Predicting the presence of adherent perinephric fat using MRI radiomics combined with machine learning.

OBJECTIVES: Adherent perinephric fat (APF) poses significant challenges to surgical procedures. This study aimed to evaluate the usefulness of machine learning algorithms combined with MRI-based radiomics features for predicting the presence of APF. MATERIALS AND METHODS: Patients with renal cell carcinoma who underwent surgery between April 2019 and February 2022 at Chonnam National University Hwasun Hospital were retrospectively screened, and 119 patients included. Twenty-one and seventeen patients were set aside for the internal and external test sets, respectively. Pre-operative T1-weighted MRI acquired at 60 s following a contrast injection (T1w-60) were collected. For each T1w-60 data, two regions of interest (ROIs) were manually drawn: the perinephric fat tissue and an aorta segment on the same level as the targeted kidney. Preprocessing steps included resizing voxels, N4 Bias Correction filtering, and aorta-based normalization. For each patient, 851 radiomics features were extracted from the ROI of perinephric fat tissue. Gender and BMI were added as clinical factors. Least Absolute Shrinkage and Selection Operator was adopted for feature selection. We trained and evaluated five models using a 4-fold cross validation. The final model was chosen based on the highest mean AUC across four folds. The performance of the final model was evaluated on the internal and external test sets. RESULTS: A total of 15 features were selected in the final set. The final model achieved the accuracy, sensitivity, specificity, and AUC of 81% (95% confidence interval, 61.9-95.2%), 72.7% (42.9-100%), 90% (66.7-100%), and 0.855 (0.615-1.0), respectively on the internal test set, and 88.2% (70.6-100%), 100% (100-100%), 80% (50%-100%), 0.971 (0.871-1.0), respectively on the external test set. CONCLUSIONS: Our study demonstrated the feasibility of machine learning algorithms trained with MRI-based radiomics features for APF prediction. Further studies with a multi-center approach are necessary to validate our findings.

Deep-learning model for evaluating histopathology of acute renal tubular injury.

Tubular injury is the most common cause of acute kidney injury. Histopathological diagnosis may help distinguish between the different types of acute kidney injury and aid in treatment. To date, a limited number of study has used deep-learning models to assist in the histopathological diagnosis of acute kidney injury. This study aimed to perform histopathological segmentation to identify the four structures of acute renal tubular injury using deep-learning models. A segmentation model was used to classify tubule-specific injuries following cisplatin treatment. A total of 45 whole-slide images with 400 generated patches were used in the segmentation model, and 27,478 annotations were created for four classes: glomerulus, healthy tubules, necrotic tubules, and tubules with casts. A segmentation model was developed using the DeepLabV3 architecture with a MobileNetv3-Large backbone to accurately identify the four histopathological structures associated with acute renal tubular injury in PAS-stained mouse samples. In the segmentation model for four structures, the highest Intersection over Union and the Dice coefficient were obtained for the segmentation of the "glomerulus" class, followed by "necrotic tubules," "healthy tubules," and "tubules with cast" classes. The overall performance of the segmentation algorithm for all classes in the test set included an Intersection over Union of 0.7968 and a Dice coefficient of 0.8772. The Dice scores for the glomerulus, healthy tubules, necrotic tubules, and tubules with cast are 91.78 ± 11.09, 87.37 ± 4.02, 88.08 ± 6.83, and 83.64 ± 20.39%, respectively. The utilization of deep learning in a predictive model has demonstrated promising performance in accurately identifying the degree of injured renal tubules. These results may provide new opportunities for the application of the proposed methods to evaluate renal pathology more effectively.

Accurate prediction of pure uric acid urinary stones in clinical context via a combination of radiomics and machine learning.

PURPOSE: Oral chemolysis is an effective and non-invasive treatment for uric acid urinary stones. This study aimed to classify urinary stones into either pure uric acid (pUA) or other composition (Others) using non-contrast-enhanced computed tomography scans (NCCTs). METHODS: Instances managed at our institution from 2019 to 2021 were screened. They were labeled as either pUA or Others based upon composition analyses, and randomly split into training or testing data set. Several instances contained multiple NCCTs which were all collected. In each of NCCTs, individual urinary stone was treated as individual sample. From manually drawn volumes of interest, we extracted original and wavelet radiomics features for each sample. The most important features were then selected via the Least Absolute Shrinkage and Selection Operator for building the final model on a Support Vector Machine. Performance on the testing set was evaluated via accuracy, sensitivity, specificity, and area under the precision-recall curve (AUPRC). RESULTS: There were 302 instances, of which 118 had pUA urinary stones, generating 576 samples in total. From 851 original and wavelet radiomics features extracted for each sample, 10 most important features were ultimately selected. On the testing data set, accuracy, sensitivity, specificity, and AUPRC were 93.9%, 97.9%, 92.2%, and 0.958, respectively, for per-sample prediction, and 90.8%, 100%, 87.5%, and 0.902, respectively, for per-instance prediction. CONCLUSION: The machine learning algorithm trained with radiomics features from NCCTs can accurately predict pUA urinary stones. Our work suggests a potential assisting tool for stone disease treatment selection.

