Chemical shift-encoded MRI with compressed sensing combined with parallel imaging for proton density fat fraction measurement of the lumbar vertebral bone marrow.

We aimed to investigate the accuracy of proton density fat fraction (PDFF) measurement of the lumbar vertebral bone marrow using chemical shift-encoded magnetic resonance imaging (CSE-MRI) with compressed sensing combined with parallel imaging (CSPI). This study recruited a commercially available phantom, and 43 patients. Fully sampled data without CSPI and under-sampled data with CSPI acceleration factors of 2.4, 3.6, and 4.8 were acquired using a 1.5T imaging system. The relationships between PDFF measurements obtained with the no-CSPI acquisition and those obtained with each CSPI acquisition were assessed using Pearson correlation coefficient (r), linear regression analyses, and Bland-Altman analysis. The intra- and inter-observer variabilities of the PDFF measurements were evaluated using the intraclass correlation coefficient. PDFF measurements obtained with all acquisitions showed a significant correlation and strong agreement with the reference PDFF measurement of the phantom. PDFF measurements obtained using CSE-MRI with and without CSPI were positively correlated (all acquisitions: r = 0.99; P < .001). The mean bias was -0.31% to -0.17% with 95% limits of agreement within ±2.02%. The intra- and inter-observer agreements were excellent (intraclass correlation coefficient: 0.988 and 0.981, respectively). A strong agreement and positive correlation were observed between the PDFF measurements obtained using CSE-MRI with and without CSPI. PDFF measurement of the lumbar vertebral bone marrow using CSE-MRI with CSPI can be acquired with a maximum reduction of approximately 75% in the acquisition time compared with a fully sampled acquisition.

[Segmentation of Mass in Mammogram Using Gaze Search Patterns].

PURPOSE: It is very difficult for a radiologist to correctly detect small lesions and lesions hidden on dense breast tissue on a mammogram. Therefore, recently, computer-aided detection (CAD) systems have been widely used to assist radiologists in interpreting images. Thus, in this study, we aimed to segment mass on the mammogram with high accuracy by using focus images obtained from a eye-tracking device. METHODS: We obtained focus images for two mammography expert radiologists and 19 mammography technologists on 8 abnormal and 8a normal mammograms published by the DDSM. Next, the auto-encoder, Pix2Pix, and UNIT learned the relationship between the actual mammogram and the focus image, and generated the focus image for the unknown mammogram. Finally, we segmented regions of mass on mammogram using the U-Net for each focus image generated by the auto-encoder, Pix2Pix, and UNIT. RESULTS: The dice coefficient in the UNIT was 0.64±0.14. The dice coefficient in the UNIT was higher than that in the auto-encoder and Pix2Pix, and there was a statistically significant difference (p<0.05). The dice coefficient of the proposed method, which combines the focus images generated by the UNIT and the original mammogram, was 0.66±0.15, which is equivalent to the method using the original mammogram. CONCLUSION: In the future, it will be necessary to increase the number of cases and further improve the segmentation.

Exercise Suppresses Head and Neck Squamous Cell Carcinoma Growth via Oncostatin M.

Major advances have been made in cancer treatment, but the prognosis for elderly cancer patients with sarcopenia and frailty remains poor. Myokines, which are thought to exert preventive effects against sarcopenia, have been reported to be associated with the prognosis of various cancers, but their effect on head and neck squamous cell carcinoma (HNSCC) is unknown. The aim of this study was to clarify the influence of exercise on the control of HNSCC and to examine the underlying mechanism involved. Mice were injected with HSC-3-M3 cells, a human cell line of highly metastatic and poorly differentiated tongue cancer, at the beginning of the study. Just prior to transplantation, blood was collected from the mice, and the levels of myokines were measured by ELISA. Oncostatin M (OSM), a selected myokine, was added to HSC-3-M3 cells, after which the cell proliferation ability, cell cycle, and protein expression were analyzed in vitro. Tumor cell viability was lower (control: 100%, exercise: 75%), tumors were smaller (control: 26.2 mm3, exercise: 6.4 mm3), and survival was longer in the exercise group than in the control group in vivo. OSM inhibited HSC-3-M3 cell proliferation in a concentration-dependent manner in vitro. The addition of OSM increased the proportion of cells in the G0/G1 phase, decreased the proportion of cells in the G2/M phase, and increased the expression of the CDK inhibitors p21 and p27. These results indicate that exercise may directly inhibit the proliferation of HNSCC cell lines via OSM.

