AI-based Virtual Synthesis of Methionine PET from Contrast-enhanced MRI: Development and External Validation Study.

Background Carbon 11 (11C)-methionine is a useful PET radiotracer for the management of patients with glioma, but radiation exposure and lack of molecular imaging facilities limit its use. Purpose To generate synthetic methionine PET images from contrast-enhanced (CE) MRI through an artificial intelligence (AI)-based image-to-image translation model and to compare its performance for grading and prognosis of gliomas with that of real PET. Materials and Methods An AI-based model to generate synthetic methionine PET images from CE MRI was developed and validated from patients who underwent both methionine PET and CE MRI at a university hospital from January 2007 to December 2018 (institutional data set). Pearson correlation coefficients for the maximum and mean tumor to background ratio (TBRmax and TBRmean, respectively) of methionine uptake and the lesion volume between synthetic and real PET were calculated. Two additional open-source glioma databases of preoperative CE MRI without methionine PET were used as the external test set. Using the TBRs, the area under the receiver operating characteristic curve (AUC) for classifying high-grade and low-grade gliomas and overall survival were evaluated. Results The institutional data set included 362 patients (mean age, 49 years ± 19 [SD]; 195 female, 167 male; training, n = 294; validation, n = 34; test, n = 34). In the internal test set, Pearson correlation coefficients were 0.68 (95% CI: 0.47, 0.81), 0.76 (95% CI: 0.59, 0.86), and 0.92 (95% CI: 0.85, 0.95) for TBRmax, TBRmean, and lesion volume, respectively. The external test set included 344 patients with gliomas (mean age, 53 years ± 15; 192 male, 152 female; high grade, n = 269). The AUC for TBRmax was 0.81 (95% CI: 0.75, 0.86) and the overall survival analysis showed a significant difference between the high (2-year survival rate, 27%) and low (2-year survival rate, 71%; P < .001) TBRmax groups. Conclusion The AI-based model-generated synthetic methionine PET images strongly correlated with real PET images and showed good performance for glioma grading and prognostication. Published under a CC BY 4.0 license. Supplemental material is available for this article.

Deep Learning-based Angiogram Generation Model for Cerebral Angiography without Misregistration Artifacts.

Background Digital subtraction angiography (DSA) generates an image by subtracting a mask image from a dynamic angiogram. However, patient movement-caused misregistration artifacts can result in unclear DSA images that interrupt procedures. Purpose To train and to validate a deep learning (DL)-based model to produce DSA-like cerebral angiograms directly from dynamic angiograms and then quantitatively and visually evaluate these angiograms for clinical usefulness. Materials and Methods A retrospective model development and validation study was conducted on dynamic and DSA image pairs consecutively collected from January 2019 through April 2019. Angiograms showing misregistration were first separated per patient by two radiologists and sorted into the misregistration test data set. Nonmisregistration angiograms were divided into development and external test data sets at a ratio of 8:1 per patient. The development data set was divided into training and validation data sets at ratio of 3:1 per patient. The DL model was created by using the training data set, tuned with the validation data set, and then evaluated quantitatively with the external test data set and visually with the misregistration test data set. Quantitative evaluations used the peak signal-to-noise ratio (PSNR) and the structural similarity (SSIM) with mixed liner models. Visual evaluation was conducted by using a numerical rating scale. Results The training, validation, nonmisregistration test, and misregistration test data sets included 10 751, 2784, 1346, and 711 paired images collected from 40 patients (mean age, 62 years ± 11 [standard deviation]; 33 women). In the quantitative evaluation, DL-generated angiograms showed a mean PSNR value of 40.2 dB ± 4.05 and a mean SSIM value of 0.97 ± 0.02, indicating high coincidence with the paired DSA images. In the visual evaluation, the median ratings of the DL-generated angiograms were similar to or better than those of the original DSA images for all 24 sequences. Conclusion The deep learning-based model provided clinically useful cerebral angiograms free from clinically significant artifacts directly from dynamic angiograms. Published under a CC BY 4.0 license. Supplemental material is available for this article.

Deep Learning for MR Angiography: Automated Detection of Cerebral Aneurysms.