Predicting Histologic Grade of Meningiomas Using a Combined Model of Radiomic and Clinical Imaging Features from Preoperative MRI.

Meningiomas are common primary brain tumors, and their accurate preoperative grading is crucial for treatment planning. This study aimed to evaluate the value of radiomics and clinical imaging features in predicting the histologic grade of meningiomas from preoperative MRI. We retrospectively reviewed patients with intracranial meningiomas from two hospitals. Preoperative MRIs were analyzed for tumor and edema volumes, enhancement patterns, margins, and tumor-brain interfaces. Radiomics features were extracted, and machine learning models were employed to predict meningioma grades. A total of 212 patients were included. In the training group (Hospital 1), significant differences were observed between low-grade and high-grade meningiomas in terms of tumor volume (p = 0.012), edema volume (p = 0.004), enhancement (p = 0.001), margin (p < 0.001), and tumor-brain interface (p < 0.001). Five radiomics features were selected for model development. The prediction model for radiomics features demonstrated an average validation accuracy of 0.74, while the model for clinical imaging features showed an average validation accuracy of 0.69. When applied to external test data (Hospital 2), the radiomics model achieved an area under the receiver operating characteristics curve (AUC) of 0.72 and accuracy of 0.69, while the clinical imaging model achieved an AUC of 0.82 and accuracy of 0.81. An improved performance was obtained from the model constructed by combining radiomics and clinical imaging features. In the combined model, the AUC and accuracy for meningioma grading were 0.86 and 0.73, respectively. In conclusion, this study demonstrates the potential value of radiomics and clinical imaging features in predicting the histologic grade of meningiomas. The combination of both radiomics and clinical imaging features achieved the highest AUC among the models. Therefore, the combined model of radiomics and clinical imaging features may offer a more effective tool for predicting clinical outcomes in meningioma patients.

Prediction of Residual Axillary Nodal Metastasis Following Neoadjuvant Chemotherapy for Breast Cancer: Radiomics Analysis Based on Chest Computed Tomography.

OBJECTIVE: To evaluate the diagnostic performance of chest computed tomography (CT)-based qualitative and radiomics models for predicting residual axillary nodal metastasis after neoadjuvant chemotherapy (NAC) for patients with clinically node-positive breast cancer. MATERIALS AND METHODS: This retrospective study included 226 women (mean age, 51.4 years) with clinically node-positive breast cancer treated with NAC followed by surgery between January 2015 and July 2021. Patients were randomly divided into the training and test sets (4:1 ratio). The following predictive models were built: a qualitative CT feature model using logistic regression based on qualitative imaging features of axillary nodes from the pooled data obtained using the visual interpretations of three radiologists; three radiomics models using radiomics features from three (intranodal, perinodal, and combined) different regions of interest (ROIs) delineated on pre-NAC CT and post-NAC CT using a gradient-boosting classifier; and fusion models integrating clinicopathologic factors with the qualitative CT feature model (referred to as clinical-qualitative CT feature models) or with the combined ROI radiomics model (referred to as clinical-radiomics models). The area under the curve (AUC) was used to assess and compare the model performance. RESULTS: Clinical N stage, biological subtype, and primary tumor response indicated by imaging were associated with residual nodal metastasis during the multivariable analysis (all P < 0.05). The AUCs of the qualitative CT feature model and radiomics models (intranodal, perinodal, and combined ROI models) according to post-NAC CT were 0.642, 0.812, 0.762, and 0.832, respectively. The AUCs of the clinical-qualitative CT feature model and clinical-radiomics model according to post-NAC CT were 0.740 and 0.866, respectively. CONCLUSION: CT-based predictive models showed good diagnostic performance for predicting residual nodal metastasis after NAC. Quantitative radiomics analysis may provide a higher level of performance than qualitative CT features models. Larger multicenter studies should be conducted to confirm their performance.

A Radiomics Approach on Chest CT Distinguishes Primary Lung Cancer from Solitary Lung Metastasis in Colorectal Cancer Patients.

