Temporal subtraction CT with nonrigid image registration improves detection of bone metastases by radiologists: results of a large-scale observer study.

To determine whether temporal subtraction (TS) CT obtained with non-rigid image registration improves detection of various bone metastases during serial clinical follow-up examinations by numerous radiologists. Six board-certified radiologists retrospectively scrutinized CT images for patients with history of malignancy sequentially. These radiologists selected 50 positive and 50 negative subjects with and without bone metastases, respectively. Furthermore, for each subject, they selected a pair of previous and current CT images satisfying predefined criteria by consensus. Previous images were non-rigidly transformed to match current images and subtracted from current images to automatically generate TS images. Subsequently, 18 radiologists independently interpreted the 100 CT image pairs to identify bone metastases, both without and with TS images, with each interpretation separated from the other by an interval of at least 30 days. Jackknife free-response receiver operating characteristics (JAFROC) analysis was conducted to assess observer performance. Compared with interpretation without TS images, interpretation with TS images was associated with a significantly higher mean figure of merit (0.710 vs. 0.658; JAFROC analysis, P = 0.0027). Mean sensitivity at lesion-based was significantly higher for interpretation with TS compared with that without TS (46.1% vs. 33.9%; P = 0.003). Mean false positive count per subject was also significantly higher for interpretation with TS than for that without TS (0.28 vs. 0.15; P < 0.001). At the subject-based, mean sensitivity was significantly higher for interpretation with TS images than that without TS images (73.2% vs. 65.4%; P = 0.003). There was no significant difference in mean specificity (0.93 vs. 0.95; P = 0.083). TS significantly improved overall performance in the detection of various bone metastases.

Optimization of prediction methods for risk assessment of pathogenic germline variants in the Japanese population.

Predicting pathogenic germline variants (PGVs) in breast cancer patients is important for selecting optimal therapeutics and implementing risk reduction strategies. However, PGV risk factors and the performance of prediction methods in the Japanese population remain unclear. We investigated clinicopathological risk factors using the Tyrer-Cuzick (TC) breast cancer risk evaluation tool to predict BRCA PGVs in unselected Japanese breast cancer patients (n = 1,995). Eleven breast cancer susceptibility genes were analyzed using target-capture sequencing in a previous study; the PGV prevalence in BRCA1, BRCA2, and PALB2 was 0.75%, 3.1%, and 0.45%, respectively. Significant associations were found between the presence of BRCA PGVs and early disease onset, number of familial cancer cases (up to third-degree relatives), triple-negative breast cancer patients under the age of 60, and ovarian cancer history (all P < .0001). In total, 816 patients (40.9%) satisfied the National Comprehensive Cancer Network (NCCN) guidelines for recommending multigene testing. The sensitivity and specificity of the NCCN criteria for discriminating PGV carriers from noncarriers were 71.3% and 60.7%, respectively. The TC model showed good discrimination for predicting BRCA PGVs (area under the curve, 0.75; 95% confidence interval, 0.69-0.81). Furthermore, use of the TC model with an optimized cutoff of TC score ≥0.16% in addition to the NCCN guidelines improved the predictive efficiency for high-risk groups (sensitivity, 77.2%; specificity, 54.8%; about 11 genes). Given the influence of ethnic differences on prediction, we consider that further studies are warranted to elucidate the role of environmental and genetic factors for realizing precise prediction.

Imaging findings of granulocyte colony-stimulating factor-producing tumors: a case series and review of the literature.

Granulocyte colony-stimulating factor (G-CSF)-producing tumors have an aggressive clinical course. Here, we report five cases of G-CSF-producing tumors and review the literature, focusing on imaging findings related to tumor-produced G-CSF. In addition to our cases, we identified 30 previous reports of G-CSF-producing tumors on which 18F-fluorodeoxyglucose positron emission tomography (FDG-PET)/CT, bone scintigraphy, or evaluation of bone marrow MR findings was performed. White blood cell count, serum C-reactive protein, and serum interleukin-6 were elevated in all cases for which these parameters were measured. G-CSF-producing tumors presented large necrotic masses (mean diameter 83.2 mm, range 17-195 mm) with marked FDG uptake (mean maximum standardized uptake value: 20.09). Diffuse FDG uptake into the bone marrow was shown in 28 of the 31 cases in which FDG-PET/CT was performed. The signal intensity of bone marrow suggested marrow reconversion in all seven MRI-assessable cases. Bone scintigraphy demonstrated no significant uptake, except in two cases with bone metastases. Splenic FDG uptake was increased in 8 of 10 cases in which it was evaluated. These imaging findings may reflect the effects of tumor-produced G-CSF. The presence of G-CSF-producing tumors should be considered in patients with cancer who show these imaging findings and marked inflammatory features of unknown origin.

