Electron density dual-energy CT can improve the detection of lumbar disc herniation with higher image quality than standard and virtual non-calcium images.

OBJECTIVES: To compare the diagnostic performance and image quality of dual-energy computed tomography (DECT) with electron density (ED) image reconstruction with those of DECT with standard CT (SC) and virtual non-calcium (VNCa) image reconstructions, for diagnosing lumbar disc herniation (L-HIVD). METHODS: A total of 59 patients (354 intervertebral discs from T12/L1 to L5/S1; mean age, 60 years; 30 women and 29 men) who underwent DECT with spectral reconstruction and 3-T MRI within 2 weeks were enrolled between March 2021 and February 2022. Four radiologists independently assessed three image sets of randomized ED, SC, and VNCa images to detect L-HIVD at 8-week intervals. The coefficient of variance (CV) and the Weber contrast of the ROIs in the normal and diseased disc to cerebrospinal fluid space (NCR-normal/-diseased, respectively) were calculated to compare the image qualities of the noiseless ED and other series. RESULTS: Overall, 129 L-HIVDs were noted on MRI. In the detection of L-HIVD, ED showed a higher AUC and sensitivity than SC and VNCa; 0.871 vs 0.807 vs 833 (p = 0.002) and 81% vs 70% vs 74% (p = 0.006 for SC), respectively. CV was much lower in all measurements of ED than those for SC and VNCa (p < 0.001). Furthermore, NCR-normal and NCR-diseased were the highest in ED (ED vs SC in NCR-normal and NCR-diseased, p = 0.001 and p = 0.004, respectively; ED vs VNCa in NCR-diseased, p = 0.044). CONCLUSION: Compared to SC and VNCa images, DECT with ED reconstruction can enhance the AUC and sensitivity of L-HIVD detection with a lower CV and higher NCR. CLINICAL RELEVANCE STATEMENT: To our knowledge, this is the first study to quantify the image quality of noiseless ED images. ED imaging may be helpful for detecting L-HIVD in patients who cannot undergo MRI. KEY POINTS: ED images have diagnostic potential, but relevant quantitative analyses of image quality are limited. ED images detect disc herniation, with a better coefficient of variance and normalized contrast ratio values. ED images could detect L-HIVD when MRI is not an option.

Olecranon bone age assessment in puberty using a lateral elbow radiograph and a deep-learning model.

OBJECTIVES: To improve pubertal bone age (BA) evaluation by developing a precise and practical elbow BA classification using the olecranon, and a deep-learning AI model. MATERIALS AND METHODS: Lateral elbow radiographs taken for BA evaluation in children under 18 years were collected from January 2020 to June 2022, retrospectively. A novel classification and the olecranon BA were established based on the morphological changes in the olecranon ossification process during puberty. The olecranon BA was compared with other elbow and hand BA methods, using intraclass correlation coefficients (ICCs), and a deep-learning AI model was developed. RESULTS: A total of 3508 lateral elbow radiographs (mean age 9.8 ± 1.8 years) were collected. The olecranon BA showed the highest applicability (100%) and interobserver agreement (ICC 0.993) among elbow BA methods. It showed excellent reliability with Sauvegrain (0.967 in girls, 0.969 in boys) and Dimeglio (0.978 in girls, 0.978 in boys) elbow BA methods, as well as Korean standard (KS) hand BA in boys (0.917), and good reliability with KS in girls (0.896) and Greulich-Pyle (GP)/Tanner-Whitehouse (TW)3 (0.835 in girls, 0.895 in boys) hand BA methods. The AI model for olecranon BA showed an accuracy of 0.96 and a specificity of 0.98 with EfficientDet-b4. External validation showed an accuracy of 0.86 and a specificity of 0.91. CONCLUSION: The olecranon BA evaluation for puberty, requiring only a lateral elbow radiograph, showed the highest applicability and interobserver agreement, and excellent reliability with other BA evaluation methods, along with a high performance of the AI model. CLINICAL RELEVANCE STATEMENT: This AI model uses a single lateral elbow radiograph to determine bone age for puberty from the olecranon ossification center and can improve pubertal bone age assessment with the highest applicability and excellent reliability compared to previous methods. KEY POINTS: Elbow bone age is valuable for pubertal bone age assessment, but conventional methods have limitations. Olecranon bone age and its AI model showed high performances for pubertal bone age assessment. Olecranon bone age system is practical and accurate while requiring only a single lateral elbow radiograph.