Purpose To develop and evaluate a supportive algorithm using deep learning for detecting cerebral aneurysms at time-of-flight MR angiography to provide a second assessment of images already interpreted by radiologists. Materials and Methods MR images reported by radiologists to contain aneurysms were extracted from four institutions for the period from November 2006 through October 2017. The images were divided into three data sets: training data set, internal test data set, and external test data set. The algorithm was constructed by deep learning with the training data set, and its sensitivity to detect aneurysms in the test data sets was evaluated. To find aneurysms that had been overlooked in the initial reports, two radiologists independently performed a blinded interpretation of aneurysm candidates detected by the algorithm. When there was disagreement, the final diagnosis was made in consensus. The number of newly detected aneurysms was also evaluated. Results The training data set, which provided training and validation data, included 748 aneurysms (mean size, 3.1 mm ± 2.0 [standard deviation]) from 683 examinations; 318 of these examinations were on male patients (mean age, 63 years ± 13) and 365 were on female patients (mean age, 64 years ± 13). Test data were provided by the internal test data set (649 aneurysms [mean size, 4.1 mm ± 3.2] in 521 examinations, including 177 male patients and 344 female patients with mean age of 66 years ± 12 and 67 years ± 13, respectively) and the external test data set (80 aneurysms [mean size, 4.1 mm ± 2.1] in 67 examinations, including 19 male patients and 48 female patients with mean age of 63 years ± 12 and 68 years ± 12, respectively). The sensitivity was 91% (592 of 649) and 93% (74 of 80) for the internal and external test data sets, respectively. The algorithm improved aneurysm detection in the internal and external test data sets by 4.8% (31 of 649) and 13% (10 of 80), respectively, compared with the initial reports. Conclusion A deep learning algorithm detected cerebral aneurysms in radiologic reports with high sensitivity and improved aneurysm detection compared with the initial reports. © RSNA, 2018 See also the editorial by Flanders in this issue.

[Using Super Resolution to Denoise on PET Images].

A high-resolution display panel comes to practical use, but the resolution of the indicated contents does not change. The up-sampling processing is applied to indication of the low-resolution contents. In the up-sampling process, the super resolution enables an up-sampling process which estimates information of high frequency components lost by sampling while analyzing input images is noticed. In this paper, we aimed at reconstructing an image of normal resolution in which the influence of statistical noise is reduced by applying super resolution after down-sampling processing is applied to positron emission tomography (PET) image with many statistical noises. To evaluate the noise reduction effect, we compared it with the Gaussian filter which is frequently used to reduce the influence of the statistical noise of the PET image. A 3D Hoffman brain phantom was used to evaluate objectively by peak signal-to-noise ratio and power spectral density. The objective index of the PET image applying super resolution is positive results, suggesting the possibility of being useful as compared with the conventional method.

Scatter Correction of Septal Penetration for 123I-IMP Cerebral Blood Flow SPECT Adding Radioactivity from the Outside of Field of View-Comparison between Simulation-based and Multi-window Scatter Corrections.

PURPOSE: The N-Isopropyl-p-[123I] Iodoamphetamine (123I-IMP) SPECT imaging reduces the image quality and quantitative accuracy due to scatter and septal penetration occurred by radioactive uptake from outside of the field of view such as the lungs. We evaluated the influence of scatter and septal penetration using phantom-simulated radioactivity from outside of the field of view, and subsequently compared the effect of scatter and septal penetration corrections between the simulation-based effective scatter source estimation (ESSE) method and the multi-window method (ellipse approximation method). METHODS: We used the phantom filled with 10 and 25 kBq/mL for the brain and lung parts corresponding to radioactive concentration in the clinical study. The SPECT images were acquired with and without lung phantom using low-energy high-resolution (LEHR) and cardiac high-resolution (CHR) collimators. We quantitatively evaluated a brain phantom by count analysis and coefficient of variation as reference data without lung phantom simulated the radioactivity from outside of the field of view, and compared between two scatter corrections by each collimator. RESULTS: The brain count in cerebral base with the ESSE method using LEHR collimator was higher than that of the ellipse approximation method. The whole brain count with the ellipse approximation method using CHR collimator shows 28.8% lower than the ESSE method, so that it suggests that the ellipse approximation method for LEHR collimator and the ESSE method for CHR collimator was close to reference counts. The coefficient of variation of the ESSE method was lower than that of the ellipse approximation method for both two collimators. CONCLUSIONS: It was possible to correct the scatter and penetration from outside the field of view with high accuracy, by using the ellipse approximation method with LEHR collimator and the ESSE method with CHR collimator.

Clinical Experience of Dual-phase Cone Beam Computed Tomography during Hepatic Arteriography to Apply 3D-DSA.