Diagnostic performance of 68Ga-DOTATOC PET/CT in tumor-induced osteomalacia.

OBJECTIVE: Tumor-induced osteomalacia (TIO) is caused by typically small tumors that secrete fibroblast growth factor 23 (FGF23). As tumor resection is the only effective treatment for TIO, it is important to detect the culprit tumor. We aimed to assess the utility of 68Gallium-DOTA-D-Phe(1)-Tyr(3)-octreotide (68Ga-DOTATOC) PET/CT in TIO and the correlation between biochemical parameters and the PET/CT results. METHODS: Thirty-five patients with clinically suspected TIO who had undergone 68Ga-DOTATOC PET/CT were retrospectively analyzed. 68Ga-DOTATOC PET/CT results were compared with biochemical parameters and the final diagnosis, including histopathology. RESULTS: 68Ga-DOTATOC PET/CT detected focal uptake consistent with TIO in 21/35 patients, one of which was considered false positive. In 16 patients, the cause of osteomalacia was confirmed histologically as phosphaturic mesenchymal tumor (n = 15) or fibrous dysplasia (n = 1). The other four patients were judged clinically as true positive by subsequent MRI and the clinical course. Overall, the detection rate of 68Ga-DOTATOC PET/CT was 57% (20/35). Median tumor maximum standardized uptake value (SUVmax) was 6.9 (range 1.5-37.7). There was no significant difference in serum intact FGF23 level between DOTATOC-positive and DOTATOC-negative cases, and no significant correlation was observed between intact FGF23 level and tumor SUVmax. CONCLUSIONS: 68Ga-DOTATOC PET/CT was clinically useful in detecting culprit tumors and subsequent patient management in TIO.

Diffusion-weighted imaging of uterine adenomyosis: Correlation with clinical backgrounds and comparison with malignant uterine tumors.

AIM: To elucidate correlation between signal intensity on diffusion-weighted images (SI-DWI) and clinical backgrounds for uterine adenomyosis and to compare SI-DWI of adenomyosis and malignant uterine tumors. METHODS: This study examined 46 adenomyosis patients diagnosed using magnetic resonance imaging and 25 patients with surgically confirmed malignant uterine myometrial tumor. First, adenomyosis cases were classified visually into high-intensity and low-intensity groups based on the SI-DWI compared with that of normal uterine myometrium. Secondly, correlation was assessed between SI-DWI of adenomyosis and patient clinical background information such as age, menopausal status, menstrual cycle and dysmenorrhea severity. Third, quantitative comparison was made of low-intensity adenomyosis (LIA), high-intensity adenomyosis (HIA) and malignant tumor groups for the signal intensity ratio (SIR) on DWI and the apparent diffusion coefficient (ADC). Their diagnostic performance was evaluated using logistic regression analysis and receiver operating characteristic (ROC) analysis. RESULTS: The 46 adenomyosis cases were classified as 26 low-intensity and 20 high-intensity cases. Significant correlation was found only for menstrual cycle phases. HIA had significantly lower SIR and higher ADC than malignant tumor. The ADC of HIA was significantly higher than that of LIA. The combination of SIR and ADC showed excellent diagnostic performance (area under ROC curve, 0.99). CONCLUSION: There is a variation in signal intensity on DWI of uterine adenomyosis and it is associated with menstrual cycle phase. Adenomyosis with high signal intensity on DWI can be differentiated from malignant lesions by its lower signal intensity on DWI and higher ADC than that found for malignant uterine tumors, however overlaps exist.

Utility of real-time image fusion technology in ultrasonography-guided fiducial marker implantation for stereotactic body radiation therapy for liver tumors.

