Predictive value of the hemispheric magnetic resonance angiography score on the development of indirect pial synangiosis after combined revascularization surgery for adult moyamoya disease.

PURPOSE: It is difficult to precisely predict indirect bypass development in the context of combined bypass procedures in moyamoya disease (MMD). We aimed to investigate the predictive value of magnetic resonance angiography (MRA) signal intensity in the peripheral portion of the major cerebral arteries for indirect bypass development in adult patients with MMD. METHODS: We studied 93 hemispheres from 62 adult patients who underwent combined direct and indirect revascularization between 2005 and 2019 and genetic analysis for RNF213 p.R4810K. The signal intensity of the peripheral portion of the major intracranial arteries during preoperative MRA was graded as a hemispheric MRA score (0-3 in the middle cerebral artery and 0-2 in the anterior cerebral and posterior cerebral arteries, with a high score representing low visibility) according to each vessel's visibility. Postoperative bypass development was qualitatively evaluated using MRA, and we evaluated the correlation between preoperative factors, including the hemispheric MRA score and bypass development, using univariate and multivariate analyses. RESULTS: A good indirect bypass was observed in 70% of the hemispheres. Hemispheric MRA scores were significantly higher in hemispheres with good indirect bypass development than in those with poor indirect bypass development (median: 3 vs. 1; p < 0.0001). Multiple logistic regression analysis revealed hemispheric MRA score as an independent predictor of good indirect bypass development (odds ratio, 2.1; 95% confidence interval, 1.3-3.6; p < 0.01). The low hemispheric MRA score (< 2) and wild-type RNF213 predicted poor indirect bypass development with a specificity of 0.92. CONCLUSION: Hemispheric MRA score was a predictive factor for indirect bypass development in adult patients who underwent a combined bypass procedure for MMD. Predicting poor indirect bypass development may lead to future tailored bypass surgeries for MMD.

Comparing accuracy of machine learning approaches to identifying parathyroid adenomas: Lessons and new directions.

PURPOSE: The purpose of this investigation is to understand the accuracy of machine learning techniques to detect biopsy-proven adenomas from similar appearing lymph nodes and factors that influence accuracy by comparing support vector machine (SVM) and bidirectional Long short-term memory (Bi-LSTM) analyses. This will provide greater insight into how these tools could integrate multidimensional data and aid the detection of parathyroid adenomas consistently and accurately. METHODS: Ninety-nine patients were identified; 93 4D-CTs of patients with pathology-proven parathyroid adenomas were reviewed; 94 parathyroid adenomas and 112 lymph nodes were analyzed. A 2D slice through the lesions in each phase was used to perform sequence classification with ResNet50 as the pre-trained network to construct the Bi-LSTM model, and the mean enhancement curves were used to form an SVM model. The model characteristics and accuracy were calculated for the training and validation data sets. RESULTS: On the training data, the area under the curve (AUC) of the Bi-LSTM was 0.99, while the SVM was 0.95 and statistically significant on the DeLong test. The overall accuracy of the Bi-LSTM on the validation data set was 92 %, while the SVM was 88 %. The accuracy for parathyroid adenomas specifically was 93 % for the Bi-LSTM and 83 % for the SVM model. CONCLUSION: Enhancement characteristics are a distinguishing feature that accurately identifies parathyroid adenomas alone. The Bi-LSTM performs statistically better in identifying parathyroid adenomas than the SVM analysis when using both morphologic and enhancement information to distinguish between parathyroid adenomas and lymph nodes. SUMMARY STATEMENT: The Bi-LSTM more accurately identifies parathyroid adenomas than the SVM analysis, which uses both morphologic and enhancement information to distinguish between parathyroid adenomas and lymph nodes, performs statistically better.

Fully automated 3D machine learning model for HPV status characterization in oropharyngeal squamous cell carcinomas based on CT images.