PURPOSE: This study utilized a radiomics approach combined with a machine learning algorithm to distinguish primary lung cancer (LC) from solitary lung metastasis (LM) in colorectal cancer (CRC) patients with a solitary pulmonary nodule (SPN). MATERIALS AND METHODS: In a retrospective study, 239 patients who underwent chest computerized tomography (CT) at three different institutions between 2011 and 2019 and were diagnosed as primary LC or solitary LM were included. The data from the first institution were divided into training and internal testing datasets. The data from the second and third institutions were used as an external testing dataset. Radiomic features were extracted from the intra and perinodular regions of interest (ROI). After a feature selection process, Support vector machine (SVM) was used to train models for classifying between LC and LM. The performances of the SVM classifiers were evaluated with both the internal and external testing datasets. The performances of the model were compared to those of two radiologists who reviewed the CT images of the testing datasets for the binary prediction of LC versus LM. RESULTS: The SVM classifier trained with the radiomic features from the intranodular ROI and achieved the sensitivity/specificity of 0.545/0.828 in the internal test dataset, and 0.833/0.964 in the external test dataset, respectively. The SVM classifier trained with the combined radiomic features from the intra- and perinodular ROIs achieved the sensitivity/specificity of 0.545/0.966 in the internal test dataset, and 0.833/1.000 in the external test data set, respectively. Two radiologists demonstrated the sensitivity/specificity of 0.545/0.966 and 0.636/0.828 in the internal test dataset, and 0.917/0.929 and 0.833/0.929 in the external test dataset, which were comparable to the performance of the model trained with the combined radiomics features. CONCLUSION: Our results suggested that the machine learning classifiers trained using radiomics features of SPN in CRC patients can be used to distinguish the primary LC and the solitary LM with a similar level of performance to radiologists.

Prediction of the histologic upgrade of ductal carcinoma in situ using a combined radiomics and machine learning approach based on breast dynamic contrast-enhanced magnetic resonance imaging.

Objective: To investigate whether support vector machine (SVM) trained with radiomics features based on breast magnetic resonance imaging (MRI) could predict the upgrade of ductal carcinoma in situ (DCIS) diagnosed by core needle biopsy (CNB) after surgical excision. Materials and methods: This retrospective study included a total of 349 lesions from 346 female patients (mean age, 54 years) diagnosed with DCIS by CNB between January 2011 and December 2017. Based on histological confirmation after surgery, the patients were divided into pure (n = 198, 56.7%) and upgraded DCIS (n = 151, 43.3%). The entire dataset was randomly split to training (80%) and test sets (20%). Radiomics features were extracted from the intratumor region-of-interest, which was semi-automatically drawn by two radiologists, based on the first subtraction images from dynamic contrast-enhanced T1-weighted MRI. A least absolute shrinkage and selection operator (LASSO) was used for feature selection. A 4-fold cross validation was applied to the training set to determine the combination of features used to train SVM for classification between pure and upgraded DCIS. Sensitivity, specificity, accuracy, and area under the receiver-operating characteristic curve (AUC) were calculated to evaluate the model performance using the hold-out test set. Results: The model trained with 9 features (Energy, Skewness, Surface Area to Volume ratio, Gray Level Non Uniformity, Kurtosis, Dependence Variance, Maximum 2D diameter Column, Sphericity, and Large Area Emphasis) demonstrated the highest 4-fold mean validation accuracy and AUC of 0.724 (95% CI, 0.619-0.829) and 0.742 (0.623-0.860), respectively. Sensitivity, specificity, accuracy, and AUC using the test set were 0.733 (0.575-0.892) and 0.7 (0.558-0.842), 0.714 (0.608-0.820) and 0.767 (0.651-0.882), respectively. Conclusion: Our study suggested that the combined radiomics and machine learning approach based on preoperative breast MRI may provide an assisting tool to predict the histologic upgrade of DCIS.

Investigating the Feasibility of In Vivo Perfusion Imaging Methods for Spinal Cord Using Hyperpolarized [13C]t-Butanol and [13C,15N2]Urea.