BACKGROUND: Ultrasonography (US) is useful when implanting fiducial markers in the liver. However, the implant position is sometimes lost. Recently, real-time image fusion technology (Volume Navigation [V-navi]; GE Healthcare, Milwaukee, WI, USA) has been introduced as a technique for using images from different modalities, and its utility for fiducial marker implantation has been hypothesized. PURPOSE: To evaluate the utility of US-guided fiducial marker implantation in the liver using V-navi compared to conventional US. MATERIAL AND METHODS: We retrospectively reviewed 35 patients who underwent fiducial marker implantation for stereotactic body radiation therapy of liver tumors in 2013-2018. To avoid artifacts obscuring the tumor, the target point of implantation was set 10 mm cranial or caudal to the tumor. Marker implantation was then performed using US alone (US group, n = 24) or V-navi with computed tomography (CT) or magnetic resonance imaging (V-navi group, n = 11). Postprocedural CT was evaluated to determine technical success, distances between marker and either tumor surface or target point, and whether marker-induced artifacts obscured the tumor. Complications were also evaluated. Results were compared between groups. RESULTS: Technical success was obtained in 33 patients. Distance between the tumor and marker showed no significant difference between groups. Distance between target point and marker was shorter in the V-navi group (P = 0.0093). Tumor-obscuring artifacts were seen in 12 patients (V-navi group, n = 1; US group, n = 11; P = 0.055). The only complication was minor bleeding in the US group (n = 1). CONCLUSION: V-navi appears useful for US-guided fiducial marker implantation in the liver compared with conventional US.

Microcystic, Elongated and Fragmented Pattern Invasion Can Adversely Influence Preoperative Staging for Low-grade Endometrial Carcinoma.

PURPOSE: To investigate the influence of microcystic, elongated and fragmented (MELF) pattern invasion on preoperative evaluation of lymph node (LN) metastasis and myometrial invasion in patients with low-grade endometrial carcinoma. METHODS: The study included 192 consecutive patients with low-grade endometrial carcinoma who underwent preoperative computed tomography (CT) and magnetic resonance imaging (MRI), followed by surgery. One hundred sixty one of 192 patients underwent LN dissection and were analyzed for LN metastasis. All patients were analyzed for myometrial invasion. Presence of enlarged LN was evaluated by using size criteria on CT. Depth of myometrial invasion was evaluated on MRI using T2-weighted imaging, diffusion-weighted imaging and contrast-enhanced T1-weighted imaging comprehensively. Sensitivity and specificity for LN metastasis and deep myometrial invasion were evaluated for MELF group and non-MELF group. The difference of sensitivity between two groups was compared using Chi-square and Fisher's exact test. RESULTS: MELF pattern invasion was identified in 43/192 patients (22%). LN metastases were observed in 18/39 patients in MELF group and 6/122 patients in non-MELF group for pelvic LN and 11/29 patients in MELF group and 4/57 patients in non-MELF group for para-aortic LN. Sensitivity for the detection of pelvic LN metastasis in MELF group was significantly lower than in non-MELF group (16.7% vs 66.7%). As for the assessment of the deep myometiral invasion, pathological deep myometrial invasion were found in 31/43 patients in MELF group and 32/149 patients in non-MELF group. Sensitivity in MELF group showed lower values than in non-MELF group (54.8% vs 78.1% for reader 1, 54.8% vs 62.5% for reader 2), although there was no statistically significant difference (P = 0.09 for reader 1 and P = 0.72 for reader 2). CONCLUSION: In case of low-grade endometrial carcinoma with MELF pattern invasion, preoperative staging by CT and MRI have a risk for underestimation.

Evaluation of Malignant Breast Lesions Using High-resolution Readout-segmented Diffusion-weighted Echo-planar Imaging: Comparison with Pathology.

