification of CNS tumors has made the combined molecular and histopathological characterization of tumors a pivotal part of glioma patient management. Recent publications on radiogenomics-based prediction of the mutational status have demonstrated the predictive potential of imaging-based, non-invasive tissue characterization algorithms. Hence, the aim of this study was to assess the potential of multiparametric 18F-FET PET-MRI including MR fingerprinting accelerated with machine learning and radiomic algorithms to predict tumor grading and mutational status of patients with cerebral gliomas. MATERIALS AND METHODS: 42 patients with suspected primary brain tumor without prior surgical or systemic treatment or biopsy underwent an 18F-FET PET-MRI examination. To differentiate the mutational status and the WHO grade of the cerebral tumors, support vector machine and random forest were trained with the radiomics signature of the multiparametric PET-MRI data including MR fingerprinting. Surgical sampling served as a gold standard for histopathological reference and assessment of mutational status. RESULTS: The 5-fold cross-validated area under the curve in predicting the ATRX mutation was 85.1%, MGMT mutation was 75.7%, IDH1 was 88.7%, and 1p19q was 97.8%. The area under the curve of differentiating low-grade glioma vs. high-grade glioma was 85.2%. CONCLUSION: 18F-FET PET-MRI and MR fingerprinting enable high-quality imaging-based tumor decoding and phenotyping for differentiation of low-grade vs. high-grade gliomas and for prediction of the mutational status of ATRX, IDH1, and 1p19q. These initial results underline the potential of 18F-FET PET-MRI to serve as an alternative to invasive tissue characterization.

An in vitro comparison of flow dynamics of the Magna Ease and the Trifecta prostheses.

Objectives: We aimed to compare the in vitro flow dynamics of the Perimount Magna Ease™ (PME) and the Trifecta™ (TF) bioprostheses.Material and methods: A new flow chamber was designed to compare the flow patterns of the PME (Edwards Lifesciences, Irvine, CA, USA) and the TF (SJM, St. Paul, MN, USA) aortic valve prostheses. This new channel offered the possibility of 2D-particle-image-velocimetry (2D-PIV) to completely evaluate the flow field downstream from the aortic valve to the middle of the aortic arch. Maximum average velocities, vorticity, shear strength, maximum orifice diameters and jet flow diameters were analyzed. Valve sizes of 21, 23 and 25 mm were evaluated.Results: Average velocity values, shear strength and vorticities were smaller in the flow field of the TF (maximum average velocity: 0.81 ± 0.03m/s, PME 23 mm vs. 0.7 ± 0.02m/s TF 23 mm, P < .001) under pulsatile flow conditions (70 Hz, 70 mL stroke volume). The evaluation of the upper orifice area revealed bigger maximum diameters during the peak flow phase for the TF, but more leaflet-flutter.Conclusions: Our flow chamber allowed a precise and highly sensitive characterization and comparison of complex fluid dynamics of different aortic valve prostheses. Both the Trifecta™ and the Perimount Magna Ease™ showed a good performance on a high level.

The effect of feature normalization methods in radiomics.

OBJECTIVES: In radiomics, different feature normalization methods, such as z-Score or Min-Max, are currently utilized, but their specific impact on the model is unclear. We aimed to measure their effect on the predictive performance and the feature selection. METHODS: We employed fifteen publicly available radiomics datasets to compare seven normalization methods. Using four feature selection and classifier methods, we used cross-validation to measure the area under the curve (AUC) of the resulting models, the agreement of selected features, and the model calibration. In addition, we assessed whether normalization before cross-validation introduces bias. RESULTS: On average, the difference between the normalization methods was relatively small, with a gain of at most + 0.012 in AUC when comparing the z-Score (mean AUC: 0.707 ± 0.102) to no normalization (mean AUC: 0.719 ± 0.107). However, on some datasets, the difference reached + 0.051. The z-Score performed best, while the tanh transformation showed the worst performance and even decreased the overall predictive performance. While quantile transformation performed, on average, slightly worse than the z-Score, it outperformed all other methods on one out of three datasets. The agreement between the features selected by different normalization methods was only mild, reaching at most 62%. Applying the normalization before cross-validation did not introduce significant bias. CONCLUSION: The choice of the feature normalization method influenced the predictive performance but depended strongly on the dataset. It strongly impacted the set of selected features. CRITICAL RELEVANCE STATEMENT: Feature normalization plays a crucial role in the preprocessing and influences the predictive performance and the selected features, complicating feature interpretation. KEY POINTS: • The impact of feature normalization methods on radiomic models was measured. • Normalization methods performed similarly on average, but differed more strongly on some datasets. • Different methods led to different sets of selected features, impeding feature interpretation. • Model calibration was not largely affected by the normalization method.