PURPOSE: This study examined the feasibility of using two novel agents, hyperpolarized [13C]t-butanol and [13C,15N2]urea, for assessing in vivo perfusion of the intact spinal cord in rodents. Due to their distinct permeabilities to blood brain barrier (BBB), we hypothesized that [13C]t-butanol and [13C,15N2]urea exhibit unique 13C signal characteristics in the spinal cord. PROCEDURES: Dynamic 13C t-butanol MRI data were acquired from healthy Long-Evans rats using a symmetric, ramp-sampled, partial-Fourier 13C echo-planar imaging sequence after the injection of hyperpolarized [13C]t-butanol solution. In subsequent scans, dynamic 13C urea MRI data were acquired after the injection of hyperpolarized [13C,15N2]urea. The SNRs of t-butanol and urea were calculated for regions corresponding to spine, supratentorial brain, and blood vessels and plotted over time. Mean peak SNR and AUC were calculated from the dynamic plots for each region and compared between t-butanol and urea. RESULTS: In spine and supratentorial brain, the mean peak SNR and AUC of t-butanol were significantly higher than those of urea (p < 0.05). In contrast, urea was predominantly contained within vasculature and exhibited significantly higher levels of mean peak SNR and AUC compared to t-butanol in blood vessels (p < 0.05). CONCLUSION: This study has demonstrated the feasibility of using hyperpolarized [13C]t-butanol and [13C,15N2]urea for assessing in vivo perfusion in cervical spinal cord. Due to differences in blood-brain barrier permeability, t-butanol rapidly crossed the blood-brain barrier and diffused into spine and brain tissue, while urea predominantly remained in vasculature. The results from this study suggest that this technique may provide unique non-invasive imaging tracers that are able to directly monitor hemodynamic processes in the normal and injured spinal cord.

Predicting Underestimation of Invasive Cancer in Patients with Core-Needle-Biopsy-Diagnosed Ductal Carcinoma In Situ Using Deep Learning Algorithms.

The prediction of an occult invasive component in ductal carcinoma in situ (DCIS) before surgery is of clinical importance because the treatment strategies are different between pure DCIS without invasive component and upgraded DCIS. We demonstrated the potential of using deep learning models for differentiating between upgraded versus pure DCIS in DCIS diagnosed by core-needle biopsy. Preoperative axial dynamic contrast-enhanced magnetic resonance imaging (MRI) data from 352 lesions were used to train, validate, and test three different types of deep learning models. The highest performance was achieved by Recurrent Residual Convolutional Neural Network using Regions of Interest (ROIs) with an accuracy of 75.0% and area under the receiver operating characteristic curve (AUC) of 0.796. Our results suggest that the deep learning approach may provide an assisting tool to predict the histologic upgrade of DCIS and provide personalized treatment strategies to patients with underestimated invasive disease.

Classification of MR-Detected Additional Lesions in Patients With Breast Cancer Using a Combination of Radiomics Analysis and Machine Learning.

OBJECTIVE: This study was conducted in order to investigate the feasibility of using radiomics analysis (RA) with machine learning algorithms based on breast magnetic resonance (MR) images for discriminating malignant from benign MR-detected additional lesions in patients with primary breast cancer. MATERIALS AND METHODS: One hundred seventy-four MR-detected additional lesions (benign, n = 86; malignancy, n = 88) from 158 patients with ipsilateral primary breast cancer from a tertiary medical center were included in this retrospective study. The entire data were randomly split to training (80%) and independent test sets (20%). In addition, 25 patients (benign, n = 21; malignancy, n = 15) from another tertiary medical center were included for the external test. Radiomics features that were extracted from three regions-of-interest (ROIs; intratumor, peritumor, combined) using fat-saturated T1-weighted images obtained by subtracting pre- from postcontrast images (SUB) and T2-weighted image (T2) were utilized to train the support vector machine for the binary classification. A decision tree method was utilized to build a classifier model using clinical imaging interpretation (CII) features assessed by radiologists. Area under the receiver operating characteristic curve (AUROC), accuracy, sensitivity, and specificity were used to compare the diagnostic performance. RESULTS: The RA models trained using radiomics features from the intratumor-ROI showed comparable performance to the CII model (accuracy, AUROC: 73.3%, 69.6% for the SUB RA model; 70.0%, 75.1% for the T2 RA model; 73.3%, 72.0% for the CII model). The diagnostic performance increased when the radiomics and CII features were combined to build a fusion model. The fusion model that combines the CII features and radiomics features from multiparametric MRI data demonstrated the highest performance with an accuracy of 86.7% and an AUROC of 91.1%. The external test showed a similar pattern where the fusion models demonstrated higher levels of performance compared with the RA- or CII-only models. The accuracy and AUROC of the SUB+T2 RA+CII model in the external test were 80.6% and 91.4%, respectively. CONCLUSION: Our study demonstrated the feasibility of using RA with machine learning approach based on multiparametric MRI for quantitatively characterizing MR-detected additional lesions. The fusion model demonstrated an improved diagnostic performance over the models trained with either RA or CII alone.