Patient Re-Identification Based on Deep Metric Learning in Trunk Computed Tomography Images Acquired from Devices from Different Vendors.

During radiologic interpretation, radiologists read patient identifiers from the metadata of medical images to recognize the patient being examined. However, it is challenging for radiologists to identify "incorrect" metadata and patient identification errors. We propose a method that uses a patient re-identification technique to link correct metadata to an image set of computed tomography images of a trunk with lost or wrongly assigned metadata. This method is based on a feature vector matching technique that uses a deep feature extractor to adapt to the cross-vendor domain contained in the scout computed tomography image dataset. To identify "incorrect" metadata, we calculated the highest similarity score between a follow-up image and a stored baseline image linked to the correct metadata. The re-identification performance tests whether the image with the highest similarity score belongs to the same patient, i.e., whether the metadata attached to the image are correct. The similarity scores between the follow-up and baseline images for the same "correct" patients were generally greater than those for "incorrect" patients. The proposed feature extractor was sufficiently robust to extract individual distinguishable features without additional training, even for unknown scout computed tomography images. Furthermore, the proposed augmentation technique further improved the re-identification performance of the subset for different vendors by incorporating changes in width magnification due to changes in patient table height during each examination. We believe that metadata checking using the proposed method would help detect the metadata with an "incorrect" patient identifier assigned due to unavoidable errors such as human error.

Chemical Shift-Encoded MRI of the Lumbar Vertebral Bone Marrow for Detecting Osteoporosis With Low Trabecular Bone Quality in Patients With Breast Cancer Receiving Aromatase Inhibitors.

BACKGROUND: Osteoporosis with low trabecular bone quality (OLB) in patients with breast cancer receiving aromatase inhibitor (AI) therapy is associated with an increased risk of vertebral fractures. The capability of chemical shift-encoded MRI (CSE-MRI) in detecting OLB needs to be investigated. PURPOSE: To assess the diagnostic performance of proton density fat fraction (PDFF) and R2* measurements from CSE-MRI for detecting OLB in postmenopausal women with breast cancer undergoing AI therapy. STUDY TYPE: Prospective. POPULATION: 126 postmenopausal females (mean age: 69.5 ± 8.8 years) receiving AIs (average period: 41.6 ± 26.5 months) after breast cancer surgery. FIELD STRENGTH/SEQUENCE: 1.5-T, three-dimensional CSE-MRI (six echoes), T1-weighted Dixon, short tau inversion recovery, and diffusion-weighted images. ASSESSMENT: Both CSE-MRI and dual-energy x-ray absorptiometry were performed on the same day. Measurements included averaged PDFF, R2*, bone mineral density (BMD), and trabecular bone score (TBS) from L1 to L4 vertebrae. A T-score ≤ -2.5 from BMD measurements indicated osteoporosis, whereas T-scores of ≤ - 2.5 plus TBS ≤-3.7 indicated OLB. The diagnostic performance of PDFF, R2*, and the combination of PDFF and R2* for identifying osteoporosis or OLB was assessed. STATISTICAL TESTS: Student's t-test; Mann-Whitney U test; χ2 or Fisher exact tests; Pearson correlation; multivariate analysis; Receiver operating characteristic (ROC) analysis with the area under the curve (AUC); logistic regression model; intraclass correlation coefficient. A P-value <0.05 was considered statistically significant. RESULTS: For detecting osteoporosis, AUC values were 0.59 (PDFF), 0.66 (R2*), and 0.65 (combined PDFF and R2*). Significant mean differences were noted between patients with and without OLB for PDFF (66.11 ± 5.36 vs. 57.49 ± 6.43) and R2* (46.62 ± 9.24 vs. 63.36 ± 12.44). AUC values for detecting OLB were 0.75 (PDFF), 0.82 (R2*), and 0.84 (combined PDFF and R2*). DATA CONCLUSION: R2* may perform better than PDFF for identifying OLB in patients with breast cancer receiving AIs. LEVEL OF EVIDENCE: 2 TECHNICAL EFFICACY: Stage 4.

Continuous Condensed Triplet Accumulation for Irradiance-Induced Anticounterfeit Afterglow.