Applying oversampling before cross-validation will lead to high bias in radiomics.

Class imbalance is often unavoidable for radiomic data collected from clinical routine. It can create problems during classifier training since the majority class could dominate the minority class. Consequently, resampling methods like oversampling or undersampling are applied to the data to class-balance the data. However, the resampling must not be applied upfront to all data because it would lead to data leakage and, therefore, to erroneous results. This study aims to measure the extent of this bias. Five-fold cross-validation with 30 repeats was performed using a set of 15 radiomic datasets to train predictive models. The training involved two scenarios: first, the models were trained correctly by applying the resampling methods during the cross-validation. Second, the models were trained incorrectly by performing the resampling on all the data before cross-validation. The bias was defined empirically as the difference between the best-performing models in both scenarios in terms of area under the receiver operating characteristic curve (AUC), sensitivity, specificity, balanced accuracy, and the Brier score. In addition, a simulation study was performed on a randomly generated dataset for verification. The results demonstrated that incorrectly applying the oversampling methods to all data resulted in a large positive bias (up to 0.34 in AUC, 0.33 in sensitivity, 0.31 in specificity, and 0.37 in balanced accuracy). The bias depended on the data balance, and approximately an increase of 0.10 in the AUC was observed for each increase in imbalance. The models also showed a bias in calibration measured using the Brier score, which differed by up to -0.18 between the correctly and incorrectly trained models. The undersampling methods were not affected significantly by bias. These results emphasize that any resampling method should be applied correctly only to the training data to avoid data leakage and, subsequently, biased model performance and calibration.

The effect of data resampling methods in radiomics.

Radiomic datasets can be class-imbalanced, for instance, when the prevalence of diseases varies notably, meaning that the number of positive samples is much smaller than that of negative samples. In these cases, the majority class may dominate the model's training and thus negatively affect the model's predictive performance, leading to bias. Therefore, resampling methods are often utilized to class-balance the data. However, several resampling methods exist, and neither their relative predictive performance nor their impact on feature selection has been systematically analyzed. In this study, we aimed to measure the impact of nine resampling methods on radiomic models utilizing a set of fifteen publicly available datasets regarding their predictive performance. Furthermore, we evaluated the agreement and similarity of the set of selected features. Our results show that applying resampling methods did not improve the predictive performance on average. On specific datasets, slight improvements in predictive performance (+ 0.015 in AUC) could be seen. A considerable disagreement on the set of selected features was seen (only 28.7% of features agreed), which strongly impedes feature interpretability. However, selected features are similar when considering their correlation (82.9% of features correlated on average).

Radiomics Analysis of Multiparametric PET/MRI for N- and M-Staging in Patients with Primary Cervical Cancer.

PURPOSE: The aim of this study was to investigate the potential of multiparametric 18F-FDG PET/MR imaging as a platform for radiomics analysis and machine learning algorithms based on primary cervical cancers to predict N- and M-stage in patients. MATERIALS AND METHODS: A total of 30 patients with histopathological confirmation of primary and untreated cervical cancer were prospectively enrolled for a multiparametric 18F-FDG PET/MR examination, comprising a dedicated protocol for imaging of the female pelvis. The primary tumor in the uterine cervix was manually segmented on post-contrast T1-weighted images. Quantitative features were extracted from the segmented tumors using the Radiomic Image Processing Toolbox for the R software environment for statistical computing and graphics. 45 different image features were calculated from non-enhanced as well as post-contrast T1-weighted TSE images, T2-weighted TSE images, the ADC map, the parametric Ktrans, Kep, Ve and iAUC maps and PET images, respectively. Statistical analysis and modeling was performed using Python 3.5 and the scikit-learn software machine learning library for the Python programming language. RESULTS: Prediction of M-stage was superior when compared to N-stage. Prediction of M-stage using SVM with SVM-RFE as feature selection obtained the highest performance providing sensitivity of 91 % and specificity of 92 %. Using receiver operating characteristic (ROC) analysis of the pooled predictions, the area under the curve (AUC) was 0.97. Prediction of N-stage using RBF-SVM with MIFS as feature selection reached sensitivity of 83 %, specificity of 67 % and an AUC of 0.82. CONCLUSION: M- and N-stage can be predicted based on isolated radiomics analyses of the primary tumor in cervical cancers, thus serving as a template for noninvasive tumor phenotyping and patient stratification using high-dimensional feature vectors extracted from multiparametric PET/MRI data. KEY POINTS: · Radiomics analysis based on multiparametric PET/MRI enables prediction of the metastatic status of cervical cancers. · Prediction of M-stage is superior to N-stage. · Multiparametric PET/MRI displays a valuable platform for radiomics analyses .