BACKGROUND: Human papillomavirus (HPV) status plays a major role in predicting oropharyngeal squamous cell carcinoma (OPSCC) survival. This study assesses the accuracy of a fully automated 3D convolutional neural network (CNN) in predicting HPV status using CT images. METHODS: Pretreatment CT images from OPSCC patients were used to train a 3D DenseNet-121 model to predict HPV-p16 status. Performance was evaluated by the ROC Curve (AUC), sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and F1 score. RESULTS: The network achieved a mean AUC of 0.80 ± 0.06. The best-preforming fold had a sensitivity of 0.86 and specificity of 0.92 at the Youden's index. The PPV, NPV, and F1 scores are 0.97, 0.71, and 0.82, respectively. CONCLUSIONS: A fully automated CNN can characterize the HPV status of OPSCC patients with high sensitivity and specificity. Further refinement of this algorithm has the potential to provide a non-invasive tool to guide clinical management.

Mechanical Properties of a 3 Dimensional-Printed Transparent Flexible Resin Used for Vascular Model Simulation Compared with Those of Porcine Arteries.

PURPOSE: To develop a vascular intervention simulation model that replicates the characteristics of a human patient and to compare the mechanical properties of a 3-dimensional (3D)-printed transparent flexible resin with those of porcine arteries using the elastic modulus (E) and kinetic friction coefficient (µk). MATERIALS AND METHODS: Resin plates were created from a transparent flexible resin using a 3D printer. Porcine artery plates were prepared by excising the aorta. E values and the adhesive strengths of the resin and arterial surfaces toward a polyethylene plate, were measured with a tensile-compressive mechanical tester. Resin transparency was measured using an ultraviolet-visible light spectrometer. The µk value of the resin plate surface after applying silicone spray for 1-5 seconds and that of the artery were measured using a translational friction tester. RESULTS: E values differed significantly between the arteries and resin plates at each curing time (0.20 MPa ± 0.04 vs 8.53 MPa ± 2.37 for a curing time of 1 minute; P < .05). The resin was stiffer than the arteries, regardless of the curing times. The visible light transmittance and adhesive strength of the resin decreased as the curing time increased. The adhesive strength of the artery was the lowest. The µk value of the silicone-coated resin surface created by applying silicone for 2-3 seconds (thickness of the silicone layer, 1.6-2.0 µm) was comparable with that of the artery, indicating that the coating imparted a similar slippage to the resin as to the living artery. CONCLUSIONS: A transparent flexible resin is useful for creating a transparent and slippery vascular model for vascular intervention simulation.

Improvement of image quality in diffusion-weighted imaging with model-based deep learning reconstruction for evaluations of the head and neck.

OBJECTIVES: To investigate the utility of deep learning (DL)-based image reconstruction using a model-based approach in head and neck diffusion-weighted imaging (DWI). MATERIALS AND METHODS: We retrospectively analyzed the cases of 41 patients who underwent head/neck DWI. The DWI in 25 patients demonstrated an untreated lesion. We performed qualitative and quantitative assessments in the DWI analyses with both deep learning (DL)- and conventional parallel imaging (PI)-based reconstructions. For the qualitative assessment, we visually evaluated the overall image quality, soft tissue conspicuity, degree of artifact(s), and lesion conspicuity based on a five-point system. In the quantitative assessment, we measured the signal-to-noise ratio (SNR) of the bilateral parotid glands, submandibular gland, the posterior muscle, and the lesion. We then calculated the contrast-to-noise ratio (CNR) between the lesion and the adjacent muscle. RESULTS: Significant differences were observed in the qualitative analysis between the DWI with PI-based and DL-based reconstructions for all of the evaluation items (p < 0.001). In the quantitative analysis, significant differences in the SNR and CNR between the DWI with PI-based and DL-based reconstructions were observed for all of the evaluation items (p = 0.002 ~ p < 0.001). DISCUSSION: DL-based image reconstruction with the model-based technique effectively provided sufficient image quality in head/neck DWI.

MR vessel-encoded arterial spin labeling with the placement of metallic items to visualize the territorial blood flow after extracranial-intracranial bypass surgery: a proof-of-concept study.