Characterization of Distinctive In Vivo Metabolism between Enhancing and Non-Enhancing Gliomas Using Hyperpolarized Carbon-13 MRI.

The development of hyperpolarized carbon-13 (13C) metabolic MRI has enabled the sensitive and noninvasive assessment of real-time in vivo metabolism in tumors. Although several studies have explored the feasibility of using hyperpolarized 13C metabolic imaging for neuro-oncology applications, most of these studies utilized high-grade enhancing tumors, and little is known about hyperpolarized 13C metabolic features of a non-enhancing tumor. In this study, 13C MR spectroscopic imaging with hyperpolarized [1-13C]pyruvate was applied for the differential characterization of metabolic profiles between enhancing and non-enhancing gliomas using rodent models of glioblastoma and a diffuse midline glioma. Distinct metabolic profiles were found between the enhancing and non-enhancing tumors, as well as their contralateral normal-appearing brain tissues. The preliminary results from this study suggest that the characterization of metabolic patterns from hyperpolarized 13C imaging between non-enhancing and enhancing tumors may be beneficial not only for understanding distinct metabolic features between the two lesions, but also for providing a basis for understanding 13C metabolic processes in ongoing clinical trials with neuro-oncology patients using this technology.

In Vivo Assessment of Metabolic Abnormality in Alport Syndrome Using Hyperpolarized [1-13C] Pyruvate MR Spectroscopic Imaging.

Alport Syndrome (AS) is a genetic disorder characterized by impaired kidney function. The development of a noninvasive tool for early diagnosis and monitoring of renal function during disease progression is of clinical importance. Hyperpolarized 13C MRI is an emerging technique that enables non-invasive, real-time measurement of in vivo metabolism. This study aimed to investigate the feasibility of using this technique for assessing changes in renal metabolism in the mouse model of AS. Mice with AS demonstrated a significant reduction in the level of lactate from 4- to 7-week-old, while the levels of lactate were unchanged in the control mice over time. This reduction in lactate production in the AS group accompanied a significant increase of PEPCK expression levels, indicating that the disease progression in AS triggered the gluconeogenic pathway and might have resulted in a decreased lactate pool size and a subsequent reduction in pyruvate-to-lactate conversion. Additional metabolic imaging parameters, including the level of lactate and pyruvate, were found to be different between the AS and control groups. These preliminary results suggest that hyperpolarized 13C MRI might provide a potential noninvasive tool for the characterization of disease progression in AS.

Differentiating Radiation Necrosis from Brain Tumor Using Hyperpolarized Carbon-13 MR Metabolic Imaging.

PURPOSE: Differentiation between radiation-induced necrosis and tumor recurrence is crucial to determine proper management strategies but continues to be one of the central challenges in neuro-oncology. We hypothesized that hyperpolarized 13C MRI, a unique technique to measure real-time in vivo metabolism, would distinguish radiation necrosis from tumor on the basis of cell-intrinsic metabolic differences. The purpose of this study was to explore the feasibility of using hyperpolarized [1-13C]pyruvate for differentiating radiation necrosis from brain tumors. PROCEDURES: Radiation necrosis was initiated by employing a CT-guided 80-Gy single-dose irradiation of a half cerebrum in mice (n = 7). Intracerebral tumor was modeled with two orthotopic mouse models: GL261 glioma (n = 6) and Lewis lung carcinoma (LLC) metastasis (n = 7). 13C 3D MR spectroscopic imaging data were acquired following hyperpolarized [1-13C]pyruvate injection approximately 89 and 14 days after treatment for irradiated and tumor-bearing mice, respectively. The ratio of lactate to pyruvate (Lac/Pyr), normalized lactate, and pyruvate in contrast-enhancing lesion was compared between the radiation-induced necrosis and brain tumors. Histopathological analysis was performed from resected brains. RESULTS: Conventional MRI exhibited typical radiographic features of radiation necrosis and brain tumor with large areas of contrast enhancement and T2 hyperintensity in all animals. Normalized lactate in radiation necrosis (0.10) was significantly lower than that in glioma (0.26, P = .004) and LLC metastatic tissue (0.25, P = .00007). Similarly, Lac/Pyr in radiation necrosis (0.18) was significantly lower than that in glioma (0.55, P = .00008) and LLC metastasis (0.46, P = .000008). These results were consistent with histological findings where tumor-bearing brains were highly cellular, while irradiated brains exhibited pathological markers consistent with reparative changes from radiation necrosis. CONCLUSION: Hyperpolarized 13C MR metabolic imaging of pyruvate is a noninvasive imaging method that differentiates between radiation necrosis and brain tumors, providing a groundwork for further clinical investigation and translation for the improved management of patients with brain tumors.