Assessment of image quality and impact of deep learning-based software in non-contrast head CT scans.

In this retrospective study, we aimed to assess the objective and subjective image quality of different reconstruction techniques and a deep learning-based software on non-contrast head computed tomography (CT) images. In total, 152 adult head CT scans (77 female, 75 male; mean age 69.4 ± 18.3 years) obtained from three different CT scanners using different protocols between March and April 2021 were included. CT images were reconstructed using filtered-back projection (FBP), iterative reconstruction (IR), and post-processed using a deep learning-based algorithm (PS). Post-processing significantly reduced noise in FBP-reconstructed images (up to 15.4% reduction) depending on the protocol, leading to improvements in signal-to-noise ratio of up to 19.7%. However, when deep learning-based post-processing was applied to FBP images compared to IR alone, the differences were inconsistent and partly non-significant, which appeared to be protocol or site specific. Subjective assessments showed no significant overall improvement in image quality for all reconstructions and post-processing. Inter-rater reliability was low and preferences varied. Deep learning-based denoising software improved objective image quality compared to FBP in routine head CT. A significant difference compared to IR was observed for only one protocol. Subjective assessments did not indicate a significant clinical impact in terms of improved subjective image quality, likely due to the low noise levels in full-dose images.

Impact of AI-Based Post-Processing on Image Quality of Non-Contrast Computed Tomography of the Chest and Abdomen.

Non-contrast computed tomography (CT) is commonly used for the evaluation of various pathologies including pulmonary infections or urolithiasis but, especially in low-dose protocols, image quality is reduced. To improve this, deep learning-based post-processing approaches are being developed. Therefore, we aimed to compare the objective and subjective image quality of different reconstruction techniques and a deep learning-based software on non-contrast chest and low-dose abdominal CTs. In this retrospective study, non-contrast chest CTs of patients suspected of COVID-19 pneumonia and low-dose abdominal CTs suspected of urolithiasis were analysed. All images were reconstructed using filtered back-projection (FBP) and were post-processed using an artificial intelligence (AI)-based commercial software (PixelShine (PS)). Additional iterative reconstruction (IR) was performed for abdominal CTs. Objective and subjective image quality were evaluated. AI-based post-processing led to an overall significant noise reduction independent of the protocol (chest or abdomen) while maintaining image information (max. difference in SNR 2.59 ± 2.9 and CNR 15.92 ± 8.9, p < 0.001). Post-processing of FBP-reconstructed abdominal images was even superior to IR alone (max. difference in SNR 0.76 ± 0.5, p ≤ 0.001). Subjective assessments verified these results, partly suggesting benefits, especially in soft-tissue imaging (p < 0.001). All in all, the deep learning-based denoising-which was non-inferior to IR-offers an opportunity for image quality improvement especially in institutions using older scanners without IR availability. Further studies are necessary to evaluate potential effects on dose reduction benefits.

Impact of Vaccination Status on Outcome of Patients With COVID-19 and Acute Ischemic Stroke Undergoing Mechanical Thrombectomy.