BACKGROUND: Depiction of bypass blood flow in patients who received extracranial-intracranial (EC-IC) bypass surgery is important for patient care. PURPOSE: To develop a vessel-encoded arterial spin labeling (VE-ASL) method using surgical staples as a magnetic resonance (MR)-conditional product in patients who received EC-IC bypass surgery. MATERIAL AND METHODS: Pseudo-continuous labeling was used for VE-ASL acquisition with a 3-T MR unit. First, an experimental study was conducted to determine the appropriate number of surgical staples to obtain a spatially sufficient saturation effect. Thereafter, four healthy normal volunteers underwent a VE-ASL study to confirm the sufficiency of the saturation effect to the right or left common carotid artery. Finally, VE-ASL scanning was performed in seven patients after EC-IC bypass surgery to confirm the ability of VE-ASL to visualize the territorial bypass perfusion. All qualitative evaluation was performed by two neuroradiologists using a 3-point grading system (2 = good, 1 = moderate, 0 = poor). RESULTS: A quantity of 200 staples was found to be appropriate for VE-ASL scanning. In healthy volunteers, one neuroradiologist rated the images of all four cases as good, while the other rated three cases as good and one case as moderate. For the seven patients after EC-IC bypass surgery, one neuroradiologist rated all seven cases as good, and the other rated six cases as good and one case as moderate. CONCLUSION: VE-ASL using surgical staples might be useful for the evaluation of territorial bypass perfusion in patients after EC-IC bypass surgery.

New trend in artificial intelligence-based assistive technology for thoracic imaging.

Although there is no solid agreement for artificial intelligence (AI), it refers to a computer system with intelligence similar to that of humans. Deep learning appeared in 2006, and more than 10 years have passed since the third AI boom was triggered by improvements in computing power, algorithm development, and the use of big data. In recent years, the application and development of AI technology in the medical field have intensified internationally. There is no doubt that AI will be used in clinical practice to assist in diagnostic imaging in the future. In qualitative diagnosis, it is desirable to develop an explainable AI that at least represents the basis of the diagnostic process. However, it must be kept in mind that AI is a physician-assistant system, and the final decision should be made by the physician while understanding the limitations of AI. The aim of this article is to review the application of AI technology in diagnostic imaging from PubMed database while particularly focusing on diagnostic imaging in thorax such as lesion detection and qualitative diagnosis in order to help radiologists and clinicians to become more familiar with AI in thorax.

Clinical applications of artificial intelligence in liver imaging.

This review outlines the current status and challenges of the clinical applications of artificial intelligence in liver imaging using computed tomography or magnetic resonance imaging based on a topic analysis of PubMed search results using latent Dirichlet allocation. LDA revealed that "segmentation," "hepatocellular carcinoma and radiomics," "metastasis," "fibrosis," and "reconstruction" were current main topic keywords. Automatic liver segmentation technology using deep learning is beginning to assume new clinical significance as part of whole-body composition analysis. It has also been applied to the screening of large populations and the acquisition of training data for machine learning models and has resulted in the development of imaging biomarkers that have a significant impact on important clinical issues, such as the estimation of liver fibrosis, recurrence, and prognosis of malignant tumors. Deep learning reconstruction is expanding as a new technological clinical application of artificial intelligence and has shown results in reducing contrast and radiation doses. However, there is much missing evidence, such as external validation of machine learning models and the evaluation of the diagnostic performance of specific diseases using deep learning reconstruction, suggesting that the clinical application of these technologies is still in development.

Phase I Randomized Trial of 17 O-Labeled Water: Safety and Feasibility Study of Indirect Proton MRI for the Evaluation of Cerebral Water Dynamics.