Assessment of deep neural networks for the diagnosis of benign and malignant skin neoplasms in comparison with dermatologists: A retrospective validation study.

BACKGROUND: The diagnostic performance of convolutional neural networks (CNNs) for diagnosing several types of skin neoplasms has been demonstrated as comparable with that of dermatologists using clinical photography. However, the generalizability should be demonstrated using a large-scale external dataset that includes most types of skin neoplasms. In this study, the performance of a neural network algorithm was compared with that of dermatologists in both real-world practice and experimental settings. METHODS AND FINDINGS: To demonstrate generalizability, the skin cancer detection algorithm (https://rcnn.modelderm.com) developed in our previous study was used without modification. We conducted a retrospective study with all single lesion biopsied cases (43 disorders; 40,331 clinical images from 10,426 cases: 1,222 malignant cases and 9,204 benign cases); mean age (standard deviation [SD], 52.1 [18.3]; 4,701 men [45.1%]) were obtained from the Department of Dermatology, Severance Hospital in Seoul, Korea between January 1, 2008 and March 31, 2019. Using the external validation dataset, the predictions of the algorithm were compared with the clinical diagnoses of 65 attending physicians who had recorded the clinical diagnoses with thorough examinations in real-world practice. In addition, the results obtained by the algorithm for the data of randomly selected batches of 30 patients were compared with those obtained by 44 dermatologists in experimental settings; the dermatologists were only provided with multiple images of each lesion, without clinical information. With regard to the determination of malignancy, the area under the curve (AUC) achieved by the algorithm was 0.863 (95% confidence interval [CI] 0.852-0.875), when unprocessed clinical photographs were used. The sensitivity and specificity of the algorithm at the predefined high-specificity threshold were 62.7% (95% CI 59.9-65.1) and 90.0% (95% CI 89.4-90.6), respectively. Furthermore, the sensitivity and specificity of the first clinical impression of 65 attending physicians were 70.2% and 95.6%, respectively, which were superior to those of the algorithm (McNemar test; p < 0.0001). The positive and negative predictive values of the algorithm were 45.4% (CI 43.7-47.3) and 94.8% (CI 94.4-95.2), respectively, whereas those of the first clinical impression were 68.1% and 96.0%, respectively. In the reader test conducted using images corresponding to batches of 30 patients, the sensitivity and specificity of the algorithm at the predefined threshold were 66.9% (95% CI 57.7-76.0) and 87.4% (95% CI 82.5-92.2), respectively. Furthermore, the sensitivity and specificity derived from the first impression of 44 of the participants were 65.8% (95% CI 55.7-75.9) and 85.7% (95% CI 82.4-88.9), respectively, which are values comparable with those of the algorithm (Wilcoxon signed-rank test; p = 0.607 and 0.097). Limitations of this study include the exclusive use of high-quality clinical photographs taken in hospitals and the lack of ethnic diversity in the study population. CONCLUSIONS: Our algorithm could diagnose skin tumors with nearly the same accuracy as a dermatologist when the diagnosis was performed solely with photographs. However, as a result of limited data relevancy, the performance was inferior to that of actual medical examination. To achieve more accurate predictive diagnoses, clinical information should be integrated with imaging information.

Automatic Assessment of ASPECTS Using Diffusion-Weighted Imaging in Acute Ischemic Stroke Using Recurrent Residual Convolutional Neural Network.

The early detection and rapid quantification of acute ischemic lesions play pivotal roles in stroke management. We developed a deep learning algorithm for the automatic binary classification of the Alberta Stroke Program Early Computed Tomographic Score (ASPECTS) using diffusion-weighted imaging (DWI) in acute stroke patients. Three hundred and ninety DWI datasets with acute anterior circulation stroke were included. A classifier algorithm utilizing a recurrent residual convolutional neural network (RRCNN) was developed for classification between low (1-6) and high (7-10) DWI-ASPECTS groups. The model performance was compared with a pre-trained VGG16, Inception V3, and a 3D convolutional neural network (3DCNN). The proposed RRCNN model demonstrated higher performance than the pre-trained models and 3DCNN with an accuracy of 87.3%, AUC of 0.941, and F1-score of 0.888 for classification between the low and high DWI-ASPECTS groups. These results suggest that the deep learning algorithm developed in this study can provide a rapid assessment of DWI-ASPECTS and may serve as an ancillary tool that can assist physicians in making urgent clinical decisions.