PURPOSE: We aimed to investigate the performance of high resolution-diffusion-weighted imaging (HR-DWI) using readout-segmented echo-planar imaging in visualizing malignant breast lesions and evaluating their extent, using pathology as a reference. METHODS: This retrospective study included patients who underwent HR-DWI with surgically confirmed malignant breast lesions. Two radiologists blinded to the final diagnosis evaluated HR-DWI independently and identified the lesions, measuring their maximum diameters. Another radiologist confirmed if those lesions were identical to the pathology. The maximum diameters of the lesions between HR-DWI and pathology were compared, and their correlations were calculated using Spearman's correlation coefficient. Apparent diffusion coefficient (ADC) values of the lesions were measured. RESULTS: Ninety-five mass/64 non-mass lesions were pathologically confirmed in 104 females. Both radiologists detected the same 93 mass lesions (97.9%). Spearman's correlation coefficient for mass lesions were 0.89 and 0.90 (P < 0.0001 and 0001) for the two radiologists, respectively. The size differences within 10 mm were 90.3% (84/93) and 94.6% (88/93) respectively. One radiologist detected 35 non-mass lesions (54.7%) and another radiologist detected 32 non-mass lesions (50.0%), of which 28 lesions were confirmed as identical. Spearman's correlation coefficient for non-mass lesions were 0.59 and 0.22 (P = 0.0002 and 0.22), respectively. The mean ADC value of mass lesions and non-mass lesions were 0.80 and 0.89 × 10-3 mm2/s, respectively. CONCLUSION: Using HR-DWI, malignant mass lesions were depicted with excellent agreement with the pathological evaluation. Approximately half of the non-mass lesions could not be identified, suggesting a current limitation of HR-DWI.

Qualitative and Quantitative Assessment of Nonlocal Means Reconstruction Algorithm in a Flexible PET Scanner.

OBJECTIVE. Flexible PET (fxPET) was designed to fit existing MRI systems. The newly modified nonlocal means (NLM) algorithm is combined with the 3D dynamic row-action maximum likelihood algorithm (DRAMA). We investigated qualitative and quantitative acceptability of fxPET images reconstructed by modified NLM compared with whole-body (WB) PET/CT images and conventional 3D DRAMA reconstruction alone. MATERIALS AND METHODS. Fifty-nine patients with known or suspected malignancies underwent WB PET/CT scanning approximately 1 hour after the injection of 18F-FDG, after which they underwent fxPET scanning. Two readers rated the quality of fxPET images by consensus. Detection rate (the proportion of lesions found on PET), maximal standardized uptake value (SUVmax), metabolic tumor volume (MTV), total lesion glycolysis (TLG), tumor-to-normal liver ratio (TNR), and background liver signal-to-noise ratio (SNR) were compared among the three datasets. RESULTS. Higher image quality was obtained by modified NLM reconstruction than by conventional reconstruction without statistical significance. The detection rate was comparable among three datasets. SUVmax was significantly higher, and MTV and TLG were significantly lower in the modified NLM dataset (p < 0.002) than in the other two datasets, with significantly positive correlations (p < 0.001; Spearman rank correlation coefficient, 0.87-0.99). The TNRs in modified NLM images were significantly larger than in the other datasets (p < 0.05). The background SNRs in modified NLM images were comparable with those in WB PET/CT images, and significantly higher than in the conventional fxPET images (p < 0.005). CONCLUSION. The modified NLM algorithm was clinically acceptable, yielding higher TNR and background SNR compared with conventional reconstruction. Image quality and the lesion detection rate were comparable in this population.

Adaptive Voxel Matching for Temporal CT Subtraction.

Temporal subtraction (TS) technique calculates a subtraction image between a pair of registered images acquired from the same patient at different times. Previous studies have shown that TS is effective for visualizing pathological changes over time; therefore, TS should be a useful tool for radiologists. However, artifacts caused by partial volume effects degrade the quality of thick-slice subtraction images, even with accurate image registration. Here, we propose a subtraction method for reducing artifacts in thick-slice images and discuss its implementation in high-speed processing. The proposed method is based on voxel matching, which reduces artifacts by considering gaps in discretized positions of two images in subtraction calculations. There are two different features between the proposed method and conventional voxel matching: (1) the size of a searching region to reduce artifacts is determined based on discretized position gaps between images and (2) the searching region is set on both images for symmetrical subtraction. The proposed method is implemented by adopting an accelerated subtraction calculation method that exploit the nature of liner interpolation for calculating the signal value at a point among discretized positions. We quantitatively evaluated the proposed method using synthetic data and qualitatively using clinical data interpreted by radiologists. The evaluation showed that the proposed method was superior to conventional methods. Moreover, the processing speed using the proposed method was almost unchanged from that of the conventional methods. The results indicate that the proposed method can improve the quality of subtraction images acquired from thick-slice images.