BACKGROUND: 17 O-labeled water (PSO17) is a contrast agent developed to measure brain water dynamics and cerebral blood flow. PURPOSE: To evaluate the safety and feasibility of PSO17. STUDY TYPE: Prospective study. SUBJECTS: A total of 12 male healthy volunteers (23.1 ± 1.9 years) were assigned to three groups of four subjects: placebo (normal saline), PSO17 10%, and PSO17 20%. FIELD STRENGTH/SEQUENCE: Dynamic 3D fluid attenuated inversion recovery (FLAIR, fast spin echo with variable refocusing flip angle) scans of the brain were performed with 3-T MRI. ASSESSMENT: Contrast agents were injected 5 minutes after the start of a 10-minute scan. Any symptoms, vital signs, and blood and urine tests were evaluated at five timepoints from preinjection to 4 days after. Blood samples for pharmacokinetic analysis, including half-life (T1/2), maximum fraction (Cmax ), time-to-maximum fraction (Tmax ), and area under the curve (AUC), were collected at 13 timepoints from preinjection to 168 hours after. Regions of interest were set in the cerebral cortex (CC), basal ganglia/thalamus (BG/TM), and white matter (WM), and 17 O concentrations were calculated from signal changes and evaluated using Cmax . STATISTICAL TESTS: All items were compared among the three groups using Tukey-Kramer's honestly significant difference test. Statistical significance was defined as P < 0.5. RESULTS: No safety issues were noted with the intravenous administration of PSO17. The T1/2 was approximately 160 hours, and the AUCs were 1.77 ± 0.10 and 3.75 ± 0.36 in the PSO17 10% and 20% groups, respectively. 17 O fractions calculated from MRI signals were higher in the PSO17 20% group than in the 10% and placebo groups. Significant differences were noted between all pairs of groups in the CC and BG/TM, and between PSO17 20% and both placebo and 10% groups in the WM. DATA CONCLUSION: PSO17 might be considered safe as a contrast medium. Dynamic 3D-FLAIR might detect dose-dependent signal changes and estimate 17 O. EVIDENCE LEVEL: 1 TECHNICAL EFFICACY: Stage 1.

Quantitative Susceptibility Mapping: Basic Methods and Clinical Applications.

Quantitative susceptibility mapping (QSM), one of the advanced MRI techniques for evaluating magnetic susceptibility, offers precise quantitative measurements of spatial distributions of magnetic susceptibility. Magnetic susceptibility describes the magnetizability of a material to an applied magnetic field and is a substance-specific value. Recently, QSM has been widely used to estimate various levels of substances in the brain, including iron, hemosiderin, and deoxyhemoglobin (paramagnetism), as well as calcification (diamagnetism). By visualizing iron distribution in the brain, it is possible to identify anatomic structures that are not evident on conventional images and to evaluate various neurodegenerative diseases. It has been challenging to apply QSM in areas outside the brain because of motion artifacts from respiration and heartbeats, as well as the presence of fat, which has a different frequency to the proton. In this review, the authors provide a brief overview of the theoretical background and analyze methods of converting MRI phase images to QSM. Moreover, we provide an overview of the current clinical applications of QSM. Online supplemental material is available for this article. ©RSNA, 2022.

Prediction of the local treatment outcome in patients with oropharyngeal squamous cell carcinoma using deep learning analysis of pretreatment FDG-PET images.

BACKGROUND: This study aimed to assess the utility of deep learning analysis using pretreatment FDG-PET images to predict local treatment outcome in oropharyngeal squamous cell carcinoma (OPSCC) patients. METHODS: One hundred fifty-four OPSCC patients who received pretreatment FDG-PET were included and divided into training (n = 102) and test (n = 52) sets. The diagnosis of local failure and local progression-free survival (PFS) rates were obtained from patient medical records. In deep learning analyses, axial and coronal images were assessed by three different architectures (AlexNet, GoogLeNET, and ResNet). In the training set, FDG-PET images were analyzed after the data augmentation process for the diagnostic model creation. A multivariate clinical model was also created using a binomial logistic regression model from a patient's clinical characteristics. The test data set was subsequently analyzed for confirmation of diagnostic accuracy. Assessment of local PFS rates was also performed. RESULTS: Training sessions were successfully performed with an accuracy of 74-89%. ROC curve analyses revealed an AUC of 0.61-0.85 by the deep learning model in the test set, whereas it was 0.62 by T-stage, 0.59 by clinical stage, and 0.74 by a multivariate clinical model. The highest AUC (0.85) was obtained with deep learning analysis of ResNet architecture. Cox proportional hazards regression analysis revealed deep learning-based classification by a multivariate clinical model (P < .05), and ResNet (P < .001) was a significant predictor of the treatment outcome. In the Kaplan-Meier analysis, the deep learning-based classification divided the patient's local PFS rate better than the T-stage, clinical stage, and a multivariate clinical model. CONCLUSIONS: Deep learning-based diagnostic model with FDG-PET images indicated its possibility to predict local treatment outcomes in OPSCCs.