Afterglow room-temperature emission that is independent of autofluorescence after ceasing excitation is a promising technology for state-of-the-art bioimaging and security devices. However, the low brightness of the afterglow emission is a current limitation for using such materials in a variety of applications. Herein, the continuous formation of condensed triplet excitons for brighter afterglow room-temperature phosphorescence is reported. (S)-(-)-2,2'-Bis(diphenylphosphino)-1,1'-binaphthyl ((S)-BINAP) incorporated in a crystalline host lattice showed bright green afterglow room-temperature phosphorescence under strong excitation. The small triplet-triplet absorption cross-section of (S)-BINAP in the whole range of visible wavelengths greatly suppressed the deactivation caused by Förster resonance energy transfer from excited states of (S)-BINAP to the accumulated triplet excitons of (S)-BINAP under strong continuous excitation. The steady-state concentration of the triplet excitons for (S)-BINAP reached 2.3 × 10-2  M, producing a bright afterglow. Owing to the brighter afterglow, afterglow detection using individual particles with sizes approaching the diffraction limit in aqueous conditions and irradiance-dependent anticounterfeiting can be achieved.

Comparative efficacy and safety of adjunctive drugs to levodopa for fluctuating Parkinson's disease - network meta-analysis.

It remains unclear which adjunctive drug for Parkinson's disease (PD) in combination with levodopa is more effective, tolerable, and safe. We aimed to compare the efficacy, tolerability, and safety among anti-PD drugs from several classes in patients with fluctuating PD who received levodopa through network meta-analysis (NMA). Twelve anti-PD drugs belonging to 4 different drug classes (dopamine agonists, monoamine oxidase type B inhibitors, catechol-O-methyl transferase inhibitors, and an adenosine A2A receptor antagonist) were selected. We systematically searched PubMed, Embase, and the Cochrane Library for eligible randomized controlled trials (RCTs) comparing placebo with anti-PD drug or among anti-PD drugs in patients with PD who experienced motor fluctuations or wearing-off and received levodopa. We included 54 RCTs in the analysis. The NMA was performed under a frequentist framework using a random-effects model. The efficacy outcome was change in daily off-time, and the tolerability outcome was discontinuation due to all causes. Safety outcomes included discontinuation due to adverse events (AEs) and the incidence of AEs, dyskinesia, hallucination, and orthostatic hypotension. According to the surface under the cumulative ranking curve (SUCRA) in the NMA, ropinirole transdermal patch (SUCRA, 0.861) ranked the highest in efficacy, followed by pramipexole (0.762), ropinirole extended release (ER) (0.750), and safinamide (0.691). In terms of tolerability, ropinirole (0.954) ranked the highest, followed by pramipexole (0.857), safinamide (0.717), and ropinirole ER (0.708). Each anti-PD drug had different SUCRA ranking profiles for the safety outcomes. These findings suggest that ropinirole, pramipexole, and safinamide are well-balanced anti-PD drugs that satisfy both efficacy and tolerability outcomes.

Applicability Evaluation of Full-Reference Image Quality Assessment Methods for Computed Tomography Images.

Image quality assessments (IQA) are an important task for providing appropriate medical care. Full-reference IQA (FR-IQA) methods, such as peak signal-to-noise ratio (PSNR) and structural similarity (SSIM), are often used to evaluate imaging conditions, reconstruction conditions, and image processing algorithms, including noise reduction and super-resolution technology. However, these IQA methods may be inapplicable for medical images because they were designed for natural images. Therefore, this study aimed to investigate the correlation between objective assessment by some FR-IQA methods and human subjective assessment for computed tomography (CT) images. For evaluation, 210 distorted images were created from six original images using two types of degradation: noise and blur. We employed nine widely used FR-IQA methods for natural images: PSNR, SSIM, feature similarity (FSIM), information fidelity criterion (IFC), visual information fidelity (VIF), noise quality measure (NQM), visual signal-to-noise ratio (VSNR), multi-scale SSIM (MSSSIM), and information content-weighted SSIM (IWSSIM). Six observers performed subjective assessments using the double stimulus continuous quality scale (DSCQS) method. The performance of IQA methods was quantified using Pearson's linear correlation coefficient (PLCC), Spearman rank order correlation coefficient (SROCC), and root-mean-square error (RMSE). Nine FR-IQA methods developed for natural images were all strongly correlated with the subjective assessment (PLCC and SROCC > 0.8), indicating that these methods can apply to CT images. Particularly, VIF had the best values for all three items, PLCC, SROCC, and RMSE. These results suggest that VIF provides the most accurate alternative measure to subjective assessments for CT images.