Augmented Intelligence Dermatology: Deep Neural Networks Empower Medical Professionals in Diagnosing Skin Cancer and Predicting Treatment Options for 134 Skin Disorders.

Although deep learning algorithms have demonstrated expert-level performance, previous efforts were mostly binary classifications of limited disorders. We trained an algorithm with 220,680 images of 174 disorders and validated it using Edinburgh (1,300 images; 10 disorders) and SNU datasets (2,201 images; 134 disorders). The algorithm could accurately predict malignancy, suggest primary treatment options, render multi-class classification among 134 disorders, and improve the performance of medical professionals. The area under the curves for malignancy detection were 0.928 ± 0.002 (Edinburgh) and 0.937 ± 0.004 (SNU). The area under the curves of primary treatment suggestion (SNU) were 0.828 ± 0.012, 0.885 ± 0.006, 0.885 ± 0.006, and 0.918 ± 0.006 for steroids, antibiotics, antivirals, and antifungals, respectively. For multi-class classification, the mean top-1 and top-5 accuracies were 56.7 ± 1.6% and 92.0 ± 1.1% (Edinburgh) and 44.8 ± 1.2% and 78.1 ± 0.3% (SNU), respectively. With the assistance of our algorithm, the sensitivity and specificity of 47 clinicians (21 dermatologists and 26 dermatology residents) for malignancy prediction (SNU; 240 images) were improved by 12.1% (P < 0.0001) and 1.1% (P < 0.0001), respectively. The malignancy prediction sensitivity of 23 non-medical professionals was significantly increased by 83.8% (P < 0.0001). The top-1 and top-3 accuracies of four doctors in the multi-class classification of 134 diseases (SNU; 2,201 images) were increased by 7.0% (P = 0.045) and 10.1% (P = 0.0020), respectively. The results suggest that our algorithm may serve as augmented intelligence that can empower medical professionals in diagnostic dermatology.

Effects of sitting posture and bolus volume on activation of swallowing-related muscles.

BACKGROUND: The pharyngeal phase is a particularly important clinical factor related to swallowing dysfunctions. Head and neck posture, as well as bolus volume, are important factors affecting the pharyngeal stages of normal swallowing. OBJECTIVE: The aim of our study was to identify the effects of sitting posture and bolus volume on the activation of swallowing-related muscles. MATERIALS AND METHODS: Twenty-four subjects participated in the study. The subjects were positioned in three sitting postures-slump sitting (SS), lumbo-pelvic upright sitting (LUS), and thoracic upright sitting (TUS). While sitting in the chair, the subject was instructed to swallow 10 and 20 mL of water. Surface electromyography (EMG) was used to measure the muscle activity of the supra-hyoid (SH) and infra-hyoid (IH) muscles. Also, sitting posture alignment (head, cervical and shoulder angle) was also performed. Data were analysed with a repeated measures analysis of variance (RMANOVA) using a generalised linear model. RESULTS: There was no significant difference in terms of the head angle (P = .395). However, significant differences were found in relation to the cervical angle (P < .001) and shoulder angle (P < .001). The TUS produced the lowest SH EMG activity (P = .001), in comparison to SS and LUS. The bolus volume for 20 mL showed greater SH and IH EMG activity (P < .001) than did the bolus volume for 10 mL. CONCLUSIONS: Correcting sitting posture from SS to TUS may better assist swallowing-related muscles with less effort, irrespective of the bolus volume.

Comparison between 8- and 32-channel phased-array receive coils for in vivo hyperpolarized 13 C imaging of the human brain.