Utility of deep learning for the diagnosis of otosclerosis on temporal bone CT.

OBJECTIVE: Diagnosis of otosclerosis on temporal bone CT images is often difficult because the imaging findings are frequently subtle. Our aim was to assess the utility of deep learning analysis in diagnosing otosclerosis on temporal bone CT images. METHODS: A total of 198 temporal bone CT images were divided into the training set (n = 140) and the test set (n = 58). The final diagnosis (otosclerosis-positive or otosclerosis-negative) was determined by an experienced senior radiologist who carefully reviewed all 198 temporal bone CT images while correlating with clinical and intraoperative findings. In deep learning analysis, a rectangular target region that includes the area of the fissula ante fenestram was extracted and fed into the deep learning training sessions to create a diagnostic model. Transfer learning was used with the deep learning model architectures of AlexNet, VGGNet, GoogLeNet, and ResNet. The test data set was subsequently analyzed using these models and by another radiologist with 3 years of experience in neuroradiology following completion of a neuroradiology fellowship. The performance of the radiologist and the deep learning models was determined using the senior radiologist's diagnosis as the gold standard. RESULTS: The diagnostic accuracies were 0.89, 0.72, 0.81, 0.86, and 0.86 for the subspecialty trained radiologist, AlexNet, VGGNet, GoogLeNet, and ResNet, respectively. The performances of VGGNet, GoogLeNet, and ResNet were not significantly different compared to the radiologist. In addition, GoogLeNet and ResNet demonstrated non-inferiority compared to the radiologist. CONCLUSIONS: Deep learning technique may be a useful supportive tool in diagnosing otosclerosis on temporal bone CT. KEY POINTS: • Deep learning can be a helpful tool for the diagnosis of otosclerosis on temporal bone CT. • Deep learning analyses with GoogLeNet and ResNet demonstrate non-inferiority when compared to the subspecialty trained radiologist. • Deep learning may be particularly useful in medical institutions without experienced radiologists.

Reinterpretation of magnetic resonance imaging findings with magnetoencephalography can improve the accuracy of detecting epileptogenic cortical lesions.

OBJECTIVE: This study examined whether the application of magnetoencephalography (MEG) to interpret magnetic resonance imaging (MRI) findings can aid the diagnosis of intractable epilepsy caused by organic brain lesions. METHODS: This study included 51 patients with epilepsy who had MEG clusters but whose initial MRI findings were interpreted as being negative for organic lesions. Three board-certified radiologists reinterpreted the MRI findings, utilizing the MEG findings as a guide. The degree to which the reinterpretation of the imaging results identified an organic lesion was rated on a 5-point scale. RESULTS: Reinterpretation of the MRI data with MEG guidance helped detect an abnormality by at least one radiologist in 18 of the 51 patients (35.2%) with symptomatic localization-related epilepsy. A surgery was performed in 7 of the 51 patients, and histopathological analysis results identified focal cortical dysplasia in 5 patients (Ia: 1, IIa: 2, unknown: 2), hippocampal sclerosis in 1 patient, and dysplastic neurons/gliosis in 1 patient. CONCLUSIONS: The results of this study highlight the potential diagnostic applications of MEG to detect organic epileptogenic lesions, particularly when radiological visualization is difficult with MRI alone.

Cystic cervical lymph nodes of papillary thyroid carcinoma, tuberculosis and human papillomavirus positive oropharyngeal squamous cell carcinoma: utility of deep learning in their differentiation on CT.