Distinguishing intrahepatic mass-forming biliary carcinomas from hepatocellular carcinoma by computed tomography and magnetic resonance imaging using the Bayesian method: a bi-center study.

OBJECTIVES: To determine imaging hallmarks for distinguishing intrahepatic mass-forming biliary carcinomas (IMBCs) from hepatocellular carcinoma (HCC) and to validate their diagnostic ability using Bayesian statistics. METHODS: Study 1 retrospectively identified clinical and imaging hallmarks that distinguish IMBCs (n = 41) from HCC (n = 247) using computed tomography (CT) and magnetic resonance imaging (MRI). Study 2 retrospectively assessed the diagnostic ability of these hallmarks to distinguish IMBCs (n = 37) from HCC (n = 111) using Bayesian statistics with images obtained from a different institution. We also assessed the diagnostic ability of the hallmarks in the patient subgroup with high diagnostic confidence (≥ 80% of post-test probability). Two radiologists independently evaluated the imaging findings in studies 1 and 2. RESULTS: In study 1, arterial phase peritumoral parenchymal enhancement on CT/MRI, delayed enhancement on CT/MRI, diffusion-weighted imaging peripheral hyperintensity, and bile duct dilatation were hallmarks indicating IMBCs, whereas chronic liver disease, non-rim arterial phase hyperenhancement on CT/MRI, enhancing capsule on CT/MRI, and opposed-phase signal drop were hallmarks indicating HCC (p = 0.001-0.04). In study 2, Bayesian statistics-based post-test probability combining all hallmark features had a diagnostic accuracy of 89.2% (132/148) in distinguishing IMBCs from HCC for both readers. In the high diagnostic confidence subgroup (n = 120 and n = 124 for readers 1 and 2, respectively), the accuracy improved (95.0% (114/120) and 93.5% (116/124) for readers 1 and 2, respectively). CONCLUSIONS: Combined interpretation of CT and MRI to identify hallmark features is useful in discriminating IMBCs from HCCs. High post-test probability by Bayesian statistics allows for a more reliable non-invasive diagnosis. KEY POINTS: • Combined interpretation of CT and MRI to identify hallmark features was useful in discriminating intrahepatic mass-forming biliary carcinomas from hepatocellular carcinoma. • Bayesian method-based post-test probability combining all hallmark features determined in study 1 showed high (> 90%) sensitivity and specificity for distinguishing intrahepatic mass-forming biliary carcinomas from hepatocellular carcinoma. • If the post-test probability or the confidence was ≥ 80% when combining the imaging features of CT and MRI, the high specificity of > 95% was achieved without any loss of sensitivity to distinguish hepatocellular carcinoma from intrahepatic mass-forming biliary carcinomas.

Background parenchymal enhancement and its effect on lesion detectability in ultrafast dynamic contrast-enhanced MRI.

PURPOSE: Background parenchymal enhancement (BPE) often affects interpretation of dynamic contrast-enhanced (DCE) MRI. There is limited evidence that reduced BPE is a feature of ultrafast DCE (UF-DCE) MRI. We aimed to evaluate the effect of BPE levels on lesion detectability on UF-DCE MRI in comparison with conventional DCE MRI. METHOD: MRIs of 70 patients with histologically proven breast lesions were retrospectively evaluated. The total number of analyzed lesions was 84 (56 malignant and 28 benign). Using 3 T MRI, 20 phases of UF-DCE MRI based on the three-dimensional gradient-echo VIBE sequence combined with a compressed sensing reconstruction were acquired followed by conventional DCE MRI. Three maximum intensity projection (MIP) images were generated from the 12th phase, the 20th phase of UF-DCE MRI and the initial phase of conventional DCE MRI. Two radiologists independently evaluated the degree of BPE and lesion detectability of the three MIP images for each breast with histologically confirmed lesions. The degree of BPE was scored on a four-point scale and lesion detectability (conspicuity and confidence levels) was scored on a three-point scale. Data were analyzed using the Wilcoxon signed-rank test with Bonferroni correction. RESULTS: BPE was lower on UF-DCE MRI than on conventional DCE MRI. Lesion detectability was higher on UF-DCE MRI among patients with higher BPE on conventional DCE MRI or premenopausal women. CONCLUSIONS: Images with lower BPE can be achieved using UF-DCE MRI and may be advantageous when assessing breast lesions among patients with higher BPE or premenopausal women.