We report on the methods and experiences of the dual-phase cone beam computed tomography during hepatic arteriography (CBCTHA) to apply the 3D-DSA. A total of 32 ml contrast medium (150 mgI/ml) was injected at the rate of 2.0 ml/s for 16 s. The early phase scan was initiated 10 s after the start of contrast media injection. The delayed phase scan was started 40 s after that (24 s after the end of CM injection). When using the dual phase CBCTHA, it was able to obtain the classical hepatocellular carcinoma (HCC) images same as computed tomography during hepatic arteriography (CTHA). In the early phase, the tumor can be highly enhanced against the liver parenchyma. In delayed phase, corona enhancement was clearly appeared at the liver parenchyma. Of 58 cases of acquisitions, we experienced six cases with miss breath holding and 14 cases with over the field of view (FOV) due to hepatomegaly. We evaluated the tumor contrast in 18 cases because the other 40 cases were not applied to our criteria. The pixel values of ROIs on the tumor, coronal enhancement, and liver parenchyma were measured, respectively. Then, we calculated tumor-parenchyma contrast (T-P contrast), corona-tumor contrast (C-T contrast), and corona-parenchyma contrast (C-P contrast). The T-P contrast was 358±112, the C-T contrast was 132±51, and the C-P contrast was 168±66. The contrast was clearly visualized among them. The dual-phase CBCTHA that applies the 3D-DSA is a simple and useful technique for hepatocellular carcinoma treatment.

Understanding the Scatter Radiation Distribution during C-arm CT Examination: A Body Phantom Study.

The purpose of this study was to understand the scatter radiation distribution during C-arm CT examination in the interventional radiography (IVR) room to show the escaped area and the radiation protective method. The C-arm rotates 200° in 5 s. The tube voltage was 90 kV, and the entrance dose to the detector was 0.36 µGy/frame during C-arm CT examination. The scattered doses were measured each 50 cm from the isocenter like a grid pattern. The heights of the measurement were 50, 100, and 150 cm from the floor. The maximum scattered doses were 38.23±0.60 µGy at 50 cm, 43.86±0.20 µGy at 100 cm, and 25.78±0.37 µGy at 150 cm. The scatter radiation distribution at 100 cm was the highest scattered dose. The operator should protect their reproductive gland, thyroid, and lens. The scattered dose was low behind the C-arm body and the bed, so they will be able to become the escaped area for staff.

Diagnostic ability of (99m)Tc-HSA-DTPA scintigraphy in combination with SPECT/CT for gastrointestinal bleeding.

PURPOSE: Gastrointestinal (GI) bleeding scintigraphy in combination with single-photon emission computed tomography/computed tomography (SPECT/CT) remains to be studied in detail. This study aimed to examine the diagnostic ability of this tool. METHODS: GI bleeding scintigraphy using (99m)Tc-human serum albumin-diethylenetriaminepentaacetic acid was performed for 38 patients with suspected GI bleeding. Twenty-four patients were diagnosed using planar images alone (planar group) and 14 patients were diagnosed using planar images and additional SPECT/CT images (planar + SPECT/CT group). The diagnostic ability of each method was analyzed. RESULTS: GI bleeding was observed in 20 of the 38 patients. For the existence of GI bleeding, planar images alone showed a sensitivity of 70%, specificity of 93%, positive predictive value (PPV) of 88%, negative predictive value (NPV) of 81%, and an overall accuracy of 83%, whereas planar images + SPECT/CT showed a sensitivity of 100%, specificity of 75%, PPV of 91%, NPV of 100%, and an overall accuracy of 93%. The source of bleeding was accurately diagnosed in 50% in the planar group and 78% in the planar + SPECT/CT group. In the planar + SPECT/CT group, 44% of the evaluable patients showed correct localization of the source of GI bleeding by additional SPECT/CT images, although planar images only showed incorrect localization. CONCLUSION: GI bleeding scintigraphy in combination with SPECT/CT is a noninvasive and useful tool for the examination of GI bleeding.

Effects of canagliflozin on kidney oxygenation evaluated using blood oxygenation level-dependent MRI in patients with type 2 diabetes.