OBJECTIVES: Cervical lymph nodes with internal cystic changes are seen with several pathologies, including papillary thyroid carcinoma (PTC), tuberculosis (TB), and HPV-positive oropharyngeal squamous cell carcinoma (HPV+OPSCC). Differentiating these lymph nodes is difficult in the absence of a known primary tumor or reliable medical history. In this study, we assessed the utility of deep learning in differentiating the pathologic lymph nodes of PTC, TB, and HPV+OPSCC on CT. METHODS: A total of 173 lymph nodes (55 PTC, 58 TB, and 60 HPV+OPSCC) were selected based on pathology records and suspicious morphological features. These lymph nodes were divided into the training set (n = 131) and the test set (n = 42). In deep learning analysis, JPEG lymph node images were extracted from the CT slice that included the largest area of each node and fed into a deep learning training session to create a diagnostic model. Transfer learning was used with the deep learning model architecture of ResNet-101. Using the test set, the diagnostic performance of the deep learning model was compared against the histopathological diagnosis and to the diagnostic performances of two board-certified neuroradiologists. RESULTS: Diagnostic accuracy of the deep learning model was 0.76 (=32/42), whereas those of Radiologist 1 and Radiologist 2 were 0.48 (=20/42) and 0.41 (=17/42), respectively. Deep learning derived diagnostic accuracy was significantly higher than both of the two neuroradiologists (P < 0.01, respectively). CONCLUSION: Deep learning algorithm holds promise to become a useful diagnostic support tool in interpreting cervical lymphadenopathy.

Potential benefits of adaptive intensity-modulated proton therapy in nasopharyngeal carcinomas.

PURPOSE: To investigate potential advantages of adaptive intensity-modulated proton beam therapy (A-IMPT) by comparing it to adaptive intensity-modulated X-ray therapy (A-IMXT) for nasopharyngeal carcinomas (NPC). METHODS: Ten patients with NPC treated with A-IMXT (step and shoot approach) and concomitant chemotherapy between 2014 and 2016 were selected. In the actual treatment, 46 Gy in 23 fractions (46Gy/23Fx.) was prescribed using the initial plan and 24Gy/12Fx was prescribed using an adapted plan thereafter. New treatment planning of A-IMPT was made for the same patients using equivalent dose fractionation schedule and dose constraints. The dose volume statistics based on deformable images and dose accumulation was used in the comparison of A-IMXT with A-IMPT. RESULTS: The means of the Dmean of the right parotid gland (P < 0.001), right TM joint (P < 0.001), left TM joint (P < 0.001), oral cavity (P < 0.001), supraglottic larynx (P = 0.001), glottic larynx (P < 0.001), , middle PCM (P = 0.0371), interior PCM (P < 0.001), cricopharyngeal muscle (P = 0.03643), and thyroid gland (P = 0.00216), in A-IMPT are lower than those of A-IMXT, with statistical significance. The means of, D0.03cc , and Dmean of each sub portion of auditory apparatus and D30% for Eustachian tube and D0.5cc for mastoid volume in A-IMPT are significantly lower than those of A-IMXT. The mean doses to the oral cavity, supraglottic larynx, and glottic larynx were all reduced by more than 20 Gy (RBE = 1.1). CONCLUSIONS: An adaptive approach is suggested to enhance the potential benefit of IMPT compared to IMXT to reduce adverse effects for patients with NPC.

Microstructural Alterations in Bipolar and Major Depressive Disorders: A Diffusion Kurtosis Imaging Study.