Bilayered lanthanide squarate hydrates (Ln = Eu to Lu) and magnetization reversal barriers of 21 K for Ln = Tm and 57 K for Yb.

A squarate (sq) bridge was applied to heavy lanthanide (Ln) complexes for possible development of high-performance single-ion magnets (SIMs). A selective synthetic method for lanthanide squarate hydrates [Ln2(sq)3(H2O)8]n (abbreviated as Ln-sq) has been established, where Ln stands for Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu. As the crystallographic analysis clarified, they all form a bilayer polymeric structure, which is isomorphic to known Eu-sq and Tb-sq. The coordination structures are described as an almost ideal square antiprism with D4d symmetry. Frequency-dependent susceptibility was recorded in alternating-current magnetic measurements for Ln = Tb, Dy, Er, Tm, and Yb. In particular, as for less studied Ln-based SIMs, the effective magnetization-reversal barriers, Ueff/kB = 21.4 ± 0.4 K (in a bias field of 1800 Oe) and 57.0 ± 0.4 K (400 Oe), were characterized for Tm-sq and Yb-sq, respectively, according to the Orbach-type relaxation mechanism plus a direct or the Raman mechanism. The barrier found for Yb-sq is the highest among those for all the compounds investigated here, and also regarded as one of the largest values for the Yb-based SIMs known so far. The complete-active-space self-consistent-field (CASSCF) calculation was applied to Tm- and Yb-sq. The ground doublet states with mJ = ±6 for the Tm3+ ion and mJ = ±7/2 for the Yb3+ ion were reproduced.

Analgesic effect of safinamide mesylate in a rat model of neuropathic pain.

Pain is one of the most frequent non-motor symptoms of Parkinson's disease (PD). Neuropathic pain is highly prevalent in PD and negatively affects the quality of life of patients with PD. However, there is currently no evidence-based treatment for its control. Safinamide, a monoamine oxidase (MAO)-B inhibitor with a sodium channel inhibitory effect, showed improvement in PD-related pain in several clinical trials. However, it is unclear for which of the various types of pain in PD safinamide is effective. The aim of the present study was to examine the effect of safinamide on neuropathic pain in a rat model of chronic constriction injury (CCI). Pain was evaluated on postoperative days 14 and 21 using von Frey or weight-bearing tests. Male CCI model rats showed a decreased paw withdrawal threshold and a weight-bearing deficit on postoperative days 14 and 21. Single oral administration of safinamide (15, 30, 45 or 70 mg/kg) dose-dependently improved neuropathic pain in both pain assessments on day 14. Subsequently, the 15 and 45 mg/kg dose groups were administered safinamide orally once daily until day 21. With repeated administration, the effect of safinamide on pain was enhanced. The present findings show that safinamide improves neuropathic pain in male CCI model rats. Further animal model research and pathological and molecular pharmacological investigations are warranted.

Improving animal-specific radiotherapy quality assurance for kilovoltage X-ray radiotherapy using a 3D printed dog skull water phantom.

Background: Accurate dose assessment during animal radiotherapy is beneficial for veterinary medicine and medical education. Aim: To visualize the radiation treatment distribution of orthovoltage X-ray equipment in clinical practice using Monte Carlo simulations and create a dog skull water phantom for animal-specific radiotherapy. Methods: EGSnrc-based BEAMnrc and DOSXYZnrc codes were used to simulate orthovoltage dose distributions. At 10, 20, 30, 40, 50, and 80 mm in a water phantom, the depth dose was measured with waterproof Farmer dosimetry chambers, and the diagonal off-axis ratio was measured with Gafchromic EBT3 film to simulate orthovoltage dose distributions. Energy differences between orthovoltage and linear accelerated radiotherapy were assessed with a heterogeneous bone and tissue virtual phantom. The animal-specific phantom for radiotherapy quality assurance (QA) was created from CT scans of a dog and printed with a three-dimensional printer using polyamide 12 nylon, with insertion points for dosimetry chambers and Gafchromic EBT3 film. Results: Monte Carlo simulated and measured dose distributions differed by no more than 2.0% along the central axis up to a depth of 80 mm. The anode heel effect occurred in shallow areas. The orthovoltage radiotherapy percentage depth dose in bone was >40%. Build-up was >40%, with build-down after bone exit, whereas linear accelerator radiotherapy absorption changed little in the bone. A highly water-impermeable, animal-specific dog skull water phantom could be created to evaluate dose distribution. Conclusion: Animal-specific water phantoms and Monte Carlo simulated pre-treatment radiotherapy are useful QA for orthovoltage radiotherapy and yield a visually familiar phantom that will be useful for veterinary medical education.