Updated Primer on Generative Artificial Intelligence and Large Language Models in Medical Imaging for Medical Professionals.

The emergence of Chat Generative Pre-trained Transformer (ChatGPT), a chatbot developed by OpenAI, has garnered interest in the application of generative artificial intelligence (AI) models in the medical field. This review summarizes different generative AI models and their potential applications in the field of medicine and explores the evolving landscape of Generative Adversarial Networks and diffusion models since the introduction of generative AI models. These models have made valuable contributions to the field of radiology. Furthermore, this review also explores the significance of synthetic data in addressing privacy concerns and augmenting data diversity and quality within the medical domain, in addition to emphasizing the role of inversion in the investigation of generative models and outlining an approach to replicate this process. We provide an overview of Large Language Models, such as GPTs and bidirectional encoder representations (BERTs), that focus on prominent representatives and discuss recent initiatives involving language-vision models in radiology, including innovative large language and vision assistant for biomedicine (LLaVa-Med), to illustrate their practical application. This comprehensive review offers insights into the wide-ranging applications of generative AI models in clinical research and emphasizes their transformative potential.

Automated detection of intracranial aneurysms using skeleton-based 3D patches, semantic segmentation, and auxiliary classification for overcoming data imbalance in brain TOF-MRA.

Accurate and reliable detection of intracranial aneurysms is vital for subsequent treatment to prevent bleeding. However, the detection of intracranial aneurysms can be time-consuming and even challenging, and there is great variability among experts, especially in the case of small aneurysms. This study aimed to detect intracranial aneurysms accurately using a convolutional neural network (CNN) with 3D time-of-flight magnetic resonance angiography (TOF-MRA). A total of 154 3D TOF-MRA datasets with intracranial aneurysms were acquired, and the gold standards were manually drawn by neuroradiologists. We also obtained 113 subjects from a public dataset for external validation. These angiograms were pre-processed by using skull-stripping, signal intensity normalization, and N4 bias correction. The 3D patches along the vessel skeleton from MRA were extracted. Values of the ratio between the aneurysmal and the normal patches ranged from 1:1 to 1:5. The semantic segmentation on intracranial aneurysms was trained using a 3D U-Net with an auxiliary classifier to overcome the imbalance in patches. The proposed method achieved an accuracy of 0.910 in internal validation and external validation accuracy of 0.883 with a 2:1 ratio of normal to aneurysmal patches. This multi-task learning method showed that the aneurysm segmentation performance was sufficient to be helpful in an actual clinical setting.

Improvement of semantic segmentation through transfer learning of multi-class regions with convolutional neural networks on supine and prone breast MRI images.

Semantic segmentation of breast and surrounding tissues in supine and prone breast magnetic resonance imaging (MRI) is required for various kinds of computer-assisted diagnoses for surgical applications. Variability of breast shape in supine and prone poses along with various MRI artifacts makes it difficult to determine robust breast and surrounding tissue segmentation. Therefore, we evaluated semantic segmentation with transfer learning of convolutional neural networks to create robust breast segmentation in supine breast MRI without considering supine or prone positions. Total 29 patients with T1-weighted contrast-enhanced images were collected at Asan Medical Center and two types of breast MRI were performed in the prone position and the supine position. The four classes, including lungs and heart, muscles and bones, parenchyma with cancer, and skin and fat, were manually drawn by an expert. Semantic segmentation on breast MRI scans with supine, prone, transferred from prone to supine, and pooled supine and prone MRI were trained and compared using 2D U-Net, 3D U-Net, 2D nnU-Net and 3D nnU-Net. The best performance was 2D models with transfer learning. Our results showed excellent performance and could be used for clinical purposes such as breast registration and computer-aided diagnosis.

Fully automated condyle segmentation using 3D convolutional neural networks.