BACKGROUND: Identifying structural and functional abnormalities in bipolar (BD) and major depressive disorders (MDD) is important for understanding biological processes. HYPOTHESIS: Diffusion kurtosis imaging (DKI) may be able to detect the brain's microstructural alterations in BD and MDD and any differences between the two. STUDY TYPE: Prospective. SUBJECTS: In all, 16 BD patients, 19 MDD patients, and 20 age- and gender-matched healthy volunteers. FIELD STRENGTH/SEQUENCE: DKI at 3.0T. ASSESSMENT: The major DKI indices of the brain were compared voxel-by-voxel among the three groups. Significantly different voxels were tested for correlation with clinical variables (ie, Young Mania Rating Scale [YMRS], 17-item Hamilton Depression Rating Scale [17-HDRS], Montgomery-Åsberg Depression Rating Scale, total disease duration, duration of current episode, and the number of past manic/depressive episodes). The performance of the DKI indices in identifying microstructural alterations was estimated. STATISTICAL TESTS: One-way analysis of variance (ANOVA) was used for group comparison of DKI indices. The performance of these indices in detecting microstructural alterations was determined by receiver operating characteristic (ROC) analysis. Pearson's product-moment correlation analyses were used to test the correlations of these indices with clinical variables. RESULTS: DKI revealed widespread microstructural alterations across the brain in each disorder (P < 0.05). Some were significantly different between the two disorders. Mean kurtosis (MK) in the gray matter of the right inferior parietal lobe was able to distinguish BD and MDD with an accuracy of 0.906. A strong correlation was revealed between MK in that region and YMRS in BD patients (r = -0.641, corrected P = 0.042) or 17-HDRS in MDD patients (r = -0.613, corrected P = 0.030). There were also strong correlations between a few other DKI indices and disease duration (r = -0.676 or 0.626, corrected P < 0.05). DATA CONCLUSION: DKI detected microstructural brain alterations in BD and MDD. Its indices may be useful to distinguish the two disorders or to reflect disease severity and duration. LEVEL OF EVIDENCE: 2 TECHNICAL EFFICACY STAGE: 3 J. Magn. Reson. Imaging 2020;52:1187-1196.

Deep learning analysis using FDG-PET to predict treatment outcome in patients with oral cavity squamous cell carcinoma.

OBJECTIVE: To assess the utility of deep learning analysis using 18F-fluorodeoxyglucose (FDG) uptake by positron emission tomography (PET/CT) to predict disease-free survival (DFS) in patients with oral cavity squamous cell carcinoma (OCSCC). METHODS: One hundred thirteen patients with OCSCC who received pretreatment FDG-PET/CT were included. They were divided into training (83 patients) and test (30 patients) sets. The diagnosis of treatment control/failure and the DFS rate were obtained from patients' medical records. In deep learning analyses, three planes of axial, coronal, and sagittal FDG-PET images were assessed by ResNet-101 architecture. In the training set, image analysis was performed for the diagnostic model creation. The test data set was subsequently analyzed for confirmation of diagnostic accuracy. T-stage, clinical stage, and conventional FDG-PET parameters (the maximum and mean standardized uptake value (SUVmax and SUVmean), heterogeneity index, metabolic tumor volume (MTV), and total lesion glycolysis (TLG) were also assessed with determining the optimal cutoff from training dataset and then validated their diagnostic ability from test dataset. RESULTS: In dividing into patients with treatment control and failure, the highest diagnostic accuracy of 0.8 was obtained using deep learning classification, with a sensitivity of 0.8, specificity of 0.8, positive predictive value of 0.89, and negative predictive value of 0.67. In the Kaplan-Meier analysis, the DFS rate was significantly different only with the analysis of deep learning-based classification (p < .01). CONCLUSIONS: Deep learning-based diagnosis with FDG-PET images may predict treatment outcome in patients with OCSCC. KEY POINTS: • Deep learning-based diagnosis of FDG-PET images showed the highest diagnostic accuracy to predict the treatment outcome in patients with oral cavity squamous cell carcinoma. • Deep learning-based diagnosis was shown to differentiate patients between good and poor disease-free survival more clearly than conventional T-stage, clinical stage, and conventional FDG-PET-based parameters.

Optimization of the spatial modulation function of vessel-encoded pseudo-continuous arterial spin labeling and its application to dynamic angiography.