Safinamide as an Adjunct to Levodopa in Asian and Caucasian Patients With Parkinson's Disease and Motor Fluctuations: A Post Hoc Analysis of the SETTLE Study.

OBJECTIVE: Safinamide is a selective, reversible monoamine oxidase B inhibitor with demonstrated efficacy and tolerability in placebo-controlled studies and is clinically useful for patients with motor fluctuations. This study evaluated the efficacy and safety of safinamide as a levodopa adjunct therapy in Asian patients with Parkinson's disease. METHODS: Data from 173 Asian and 371 Caucasian patients from the international Phase III SETTLE study were included in this post hoc analysis. The safinamide dose was increased from 50 mg/day to 100 mg/day if no tolerability issues occurred at week 2. The primary outcome was the change from baseline to week 24 in daily ON-time without troublesome dyskinesia (i.e., ON-time). Key secondary outcomes included changes in Unified Parkinson's Disease Rating Scale (UPDRS) scores. RESULTS: Safinamide significantly increased daily ON-time relative to placebo in both groups (least-squares mean: 0.83 hours, p = 0.011 [Asians]; 1.05 hours, p < 0.0001 [Caucasians]). Motor function relative to placebo (UPDRS Part III) improved significantly in Asians (-2.65 points, p = 0.012) but not Caucasians (-1.44 points, p = 0.0576). Safinamide did not worsen Dyskinesia Rating Scale scores in either subgroup, regardless of the presence or absence of dyskinesia at baseline. Dyskinesia was largely mild for Asians and moderate for Caucasians. None of the Asian patients experienced adverse events leading to treatment discontinuation. CONCLUSION: Safinamide as a levodopa adjunct is well tolerated and effective in reducing motor fluctuations in both Asian and Caucasian patients. Further studies to investigate the real-world effectiveness and safety of safinamide in Asia are warranted.

Sustained response in early responders to safinamide in patients with Parkinson's disease and motor fluctuations: A post hoc analysis of the SETTLE study.

Safinamide is a selective, reversible, monoamine oxidase B inhibitor for the treatment of patients with Parkinson's disease (PD) and motor fluctuations. This was a post hoc analysis of the SETTLE study, in which patients with PD and motor fluctuations were randomly assigned to 24-week treatment with safinamide (50 mg/day for 2 weeks, increased to 100 mg/day if tolerated) or placebo. In the present analysis, responders were defined according to their treatment responses at Week 2 and Week 24 based on changes in ON-time without troublesome dyskinesia from baseline with cutoffs of 1 hour. It was found that 81% (103/127) of the responders at Week 2 maintained the response through Week 24 in the safinamide group. Other outcomes did not necessarily coincide with the ON-time response; however, "Early" responders who showed a treatment response at both Week 2 and Week 24 had substantial improvements from baseline in OFF-time, UPDRS Part II and III scores, and PDQ-39 summary index scores through Week 24. The safinamide group had a higher proportion of early responders than the placebo group (39% vs 20%, p < 0.0001). At baseline, early responders in the safinamide group had significantly higher UPDRS Part II and III scores, shorter ON-time, and longer OFF-time than the other responder populations. In conclusion, the results of the present post hoc analysis suggest that patients with a short ON-time, severe motor symptoms, and highly compromised activities of daily living can benefit from safinamide early in treatment and over the long term.

Deep Learning-based Post Hoc CT Denoising for the Coronary Perivascular Fat Attenuation Index.