The aim of this study was to develop an auto-segmentation algorithm for mandibular condyle using the 3D U-Net and perform a stress test to determine the optimal dataset size for achieving clinically acceptable accuracy. 234 cone-beam computed tomography images of mandibular condyles were acquired from 117 subjects from two institutions, which were manually segmented to generate the ground truth. Semantic segmentation was performed using basic 3D U-Net and a cascaded 3D U-Net. A stress test was performed using different sets of condylar images as the training, validation, and test datasets. Relative accuracy was evaluated using dice similarity coefficients (DSCs) and Hausdorff distance (HD). In the five stages, the DSC ranged 0.886-0.922 and 0.912-0.932 for basic 3D U-Net and cascaded 3D U-Net, respectively; the HD ranged 2.557-3.099 and 2.452-2.600 for basic 3D U-Net and cascaded 3D U-Net, respectively. Stage V (largest data from two institutions) exhibited the highest DSC of 0.922 ± 0.021 and 0.932 ± 0.023 for basic 3D U-Net and cascaded 3D U-Net, respectively. Stage IV (200 samples from two institutions) had a lower performance than stage III (162 samples from one institution). Our results show that fully automated segmentation of mandibular condyles is possible using 3D U-Net algorithms, and the segmentation accuracy increases as training data increases.

Development and Validation of a Model Using Radiomics Features from an Apparent Diffusion Coefficient Map to Diagnose Local Tumor Recurrence in Patients Treated for Head and Neck Squamous Cell Carcinoma.

OBJECTIVE: To develop and validate a model using radiomics features from apparent diffusion coefficient (ADC) map to diagnose local tumor recurrence in head and neck squamous cell carcinoma (HNSCC). MATERIALS AND METHODS: This retrospective study included 285 patients (mean age ± standard deviation, 62 ± 12 years; 220 male, 77.2%), including 215 for training (n = 161) and internal validation (n = 54) and 70 others for external validation, with newly developed contrast-enhancing lesions at the primary cancer site on the surveillance MRI following definitive treatment of HNSCC between January 2014 and October 2019. Of the 215 and 70 patients, 127 and 34, respectively, had local tumor recurrence. Radiomics models using radiomics scores were created separately for T2-weighted imaging (T2WI), contrast-enhanced T1-weighted imaging (CE-T1WI), and ADC maps using non-zero coefficients from the least absolute shrinkage and selection operator in the training set. Receiver operating characteristic (ROC) analysis was used to evaluate the diagnostic performance of each radiomics score and known clinical parameter (age, sex, and clinical stage) in the internal and external validation sets. RESULTS: Five radiomics features from T2WI, six from CE-T1WI, and nine from ADC maps were selected and used to develop the respective radiomics models. The area under ROC curve (AUROC) of ADC radiomics score was 0.76 (95% confidence interval [CI], 0.62-0.89) and 0.77 (95% CI, 0.65-0.88) in the internal and external validation sets, respectively. These were significantly higher than the AUROC values of T2WI (0.53 [95% CI, 0.40-0.67], p = 0.006), CE-T1WI (0.53 [95% CI, 0.40-0.67], p = 0.012), and clinical parameters (0.53 [95% CI, 0.39-0.67], p = 0.021) in the external validation set. CONCLUSION: The radiomics model using ADC maps exhibited higher diagnostic performance than those of the radiomics models using T2WI or CE-T1WI and clinical parameters in the diagnosis of local tumor recurrence in HNSCC following definitive treatment.

Machine learning approach for differentiating cytomegalovirus esophagitis from herpes simplex virus esophagitis.

The endoscopic features between herpes simplex virus (HSV) and cytomegalovirus (CMV) esophagitis overlap significantly, and hence the differential diagnosis between HSV and CMV esophagitis is sometimes difficult. Therefore, we developed a machine-learning-based classifier to discriminate between CMV and HSV esophagitis. We analyzed 87 patients with HSV esophagitis and 63 patients with CMV esophagitis and developed a machine-learning-based artificial intelligence (AI) system using a total of 666 endoscopic images with HSV esophagitis and 416 endoscopic images with CMV esophagitis. In the five repeated five-fold cross-validations based on the hue-saturation-brightness color model, logistic regression with a least absolute shrinkage and selection operation showed the best performance (sensitivity, specificity, positive predictive value, negative predictive value, accuracy, and area under the receiver operating characteristic curve: 100%, 100%, 100%, 100%, 100%, and 1.0, respectively). Previous history of transplantation was included in classifiers as a clinical factor; the lower the performance of these classifiers, the greater the effect of including this clinical factor. Our machine-learning-based AI system for differential diagnosis between HSV and CMV esophagitis showed high accuracy, which could help clinicians with diagnoses.