Background: Recent clinical studies suggest protective effects of SGLT2 inhibitors on kidney disease outcome. Chronic hypoxia has a critical role in kidney disease development, thus we speculated that canagliflozin, an SGLT2 inhibitor, can improve kidney oxygenation. Methods: A single-arm study was conducted to investigate the effects of canagliflozin on T2* value, which reflects oxygenation level, in patients with type 2 diabetes (T2D) using repeated blood oxygenation level-dependent MRI (BOLD MRI) examinations. Changes in cortical T2* from before (Day 0) to after single-dose treatment (Day 1) and after five consecutive treatments (Day 5) were evaluated using 12-layer concentric objects (TLCO) and region of interest (ROI) methods. Results: In the full analysis set (n=14 patients), the TLCO method showed no change of T2* with canagliflozin treatment, whereas the ROI method found that cortical T2* was significantly increased on Day 1 but not on Day 5. Sensitivity analysis using TLCO in 13 well-measured patients showed that canagliflozin significantly increased T2* on Day 1 with no change on Day 5, whereas a significant improvement in cortical T2* following canagliflozin treatment was found on both Day 1 and 5 using ROI. Conclusions: Short-term canagliflozin treatment may improve cortical oxygenation and lead to better kidney outcomes in patients with T2D.

[Bilateral filter applied to bone scintigraphy].

The aim of this study is to improve the image quality using a post process rather than a correction process at acquisition time. We used a smoothing filter that is widely used on a compact digital camera. Especially for nuclear medicine, when we use a short acquisition time, we will get images that have a large increase in statistical noise. For those images, we validated the efficiency of the smoothing filter by assessing two characteristic parameters. In addition, we defined the best smoothing filter parameters to get stable images that reduced the influence of statistical noise.

[Development of Dose Management Software Using Visual Basic for Applications and Dose Evaluation in the Field of Nuclear Medicine].

PURPOSE: The purpose of this study was to develop software for smooth dose management based on the Japan diagnostic reference levels (DRLs 2020) in the field of nuclear medicine. METHOD: Using the programming language Visual Basic for Applications (VBA), we implemented a function for calculating actual doses, a function for comparing doses at one's own facility with those of DRLs 2020, a function for calculating appropriate doses for pediatric nuclear medicine examinations, and so on. In addition, we evaluated actual doses before and after the software implementation. RESULT: The software enabled easy calculation of actual doses and comparison with DRLs 2020 for smooth dose management. Furthermore, we were able to use the results of dose evaluation to determine the dosage at our facility and to use them as a reference for optimization. CONCLUSION: In the field of nuclear medicine, it is possible to manage doses in accordance with DRLs 2020 by introducing own software into our clinical practice.

Deep learning-based detection of parathyroid adenoma by 99mTc-MIBI scintigraphy in patients with primary hyperparathyroidism.

OBJECTIVE: It is important to detect parathyroid adenomas by parathyroid scintigraphy with 99m-technetium sestamibi (99mTc-MIBI) before surgery. This study aimed to develop and validate deep learning (DL)-based models to detect parathyroid adenoma in patients with primary hyperparathyroidism, from parathyroid scintigrams with 99mTc-MIBI. METHODS: DL-based models for detecting parathyroid adenoma in early- and late-phase parathyroid scintigrams were, respectively, developed and evaluated. The training dataset used to train the models was collected from 192 patients (165 adenoma cases, mean age: 64 years ± 13, 145 women) and the validation dataset used to tune the models was collected from 45 patients (30 adenoma cases, mean age: 67 years ± 12, 37 women). The images were collected from patients who were pathologically diagnosed with parathyroid adenomas or in whom no lesions could be detected by either parathyroid scintigraphy or ultrasonography at our institution from June 2010 to March 2019. The models were tested on a dataset collected from 44 patients (30 adenoma cases, mean age: 67 years ± 12, 38 women) who took scintigraphy from April 2019 to March 2020. The models' lesion-based sensitivity and mean false positive indications per image (mFPI) were assessed with the test dataset. RESULTS: The sensitivity was 82% [95% confidence interval 72-92%] with mFPI of 0.44 for the scintigrams of the early-phase model and 83% [73-92%] with mFPI of 0.31 for the scintigrams of the delayed-phase model in the test dataset, respectively. CONCLUSIONS: The DL-based models were able to detect parathyroid adenomas with a high sensitivity using parathyroid scintigraphy with 99m-technetium sestamibi.

Visual and quantitative evaluation of microcalcifications in mammograms with deep learning-based super-resolution.