The comparison of high-resolution diffusion weighted imaging (DWI) with high-resolution contrast-enhanced MRI in the evaluation of breast cancers.

PURPOSE: We sought to investigate the performance of high resolution (HR) diffusion-weighted imaging (DWI) using readout-segmented echo-planar imaging (rs-EPI), compared with high-resolution contrast-enhanced MRI (HR CE-MRI) in terms of morphological accuracy, on the basis of the Breast Imaging and Reporting and Data System (BI-RADS) MRI descriptors and lesion size. METHODS: This retrospective study included the image data of 94 patients with surgically confirmed malignant breast lesions who had undergone high resolution diffusion-weighted imaging (HR-DWI) and HR CE-MRI. Two radiologists blinded to the final diagnosis independently identified the lesions on HR-DWI, described the morphology of the lesions according to BI-RADS descriptors, and measured lesion size. HR CE-MRI was subsequently evaluated using the same procedure. The inter-method agreement of the morphology was assessed using kappa statistics. Correlation on size was also assessed. RESULTS: Reader A detected 79 mass lesions and 37 non-mass lesions on HR-DWI and HR CE-MRI. Reader B detected 81 mass lesions and 33 non-mass lesions on HR-DWI and HR CE-MRI. Very high agreement (kappa = 0.81-0.89, p < .05) was observed in the shape and margin assessment of mass lesions, where agreement on internal enhancement/signals was moderate to substantial (kappa = 0.43-0.61, p < .05). Disagreement was mostly seen in the evaluation of rim enhancement. High agreement was observed for non-mass lesion distribution (kappa = 0.76-0.84, p < .05), and agreement on internal enhancement/signals was moderate to fair (kappa = 0.34-0.49, p < .05). Agreement among heterogeneous, clumped, and clustered-ring patterns was variable. Size assessment showed very strong correlation both in mass (Spearman's rho = 0.90-0.96, p < .0001) and non-mass lesions (Spearman's rho = 0.86, p < .0001). CONCLUSIONS: The findings in morphology and lesion extent showed high agreement between HR-DWI and HR CE-MRI for malignant breast lesions. These results imply the potential of applying HR-DWI for evaluation of malignant breast lesions using BI-RADS MRI.

A multiparametric approach to diagnosing breast lesions using diffusion-weighted imaging and ultrafast dynamic contrast-enhanced MRI.

PURPOSE: To evaluate the diagnostic performance of a multiparametric approach to breast lesions including apparent diffusion coefficient (ADC) from diffusion-weighted images (DWI), maximum slope (MS) from ultrafast dynamic contrast enhanced (UF-DCE) MRI, lesion size, and patient's age. MATERIALS AND METHODS: In total, 96 lesions (73 malignant, 23 benign) were evaluated. UF-DCE MRI was acquired using a prototype 3D-gradient-echo volumetric interpolated breath-hold examination (VIBE) with compressed sensing. Images were obtained up to 1 min after gadolinium injection. MS was calculated as the percentage relative enhancement/s. An ADC map was automatically generated from DWI at b = 0 and b = 1000 s/mm2. MS and ADC values were measured by two radiologists independently. Interrater agreement was evaluated using intraclass correlation coefficients. Univariate and multivariate logistic regression analyses were performed using MS, ADC, lesion size, and the patient's age. The parameters of the prediction model were generated from the results of the multivariate logistic regression analysis. Area under the curve (AUC) was used to compare diagnostic performance of the prediction model and each parameter. RESULTS: Interrater agreements on MS and ADC were excellent (ICC 0.99 and 0.88, respectively). MS, ADC, and patient's age remained as significant parameters after univariate and multivariate logistic regression analysis. The prediction model using these significant parameters yielded an AUC of 0.90, significantly higher than that of MS (AUC 0.74, p = 0.01). The AUCs of ADC, MS, patient's age were 0.87, 0.74 and 0.73, respectively. CONCLUSIONS: A multiparametric model using ADC from DWI, MS from UF-DCE MRI, and patient's age showed excellent diagnostic performance, with greater contribution of ADC. Combining DWI and UF-DCE MRI might reduce scanning time while preserving diagnostic performance.