Diffusion and perfusion MRI radiomics obtained from deep learning segmentation provides reproducible and comparable diagnostic model to human in post-treatment glioblastoma.

OBJECTIVES: Deep learning-based automatic segmentation (DLAS) helps the reproducibility of radiomics features, but its effect on radiomics modeling is unknown. We therefore evaluated whether DLAS can robustly extract anatomical and physiological MRI features, thereby assisting in the accurate assessment of treatment response in glioblastoma patients. METHODS: A DLAS model was trained on 238 glioblastomas and validated on an independent set of 98 pre- and 86 post-treatment glioblastomas from two tertiary hospitals. A total of 1618 radiomics features from contrast-enhanced T1-weighted images (CE-T1w) and histogram features from apparent diffusion coefficient (ADC) and cerebral blood volume (CBV) mapping were extracted. The diagnostic performance of radiomics features and ADC and CBV parameters for identifying treatment response was tested using area under the curve (AUC) from receiver operating characteristics analysis. Feature reproducibility was tested using a 0.80 cutoff for concordance correlation coefficients. RESULTS: Reproducibility was excellent for ADC and CBV features (ICC, 0.82-0.99) and first-order features (pre- and post-treatment, 100% and 94.1% remained), but lower for texture (79.0% and 69.1% remained) and wavelet-transformed (81.8% and 74.9% remained) features of CE-T1w. DLAS-based radiomics showed similar performance to human-performed segmentations in internal validation (AUC, 0.81 [95% CI, 0.64-0.99] vs. AUC, 0.81 [0.60-1.00], p = 0.80), but slightly lower performance in external validation (AUC, 0.78 [0.61-0.95] vs. AUC, 0.65 [0.46-0.84], p = 0.23). CONCLUSION: DLAS-based feature extraction showed high reproducibility for first-order features from anatomical and physiological MRI, and comparable diagnostic performance to human manual segmentations in the identification of pseudoprogression, supporting the utility of DLAS in quantitative MRI analysis. KEY POINTS: • Deep learning-based automatic segmentation (DLAS) enables fast and robust feature extraction from diffusion- and perfusion-weighted MRI. • DLAS showed high reproducibility in first-order feature extraction from anatomical, diffusion, and perfusion MRI across two centers. • DLAS-based radiomics features showed comparable diagnostic accuracy to manual segmentations in post-treatment glioblastoma.

Oropharyngeal squamous cell carcinoma: radiomic machine-learning classifiers from multiparametric MR images for determination of HPV infection status.

We investigated the ability of machine-learning classifiers on radiomics from pre-treatment multiparametric magnetic resonance imaging (MRI) to accurately predict human papillomavirus (HPV) status in patients with oropharyngeal squamous cell carcinoma (OPSCC). This retrospective study collected data of 60 patients (48 HPV-positive and 12 HPV-negative) with newly diagnosed histopathologically proved OPSCC, who underwent head and neck MRIs consisting of axial T1WI, T2WI, CE-T1WI, and apparent diffusion coefficient (ADC) maps from diffusion-weighted imaging (DWI). The median age was 59 years (the range being 35 to 85 years), and 83.3% of patients were male. The imaging data were randomised into a training set (32 HPV-positive and 8 HPV-negative OPSCC) and a test set (16 HPV-positive and 4 HPV-negative OPSCC) in each fold. 1618 quantitative features were extracted from manually delineated regions-of-interest of primary tumour and one definite lymph node in each sequence. After feature selection by using the least absolute shrinkage and selection operator (LASSO), three different machine-learning classifiers (logistic regression, random forest, and XG boost) were trained and compared in the setting of various combinations between four sequences. The highest diagnostic accuracies were achieved when using all sequences, and the difference was significant only when the combination did not include the ADC map. Using all sequences, logistic regression and the random forest classifier yielded higher accuracy compared with the that of the XG boost classifier, with mean area under curve (AUC) values of 0.77, 0.76, and 0.71, respectively. The machine-learning classifier of non-invasive and quantitative radiomics signature could guide the classification of the HPV status.