BACKGROUND: Data on impact of COVID-19 vaccination and outcomes of patients with COVID-19 and acute ischemic stroke undergoing mechanical thrombectomy are scarce. Addressing this subject, we report our multicenter experience. METHODS AND RESULTS: This was a retrospective analysis of patients with COVID-19 and known vaccination status treated with mechanical thrombectomy for acute ischemic stroke at 20 tertiary care centers between January 2020 and January 2023. Baseline demographics, angiographic outcome, and clinical outcome evaluated by the modified Rankin Scale score at discharge were noted. A multivariate analysis was conducted to test whether these variables were associated with an unfavorable outcome, defined as modified Rankin Scale score >3. A total of 137 patients with acute ischemic stroke (48 vaccinated and 89 unvaccinated) with acute or subsided COVID-19 infection who underwent mechanical thrombectomy attributable to vessel occlusion were included in the study. Angiographic outcomes between vaccinated and unvaccinated patients were similar (modified Thrombolysis in Cerebral Infarction ≥2b: 85.4% in vaccinated patients versus 86.5% in unvaccinated patients; P=0.859). The rate of functional independence (modified Rankin Scale score, ≤2) was 23.3% in the vaccinated group and 20.9% in the unvaccinated group (P=0.763). The mortality rate was 30% in both groups. In the multivariable analysis, vaccination status was not a significant predictor for an unfavorable outcome (P=0.957). However, acute COVID-19 infection remained significant (odds ratio, 1.197 [95% CI, 1.007-1.417]; P=0.041). CONCLUSIONS: Our study demonstrated no impact of COVID-19 vaccination on angiographic or clinical outcome of COVID-19-positive patients with acute ischemic stroke undergoing mechanical thrombectomy, whereas worsening attributable to COVID-19 was confirmed.

METhodological RadiomICs Score (METRICS): a quality scoring tool for radiomics research endorsed by EuSoMII.

PURPOSE: To propose a new quality scoring tool, METhodological RadiomICs Score (METRICS), to assess and improve research quality of radiomics studies. METHODS: We conducted an online modified Delphi study with a group of international experts. It was performed in three consecutive stages: Stage#1, item preparation; Stage#2, panel discussion among EuSoMII Auditing Group members to identify the items to be voted; and Stage#3, four rounds of the modified Delphi exercise by panelists to determine the items eligible for the METRICS and their weights. The consensus threshold was 75%. Based on the median ranks derived from expert panel opinion and their rank-sum based conversion to importance scores, the category and item weights were calculated. RESULT: In total, 59 panelists from 19 countries participated in selection and ranking of the items and categories. Final METRICS tool included 30 items within 9 categories. According to their weights, the categories were in descending order of importance: study design, imaging data, image processing and feature extraction, metrics and comparison, testing, feature processing, preparation for modeling, segmentation, and open science. A web application and a repository were developed to streamline the calculation of the METRICS score and to collect feedback from the radiomics community. CONCLUSION: In this work, we developed a scoring tool for assessing the methodological quality of the radiomics research, with a large international panel and a modified Delphi protocol. With its conditional format to cover methodological variations, it provides a well-constructed framework for the key methodological concepts to assess the quality of radiomic research papers. CRITICAL RELEVANCE STATEMENT: A quality assessment tool, METhodological RadiomICs Score (METRICS), is made available by a large group of international domain experts, with transparent methodology, aiming at evaluating and improving research quality in radiomics and machine learning. KEY POINTS: • A methodological scoring tool, METRICS, was developed for assessing the quality of radiomics research, with a large international expert panel and a modified Delphi protocol. • The proposed scoring tool presents expert opinion-based importance weights of categories and items with a transparent methodology for the first time. • METRICS accounts for varying use cases, from handcrafted radiomics to entirely deep learning-based pipelines. • A web application has been developed to help with the calculation of the METRICS score ( https://metricsscore.github.io/metrics/METRICS.html ) and a repository created to collect feedback from the radiomics community ( https://github.com/metricsscore/metrics ).

Are deep models in radiomics performing better than generic models? A systematic review.

BACKGROUND: Application of radiomics proceeds by extracting and analysing imaging features based on generic morphological, textural, and statistical features defined by formulas. Recently, deep learning methods were applied. It is unclear whether deep models (DMs) can outperform generic models (GMs). METHODS: We identified publications on PubMed and Embase to determine differences between DMs and GMs in terms of receiver operating area under the curve (AUC). RESULTS: Of 1,229 records (between 2017 and 2021), 69 studies were included, 61 (88%) on tumours, 68 (99%) retrospective, and 39 (56%) single centre; 30 (43%) used an internal validation cohort; and 18 (26%) applied cross-validation. Studies with independent internal cohort had a median training sample of 196 (range 41-1,455); those with cross-validation had only 133 (43-1,426). Median size of validation cohorts was 73 (18-535) for internal and 94 (18-388) for external. Considering the internal validation, in 74% (49/66), the DMs performed better than the GMs, vice versa in 20% (13/66); no difference in 6% (4/66); and median difference in AUC 0.045. On the external validation, DMs were better in 65% (13/20), GMs in 20% (4/20) cases; no difference in 3 (15%); and median difference in AUC 0.025. On internal validation, fused models outperformed GMs and DMs in 72% (20/28), while they were worse in 14% (4/28) and equal in 14% (4/28); median gain in AUC was + 0.02. On external validation, fused model performed better in 63% (5/8), worse in 25% (2/8), and equal in 13% (1/8); median gain in AUC was + 0.025. CONCLUSIONS: Overall, DMs outperformed GMs but in 26% of the studies, DMs did not outperform GMs.