PURPOSE: In vessel-encoded pseudo-continuous arterial spin labeling (ve-pCASL), vessel-selective labeling is achieved by modulation of the inversion efficiency across space. However, the spatial transition between the labeling and control conditions is rather gradual, which can cause partial labeling of vessels, reducing SNR-efficiency and necessitating complex postprocessing to decode the vessel-selective signals. The purpose of this study is to optimize the pCASL labeling parameters to obtain a sharper spatial inversion profile of the labeling and thereby minimizing the risk of partial labeling of untargeted arteries. METHODS: Bloch simulations were performed to investigate how the inversion profile was influenced by the pCASL labeling parameters: the maximum (Gmax ) and mean (Gmean ) labeling gradient were varied for ve-pCASL with unipolar and bipolar gradients. The findings in the simulation study were subsequently confirmed in an in vivo volunteer study. Moreover, conventional and optimized settings were compared for 4D-MRA using four-cycle Hadamard ve-pCASL; the visualization of arteries and the presence of the partial labeling were assessed by an expert observer. RESULTS: When using unipolar gradient, lower Gmean resulted in a steeper spatial transition, whereas the width of the control region was broader for higher Gmax . The in vivo study confirmed these findings. When using bipolar gradients, the control region was always very narrow. Qualitative comparison of the 4D-MRA demonstrated lower occurrence of partial labeling when using the optimized gradient parameters. CONCLUSION: The shape of the ve-pCASL inversion profile can be optimized by changing Gmean and Gmax to reduce partial labeling of untargeted arteries.

Acceleration of vessel-selective dynamic MR Angiography by pseudocontinuous arterial spin labeling in combination with Acquisition of ConTRol and labEled images in the Same Shot (ACTRESS).

PURPOSE: The recently introduced "Acquisition of ConTRol and labEled imaging in the Same Shot" (ACTRESS) approach was designed to halve the scan time of arterial spin labeling (ASL) -based 4D-MRA by obtaining both labeled and control images in a single Look-Locker readout. However, application for vessel-selective labeling remains difficult. The aim of this study was to achieve a combination of ACTRESS and vessel-selective labeling to halve the scan time of vessel-selective 4D-MRA. METHODS: By Bloch equation simulations, Look-Locker pseudocontinuous-ASL (pCASL) was optimized to achieve constant static tissue signal across the multidelay readout, which is essential for the ACTRESS approach. Additionally, a new subtraction scheme was proposed to achieve visualization of the inflow phase even when labeled blood will have already arrived in the distal arteries during the first phase acquisition due to the long duration of the pCASL labeling module. In vivo studies were performed to investigate the signal variation of the static tissue, as well as to assess image quality of vessel-selective 4D-MRA with ACTRESS. RESULTS: In in vivo studies, the mean signal variation of the static tissue was 8.98% over the Look-Locker phases, thereby minimizing the elevation of background signal. This allowed visualization of peripheral arteries and slowly arriving arterial blood with image quality as good as conventional pCASL within half the acquisition time. Vessel-selective pCASL-ACTRESS enabled the separated visualization of vessels arising from internal and external carotid arteries within this shortened acquisition time. CONCLUSION: By combining vessel-selective pCASL and ACTRESS approach, 4D-MRA of a single targeted arterial tree was achieved in a few minutes.

Estimation of the number of feeding arteries of spinal arteriovenous malformations by using three-dimensional digital subtraction angiography.

PURPOSE: Spinal angiography is the gold standard for evaluation or diagnosis of spinal arteriovenous malformations (AVMs). However, some feeding arteries might be overlooked when multiple feeders exist. This study aimed to retrospectively review cases of spinal intra-dural AVMs, which were identified by three-dimensional digital subtraction angiography (3D-DSA), and attempted to estimate the number of feeding arteries. METHODS: We retrospectively reviewed patients with spinal intra-dural AVMs who underwent 3D-DSA at Hokkaido University Hospital from January 2005 to December 2016. We selected 9 patients in whom we could obtain data of multi-planar reconstruction of 3D-DSA. We measured the computed tomography (CT) values of feeding arteries and draining veins. The CT values represented the averages of maximum CT values of 5 continuous axial slices. The ratio of the CT value of feeders to that of drainers (F/D ratio) was calculated. The correlation between the F/D ratio and the number of feeders was examined with Pearson's correlation coefficient. RESULTS: The average number of feeders was 2.3 (1-4), and the number of feeders was significantly positively correlated with the F/D ratio (r = 0.855, P = .003). CONCLUSIONS: We conclude that the number of feeding arteries of spinal intra-dural AVMs can be estimated by using the F/D ratio obtained from 3D-DSA. These slides can be retrieved under Electronic Supplementary Material.