Active learning for accuracy enhancement of semantic segmentation with CNN-corrected label curations: Evaluation on kidney segmentation in abdominal CT.

Segmentation is fundamental to medical image analysis. Recent advances in fully convolutional networks has enabled automatic segmentation; however, high labeling efforts and difficulty in acquiring sufficient and high-quality training data is still a challenge. In this study, a cascaded 3D U-Net with active learning to increase training efficiency with exceedingly limited data and reduce labeling efforts is proposed. Abdominal computed tomography images of 50 kidneys were used for training. In stage I, 20 kidneys with renal cell carcinoma and four substructures were used for training by manually labelling ground truths. In stage II, 20 kidneys from the previous stage and 20 newly added kidneys were used with convolutional neural net (CNN)-corrected labelling for the newly added data. Similarly, in stage III, 50 kidneys were used. The Dice similarity coefficient was increased with the completion of each stage, and shows superior performance when compared with a recent segmentation network based on 3D U-Net. The labeling time for CNN-corrected segmentation was reduced by more than half compared to that in manual segmentation. Active learning was therefore concluded to be capable of reducing labeling efforts through CNN-corrected segmentation and increase training efficiency by iterative learning with limited data.

Machine Learning Approach to Identify Stroke Within 4.5 Hours.

Background and Purpose- We aimed to investigate the ability of machine learning (ML) techniques analyzing diffusion-weighted imaging (DWI) and fluid-attenuated inversion recovery (FLAIR) magnetic resonance imaging to identify patients within the recommended time window for thrombolysis. Methods- We analyzed DWI and FLAIR images of consecutive patients with acute ischemic stroke within 24 hours of clear symptom onset by applying automatic image processing approaches. These processes included infarct segmentation, DWI, and FLAIR imaging registration and image feature extraction. A total of 89 vector features from each image sequence were captured and used in the ML. Three ML models were developed to estimate stroke onset time for binary classification (≤4.5 hours): logistic regression, support vector machine, and random forest. To evaluate the performance of ML models, the sensitivity and specificity for identifying patients within 4.5 hours were compared with the sensitivity and specificity of human readings of DWI-FLAIR mismatch. Results- Data from a total of 355 patients were analyzed. DWI-FLAIR mismatch from human readings identified patients within 4.5 hours of symptom onset with 48.5% sensitivity and 91.3% specificity. ML algorithms had significantly greater sensitivities than human readers (75.8% for logistic regression, P=0.020; 72.7% for support vector machine, P=0.033; 75.8% for random forest, P=0.013) in detecting patients within 4.5 hours, but their specificities were comparable (82.6% for logistic regression, P=0.157; 82.6% for support vector machine, P=0.157; 82.6% for random forest, P=0.157). Conclusions- ML algorithms using multiple magnetic resonance imaging features were feasible even more sensitive than human readings in identifying patients with stroke within the time window for acute thrombolysis.

Radiomic features and multilayer perceptron network classifier: a robust MRI classification strategy for distinguishing glioblastoma from primary central nervous system lymphoma.

We aimed to establish a high-performing and robust classification strategy, using magnetic resonance imaging (MRI), along with combinations of feature extraction and selection in human and machine learning using radiomics or deep features by employing a small dataset. Using diffusion and contrast-enhanced T1-weighted MR images obtained from patients with glioblastomas and primary central nervous system lymphomas, classification task was assigned to a combination of radiomic features and (1) supervised machine learning after feature selection or (2) multilayer perceptron (MLP) network; or MR image input without radiomic feature extraction to (3) two neuro-radiologists or (4) an end-to-end convolutional neural network (CNN). The results showed similar high performance in generalized linear model (GLM) classifier and MLP using radiomics features in the internal validation set, but MLP network remained robust in the external validation set obtained using different MRI protocols. CNN showed the lowest performance in both validation sets. Our results reveal that a combination of radiomic features and MLP network classifier serves a high-performing and generalizable model for classification task for a small dataset with heterogeneous MRI protocols.