Deep Features from Pretrained Networks Do Not Outperform Hand-Crafted Features in Radiomics.

In radiomics, utilizing features extracted from pretrained deep networks could result in models with a higher predictive performance than those relying on hand-crafted features. This study compared the predictive performance of models trained with either deep features, hand-crafted features, or a combination of these features in terms of the area under the receiver-operating characteristic curve (AUC) and other metrics. We trained models on ten radiological datasets using five feature selection methods and three classifiers. Our results indicate that models based on deep features did not show an improved AUC compared to those utilizing hand-crafted features (deep: AUC 0.775, hand-crafted: AUC 0.789; p = 0.28). Including morphological features alongside deep features led to overall improvements in prediction performance for all models (+0.02 gain in AUC; p < 0.001); however, the best model did not benefit from this (+0.003 gain in AUC; p = 0.57). Using all hand-crafted features in addition to the deep features resulted in a further overall improvement (+0.034 in AUC; p < 0.001), but only a minor improvement could be observed for the best model (deep: AUC 0.798, hand-crafted: AUC 0.789; p = 0.92). Furthermore, our results show that models based on deep features extracted from networks pretrained on medical data have no advantage in predictive performance over models relying on features extracted from networks pretrained on ImageNet data. Our study contributes a benchmarking analysis of models trained on hand-crafted and deep features from pretrained networks across multiple datasets. It also provides a comprehensive understanding of their applicability and limitations in radiomics. Our study shows, in conclusion, that models based on features extracted from pretrained deep networks do not outperform models trained on hand-crafted ones.

Pediatric age estimation from thoracic and abdominal CT scout views using deep learning.

Age assessment is regularly used in clinical routine by pediatric endocrinologists to determine the physical development or maturity of children and adolescents. Our study investigates whether age assessment can be performed using CT scout views from thoracic and abdominal CT scans using a deep neural network. Hence, we retrospectively collected 1949 CT scout views from pediatric patients (acquired between January 2013 and December 2018) to train a deep neural network to predict the chronological age from CT scout views. The network was then evaluated on an independent test set of 502 CT scout views (acquired between January 2019 and July 2020). The trained model showed a mean absolute error of 1.18 ± 1.14 years on the test data set. A one-sided t-test to determine whether the difference between the predicted and actual chronological age was less than 2.0 years was statistically highly significant (p < 0.001). In addition, the correlation coefficient was very high (R = 0.97). In conclusion, the chronological age of pediatric patients can be assessed with high accuracy from CT scout views using a deep neural network.

Determining body height and weight from thoracic and abdominal CT localizers in pediatric and young adult patients using deep learning.

In this retrospective study, we aimed to predict the body height and weight of pediatric patients using CT localizers, which are overview scans performed before the acquisition of the CT. We trained three commonly used networks (EfficientNetV2-S, ResNet-18, and ResNet-34) on a cohort of 1009 and 1111 CT localizers of pediatric patients with recorded body height and weight (between January 2013 and December 2019) and validated them in an additional cohort of 116 and 127 localizers (acquired in 2020). The best-performing model was then tested in an independent cohort of 203 and 225 CT localizers (acquired between January 2021 and March 2023). In addition, a cohort of 1401 and 1590 localizers from younger adults (acquired between January 2013 and December 2013) was added to the training set to determine if it could improve the overall accuracy. The EfficientNetV2-S using the additional adult cohort performed best with a mean absolute error of 5.58 ± 4.26 cm for height and 4.25 ± 4.28 kg for weight. The relative error was 4.12 ± 4.05% for height and 11.28 ± 12.05% for weight. Our study demonstrated that automated estimation of height and weight in pediatric patients from CT localizers can be performed.

Real-Life Experience in the Treatment of Intrahepatic Cholangiocarcinoma by 90Y Radioembolization: A Multicenter Retrospective Study.