PURPOSE: To compare the performance of an 8-channel surface coil/clamshell transmitter and 32-channel head array coil/birdcage transmitter for hyperpolarized 13 C brain metabolic imaging. METHODS: To determine the field homogeneity of the radiofrequency transmitters, B1 + mapping was performed on an ethylene glycol head phantom and evaluated by means of the double angle method. Using a 3D echo-planar imaging sequence, coil sensitivity and noise-only phantom data were acquired with the 8- and 32-channel receiver arrays, and compared against data from the birdcage in transceiver mode. Multislice frequency-specific 13 C dynamic echo-planar imaging was performed on a patient with a brain tumor for each hardware configuration following injection of hyperpolarized [1-13 C]pyruvate. Signal-to-noise ratio (SNR) was evaluated from pre-whitened phantom and temporally summed patient data after coil combination based on optimal weights. RESULTS: The birdcage transmitter produced more uniform B1 + compared with the clamshell: 0.07 versus 0.12 (fractional error). Phantom experiments conducted with matched lateral housing separation demonstrated 8- versus 32-channel mean transceiver-normalized SNR performance: 0.91 versus 0.97 at the head center; 6.67 versus 2.08 on the sides; 0.66 versus 2.73 at the anterior; and 0.67 versus 3.17 on the posterior aspect. While the 8-channel receiver array showed SNR benefits along lateral aspects, the 32-channel array exhibited greater coverage and a more uniform coil-combined profile. Temporally summed, parameter-normalized patient data showed SNRmean,slice ratios (8-channel/32-channel) ranging 0.5-2.00 from apical to central brain. White matter lactate-to-pyruvate ratios were conserved across hardware: 0.45 ± 0.12 (8-channel) versus 0.43 ± 0.14 (32-channel). CONCLUSION: The 8- and 32-channel hardware configurations each have advantages in particular brain anatomy.

Metabolic Changes in Different Stages of Liver Fibrosis: In vivo Hyperpolarized 13C MR Spectroscopy and Metabolic Imaging.

PURPOSE: The objective was to assess metabolic changes in different stages of liver fibrosis using hyperpolarized C-13 magnetic resonance spectroscopy (MRS) and metabolic imaging. PROCEDURES: Mild and severe liver fibrosis were induced in C3H/HeN mice (n = 14) by injecting thioacetamide (TAA). Other C3H/HeN mice (n = 7) were injected with phosphate buffer saline (PBS) (7.4 pH) as normal controls. Hyperpolarized C-13 MRS was performed on the livers of the mice, which was accompanied by intravoxel incoherent motion (IVIM) diffusion-weighted imaging with 12 b values. The differential metabolite ratios, apparent diffusion coefficient values, and IVIM parameters among the three groups were analyzed by a one-way analysis of variance test. RESULTS: The ratios of [1-13C]lactate/pyruvate, [1-13C]lactate/total carbon (tC), [1-13C]alanine/pyruvate, and [1-13C] alanine/tC were significantly higher in both the mild and severe fibrosis groups than in the normal control group (p < 0.05). While the [1-13C]lactate/pyruvate and [1-13C]lactate/tC ratios were not significantly different between mild and severe fibrosis groups, the ratios of [1-13C]alanine/pyruvate and [1-13C]alanine/tC were significantly higher in the severe fibrosis group than in the mild fibrosis group (p < 0.05). In addition, D* showed a significantly lower value in the severe fibrosis group than in the normal or mild fibrosis groups and negatively correlated with the levels of [1-13C] lactate and [1-13C]alanine. CONCLUSIONS: Our findings suggest that it might be possible to differentiate mild from severe liver fibrosis using the cellular metabolic changes with hyperpolarized C-13 MRS and metabolic imaging.

Translation of Carbon-13 EPI for hyperpolarized MR molecular imaging of prostate and brain cancer patients.

PURPOSE: To develop and translate a metabolite-specific imaging sequence using a symmetric echo planar readout for clinical hyperpolarized (HP) Carbon-13 (13 C) applications. METHODS: Initial data were acquired from patients with prostate cancer (N = 3) and high-grade brain tumors (N = 3) on a 3T scanner. Samples of [1-13 C]pyruvate were polarized for at least 2 h using a 5T SPINlab system operating at 0.8 K. Following injection of the HP substrate, pyruvate, lactate, and bicarbonate (for brain studies) were sequentially excited with a singleband spectral-spatial RF pulse and signal was rapidly encoded with a single-shot echo planar readout on a slice-by-slice basis. Data were acquired dynamically with a temporal resolution of 2 s for prostate studies and 3 s for brain studies. RESULTS: High pyruvate signal was seen throughout the prostate and brain, with conversion to lactate being shown across studies, whereas bicarbonate production was also detected in the brain. No Nyquist ghost artifacts or obvious geometric distortion from the echo planar readout were observed. The average error in center frequency was 1.2 ± 17.0 and 4.5 ± 1.4 Hz for prostate and brain studies, respectively, below the threshold for spatial shift because of bulk off-resonance. CONCLUSION: This study demonstrated the feasibility of symmetric EPI to acquire HP 13 C metabolite maps in a clinical setting. As an advance over prior single-slice dynamic or single time point volumetric spectroscopic imaging approaches, this metabolite-specific EPI acquisition provided robust whole-organ coverage for brain and prostate studies while retaining high SNR, spatial resolution, and dynamic temporal resolution.