Bone segmentation on whole-body CT using convolutional neural network with novel data augmentation techniques.

BACKGROUND: The purpose of this study was to develop and evaluate an algorithm for bone segmentation on whole-body CT using a convolutional neural network (CNN). METHODS: Bone segmentation was performed using a network based on U-Net architecture. To evaluate its performance and robustness, we prepared three different datasets: (1) an in-house dataset comprising 16,218 slices of CT images from 32 scans in 16 patients; (2) a secondary dataset comprising 12,529 slices of CT images from 20 scans in 20 patients, which were collected from The Cancer Imaging Archive; and (3) a publicly available labelled dataset comprising 270 slices of CT images from 27 scans in 20 patients. To improve the network's performance and robustness, we evaluated the efficacy of three types of data augmentation technique: conventional method, mixup, and random image cropping and patching (RICAP). RESULTS: The network trained on the in-house dataset achieved a mean Dice coefficient of 0.983 ± 0.005 on cross validation with the in-house dataset, and 0.943 ± 0.007 with the secondary dataset. The network trained on the public dataset achieved a mean Dice coefficient of 0.947 ± 0.013 on 10 randomly generated 15-3-9 splits of the public dataset. These results outperform those reported previously. Regarding augmentation technique, the conventional method, RICAP, and a combination of these were effective. CONCLUSIONS: The CNN-based model achieved accurate bone segmentation on whole-body CT, with generalizability to various scan conditions. Data augmentation techniques enabled construction of an accurate and robust model even with a small dataset.

Improving breast mass classification by shared data with domain transformation using a generative adversarial network.

Training of a convolutional neural network (CNN) generally requires a large dataset. However, it is not easy to collect a large medical image dataset. The purpose of this study is to investigate the utility of synthetic images in training CNNs and to demonstrate the applicability of unrelated images by domain transformation. Mammograms showing 202 benign and 212 malignant masses were used for evaluation. To create synthetic data, a cycle generative adversarial network was trained with 599 lung nodules in computed tomography (CT) and 1430 breast masses on digitized mammograms (DDSM). A CNN was trained for classification between benign and malignant masses. The classification performance was compared between the networks trained with the original data, augmented data, synthetic data, DDSM images, and natural images (ImageNet dataset). The results were evaluated in terms of the classification accuracy and the area under the receiver operating characteristic curves (AUC). The classification accuracy improved from 65.7% to 67.1% with data augmentation. The use of an ImageNet pretrained model was useful (79.2%). Performance was slightly improved when synthetic images or the DDSM images only were used for pretraining (67.6 and 72.5%, respectively). When the ImageNet pretrained model was trained with the synthetic images, the classification performance slightly improved (81.4%), although the difference in AUCs was not statistically significant. The use of the synthetic images had an effect similar to the DDSM images. The results of the proposed study indicated that the synthetic data generated from unrelated lesions by domain transformation could be used to increase the training samples.

Hyperestrogenism on 18F-FDG PET/CT in a Patient With Estrogen-Producing Ovarian Clear Cell Carcinoma.

A 78-year-old woman was referred to the gynecologic outpatient department because she was suspected of having ovarian cancer based on an imaging study performed during a general medical examination. Further examination using F-FDG PET/CT revealed a bulky mass lesion with low FDG avidity, as well as surprisingly strong bilateral breast radiotracer uptake despite the patient's age. Thus, an estrogen-producing tumor was suspected. Bilateral salpingo-oophorectomy was performed, and surgical pathology diagnosed the tumor as clear cell carcinoma of the ovary.

[MRI Connecting Functions and Anatomy: A New Bridge for Radiotherapy].