Limited treatment options in patients with intrahepatic cholangiocarcinoma (iCCA) demand the introduction of new, catheter-based treatment options. Especially, 90Y radioembolization may expand therapeutic abilities beyond surgery or chemotherapy. Therefore, the purpose of this study was to identify factors associated with an improved median overall survival (mOS) in iCCA patients receiving radioembolization in a retrospective study at 5 major tertiary-care centers. Methods: In total, 138 radioembolizations in 128 patients with iCCA (female, 47.7%; male, 52.3%; mean age ± SD, 61.1 ± 13.4 y) were analyzed. Clinical data, imaging characteristics, and radioembolization reports, as well as data from RECIST, version 1.1, analysis performed 3, 6, and 12 mo after radioembolization, were collected. mOS was compared among different subgroups using Kaplan-Meier curves and the log-rank test. Results: Radioembolization was performed as first-line treatment in 25.4%, as second-line treatment in 38.4%, and as salvage treatment in 36.2%. In patients receiving first-line, second-line, and salvage radioembolization, the disease control rate was 68.6%, 52.8%, and 54.0% after 3 mo; 31.4%, 15.1%, and 12.0% after 6 mo; and 17.1%, 5.7%, and 6.0% after 1 y, respectively. In patients receiving radioembolization as first-line, second-line, and salvage treatment, mOS was 12.0 mo (95% CI, 7.6-23.4 mo), 11.8 mo (95% CI, 9.1-16.6 mo), and 8.4 mo (95% CI, 6.3-12.7 mo), respectively. No significant differences among the 3 groups were observed (P = 0.15). Hepatic tumor burden did not significantly influence mOS (P = 0.12). Conclusion: Especially in advanced iCCA, second-line and salvage radioembolization may be important treatment options. In addition to ongoing studies investigating the role of radioembolization as first-line treatment, the role of radioembolization in the later treatment stages of the disease demands further attention.

Evaluation of the dependence of radiomic features on the machine learning model.

BACKGROUND: In radiomic studies, several models are often trained with different combinations of feature selection methods and classifiers. The features of the best model are usually considered relevant to the problem, and they represent potential biomarkers. Features selected from statistically similarly performing models are generally not studied. To understand the degree to which the selected features of these statistically similar models differ, 14 publicly available datasets, 8 feature selection methods, and 8 classifiers were used in this retrospective study. For each combination of feature selection and classifier, a model was trained, and its performance was measured with AUC-ROC. The best-performing model was compared to other models using a DeLong test. Models that were statistically similar were compared in terms of their selected features. RESULTS: Approximately 57% of all models analyzed were statistically similar to the best-performing model. Feature selection methods were, in general, relatively unstable (0.58; range 0.35-0.84). The features selected by different models varied largely (0.19; range 0.02-0.42), although the selected features themselves were highly correlated (0.71; range 0.4-0.92). CONCLUSIONS: Feature relevance in radiomics strongly depends on the model used, and statistically similar models will generally identify different features as relevant. Considering features selected by a single model is misleading, and it is often not possible to directly determine whether such features are candidate biomarkers.

Benchmarking Feature Selection Methods in Radiomics.

OBJECTIVES: A critical problem in radiomic studies is the high dimensionality of the datasets, which stems from small sample sizes and many generic features extracted from the volume of interest. Therefore, feature selection methods are used, which aim to remove redundant as well as irrelevant features. Because there are many feature selection algorithms, it is key to understand their performance in the context of radiomics. MATERIALS AND METHODS: A total of 29 feature selection algorithms and 10 classifiers were evaluated on 10 publicly available radiomic datasets. Feature selection methods were compared for training times, for the stability of the selected features, and for ranking, which measures the pairwise similarity of the methods. In addition, the predictive performance of the algorithms was measured by utilizing the area under the receiver operating characteristic curve of the best-performing classifier. RESULTS: Feature selections differed largely in training times as well as stability and similarity. No single method was able to outperform another one consistently in predictive performance. CONCLUSION: Our results indicated that simpler methods are more stable than complex ones and do not perform worse in terms of area under the receiver operating characteristic curve. Analysis of variance, least absolute shrinkage and selection operator, and minimum redundancy, maximum relevance ensemble appear to be good choices for radiomic studies in terms of predictive performance, as they outperformed most other feature selection methods.