PURPOSE: To evaluate visually and quantitatively the performance of a deep-learning-based super-resolution (SR) model for microcalcifications in digital mammography. METHOD: Mammograms were consecutively collected from 5080 patients who underwent breast cancer screening from January 2015 to March 2017. Of these, 93 patients (136 breasts, mean age, 50 ± 7 years) had microcalcifications in their breasts on mammograms. We applied an artificial intelligence model known as a fast SR convolutional neural network to the mammograms. SR and original mammograms were visually evaluated by four breast radiologists using a 5-point scale (1: original mammograms are strongly preferred, 5: SR mammograms are strongly preferred) for the detection, diagnostic quality, contrast, sharpness, and noise of microcalcifications. Mammograms were quantitatively evaluated using a perception-based image-quality evaluator (PIQE). RESULTS: All radiologists rated the SR mammograms better than the original ones in terms of detection, diagnostic quality, contrast, and sharpness of microcalcifications. These ratings were significantly different according to the Wilcoxon signed-rank test (p <.001), while the noise score of the three radiologists was significantly lower (p <.001). According to PIQE, SR mammograms were rated better than the original mammograms, showing a significant difference by paired t-test (p <.001). CONCLUSION: An SR model based on deep learning can improve the visibility of microcalcifications in mammography and help detect and diagnose them in mammograms.

Differentiation of Brain Metastases and Gliomas Based on Color Map of Phase Difference Enhanced Imaging.

Background and objective: Phase difference enhanced imaging (PADRE), a new phase-related MRI technique, can enhance both paramagnetic and diamagnetic substances, and select which phases to be enhanced. Utilizing these characteristics, we developed color map of PADRE (Color PADRE), which enables simultaneous visualization of myelin-rich structures and veins. Our aim was to determine whether Color PADRE is sufficient to delineate the characteristics of non-gadolinium-enhancing T2-hyperintense regions related with metastatic tumors (MTs), diffuse astrocytomas (DAs) and glioblastomas (GBs), and whether it can contribute to the differentiation of MTs from GBs. Methods: Color PADRE images of 11 patients with MTs, nine with DAs and 17 with GBs were created by combining tissue-enhanced, vessel-enhanced and magnitude images of PADRE, and then retrospectively reviewed. First, predominant visibility of superficial white matter and deep medullary veins within non-gadolinium-enhancing T2-hyperintense regions were compared among the three groups. Then, the discriminatory power to differentiate MTs from GBs was assessed using receiver operating characteristic analysis. Results: The degree of visibility of superficial white matter was significantly better in MTs than in GBs (p = 0.017), better in GBs than in DAs (p = 0.014), and better in MTs than in DAs (p = 0.0021). On the contrary, the difference in the visibility of deep medullary veins was not significant (p = 0.065). The area under the receiver operating characteristic curve to discriminate MTs from GBs was 0.76 with a sensitivity of 80% and specificity of 64%. Conclusion: Visibility of superficial white matter on Color PADRE reflects inferred differences in the proportion of vasogenic edema and tumoral infiltration within non-gadolinium-enhancing T2-hyperintense regions of MTs, DAs and GBs. Evaluation of peritumoral areas on Color PADRE can help to distinguish MTs from GBs.

Usefulness of standardized uptake value normalized by individual CT-based lean body mass in application of PET response criteria in solid tumors (PERCIST).

Our aim in this study was to verify the usefulness of the standardized uptake value (SUV) normalized by individual CT-based lean body mass (LBMCT) in application of PET response criteria in solid tumors (PERCIST).We retrospectively investigated 14 patients (4 male and 10 female) with malignant lymphoma who were undergoing chemotherapy. (18)F-FDG PET/CT examinations were performed before and after chemotherapy. The LBMCT was calculated by estimation of fat weight from CT data (from skull base to pelvis). The mean ± standard deviation (SD) and the Bland-Altman plot were used for comparison among body weight, LBMCT, and LBM derived from a predictive equation (LBMPE). Indices for FDG uptake in the liver were: SUV, SUV based on LBMPE (SULPE), and SUV based on LBMCT (SULCT). Overall differences between the uptake values were analyzed by one-way ANOVA. If the ANOVA showed significance, differences between uptake values were investigated further by use of the Tukey-Kramer test. The mean values of body weight, LBMPE, and LBMCT were: 55.4 ± 14.9 (39.0-112.0), 43.0 ± 10.5 (31.3-75.2), and 35.3 ± 9.8 (23.4-75.8) kg, respectively. There was a wide dispersion between LBMPE and LBMCT (differences, 7.6 ± 3.6 kg; 95 % CI, 6.42-8.85). LBMPE was higher than LBMCT in all the cases except in Case 11. The mean uptake values significantly differed among SUV, SULPE, and SULCT (F = 68.3, p < 0.05). Whereas SULPE deviated from PERCIST criteria in seven patients, SULCT satisfied the criteria except in one case. These results suggest that liver SULCT is useful for application of PERCIST.