Advances in medical devices have allowed the use of CT, MRI, and PET-CT for the diagnosis of tumors and the detailed evaluation of the extent of lesions. For several decades, CT has been established as the gold standard modality for the treatment planning of radiotherapy, while MRI has emerged as a tool to evaluate the functional characteristics of tumors without radiation exposure. To further optimize precision radiation therapy, we should consider how functional images can be used in the workflow for radiation therapy. In this regard, MRI, as a modality without the need for a contrast agent, may allow more frequent scans and more detailed dose painting, such as increasing the dose to viable lesion parts while reducing the dose to less aggressive parts. Thus, a more personalized treatment based on precision radiation medicine might be realized. In recent years, MR-Linac systems (MRI integrated linear accelerator radiation therapy systems) have been applied in clinical settings by fusing MRI with Linac planning, and further development of radiation therapy utilizing MRI-derived functional images is expected. The use of MR-Linac techniques allows the characteristics of the tumor to be evaluated in more detail before treatment, and the treatment planning can be modified according to the position and characteristics of the tumor (which may change daily during irradiation) to avoid harming normal tissue. Compared with conventional cone beam CT, MR-Linac can offer MR images with much better contrast of soft tissue for image-guided radiation therapy, even when acquired at 0.35 T. A multicenter study of liver tumors using MR-Linac was recently reported. In current tumor imaging, various MRI sequences can be used to evaluate tumor functional information such as tumor heterogeneity, cell density, microenvironment, angiogenesis, necrosis, hypoxic status, and microstructure. In this article, we introduce state-of-the-art acquisition methods for MRI imaging, and discuss how the functional information obtained from these imaging methods can be useful for radiation therapy.

Development of Novel PET Imaging Probes for Detection of Amylin Aggregates in the Pancreas.

The deposition of islet amyloid is associated with ß-cell mass dysfunction in type 2 diabetes mellitus (T2DM). Since the amylin aggregate is the main component of islet amyloid, in vivo imaging of amylin may be useful for diagnosis and elucidation of the pathogenic mechanism of T2DM. In the present study, we newly designed, synthesized, and evaluated two 18F labeled compounds ([18F]DANIR-F 2b and [18F]DANIR-F 2c) as positron emission tomography (PET) probes targeting amylin aggregates. In an in vitro binding study, DANIR-F 2b and DANIR-F 2c showed binding affinity for amylin aggregates (Ki = 160 and 29 nM, respectively). In addition, [18F]DANIR-F 2b and [18F]DANIR-F 2c clearly labeled islet amyloids in in vitro autoradiography of T2DM pancreatic sections. In the biodistribution study using normal mice, [18F]DANIR-F 2b and [18F]DANIR-F 2c displayed favorable pharamacokinetics in the pancreas and some organs located near the pancreas. Furthermore, in an ex vivo autoradiographic study, [18F]DANIR-F 2c also bound to amylin aggregates in the pancreas of the amylin transplanted mice. The results of this study suggest that [18F]DANIR-F 2c shows fundamental properties as a PET imaging probe targeting amylin aggregates in the T2DM pancreas.

Lobar distribution of non-emphysematous gas trapping and lung hyperinflation in chronic obstructive pulmonary disease.

BACKGROUND: Lung hyperinflation in chronic obstructive pulmonary disease (COPD) is closely associated with emphysema and non-emphysematous gas trapping, termed functional small airway disease (fSAD), on inspiratory and expiratory computed tomography (CT). Because the cranial-caudal emphysema distribution affects pulmonary function and fSAD may precede emphysema on CT, we tested the hypothesis that lobar fSAD distribution would affect lung hyperinflation differently in COPD with minimal and established emphysema. METHODS: The volume percentages of fSAD and emphysema (fSAD% and Emph%) over the upper and lower lobes were measured using inspiratory and expiratory CT in 70 subjects with COPD. Subjects were divided into those with minimal and established emphysema (n = 36 and 34) using a threshold of 10% Emph% in the whole lung. RESULTS: In the minimal emphysema group, fSAD% in the upper and lower lobes was positively correlated with functional residual capacity (FRC) and residual volume to total lung capacity ratio (RV/TLC), and the correlation of fSAD% with RV/TLC was greater in the lower lobes. Conversely, in the established emphysema group, fSAD% in the upper and lower lobes was correlated with RV/TLC, but not with FRC. In multivariate analysis, fSAD% in the lower lobes, but not in the upper lobes, was associated with RV/TLC in subjects with minimal emphysema after adjusting for age, smoking status, and bronchodilator use. CONCLUSION: Non-emphysematous gas trapping in the upper and lower lobes has a distinct physiological effect, especially in COPD with minimal emphysema. This local evaluation might allow sensitive detection of changes in lung hyperinflation in COPD.