RATIONALE AND OBJECTIVES: Coronary inflammation related to high-risk hemorrhagic plaques can be captured by the perivascular fat attenuation index (FAI) using coronary computed tomography angiography (CCTA). Since the FAI is susceptible to image noise, we believe deep learning (DL)-based post hoc noise reduction can improve diagnostic capability. We aimed to assess the diagnostic performance of the FAI in DL-based denoised high-fidelity CCTA images compared with coronary plaque magnetic resonance imaging (MRI) delivered high-intensity hemorrhagic plaques (HIPs). MATERIALS AND METHODS: We retrospectively reviewed 43 patients who underwent CCTA and coronary plaque MRI. We generated high-fidelity CCTA images by denoising the standard CCTA images using a residual dense network that supervised the denoising task by averaging three cardiac phases with nonrigid registration. We measured the FAIs as the mean CT value of all voxels (range of -190 to -30 HU) located within a radial distance from the outer proximal right coronary artery wall. The diagnostic reference standard was defined as HIPs (high-risk hemorrhagic plaques) using MRI. The diagnostic performance of the FAI in the original and denoised images was assessed using receiver operating characteristic curves. RESULTS: Of 43 patients, 13 had HIPs. The denoised CCTA improved the area under the curve (0.89 [95% confidence interval (CI) 0.78-0.99]) of the FAI compared with that in the original image (0.77 [95% CI, 0.62-0.91], p = 0.008). The optimal cutoff value for predicting HIPs in denoised CCTA was -69 HU with 0.85 (11/13) sensitivity, 0.79 (25/30) specificity, and 0.80 (36/43) accuracy. CONCLUSION: DL-based denoised high-fidelity CCTA improved the AUC and specificity of the FAI for predicting HIPs.

WT1 Trio Peptide-Based Cancer Vaccine for Rare Cancers Expressing Shared Target WT1.

No standard treatment has been established for most rare cancers. Here, we report a clinical trial of a biweekly WT1 tri-peptide-based vaccine for recurrent or advanced rare cancers. Due to the insufficient number of patients available for a traditional clinical trial, the trial was designed for rare cancers expressing shared target molecule WT1. The recruitment criteria included WT1-expressing tumors as well as HLA-A*24:02 or 02:01. The primary endpoints were immunoglobulin G (IgG) antibody (Ab) production against the WT1-235 cytotoxic T lymphocyte (CTL) epitope and delayed-type hypersensitivity (DTH) skin reactions to targeted WT1 CTL epitopes. The secondary endpoints were safety and clinical efficacy. Forty-five patients received WT1 Trio, and 25 (55.6%) completed the 3-month protocol treatment. WT1-235 IgG Ab was positive in 88.0% of patients treated with WT1 Trio at 3 months, significantly higher than 62.5% of the weekly WT1-235 CTL peptide vaccine. The DTH positivity rate in WT1 Trio was 62.9%, which was not significantly different from 60.7% in the WT1-235 CTL peptide vaccine. The WT1 Trio safety was confirmed without severe treatment-related adverse events, except grade 3 myasthenia gravis-like symptoms observed in a patient with thymic cancer. Fifteen (33.3%) patients achieved stable disease after 3 months of treatment. In conclusion, the biweekly WT1 Trio vaccine containing the WT1-332 helper T lymphocyte peptide induced more robust immune responses targeting WT1 than the weekly WT1-235 CTL peptide vaccine. Therefore, WT1-targeted immunotherapy may be a potential therapeutic strategy for rare cancers.

Performance improvement in multi-label thoracic abnormality classification of chest X-rays with noisy labels.

PURPOSE: This study aimed at developing a deep learning-based method for multi-label thoracic abnormality classification on frontal view chest X-ray (CXR). To improve the performance of classification, issues of class imbalance, noisy labels and ensemble of networks are addressed in the paper. METHODS: The experiments were performed on a public dataset called Chest X-ray 14 (CXR14), which includes 112,120 frontal view CXRs from 30,805 patients. We came up with an ensemble learning framework to improve the classification and a noisy label detection method to detect the CXRs with noisy labels. The detected CXRs were reviewed by two board-certificated radiologists in a consensus fashion to evaluate detected noisy labels. The classification was assessed on CXR14 with area under the receiver operating characteristic curve (AUC). RESULTS: Report from the radiologists indicated that detected noisy labels had high possibility to be true positives. A notable improvement from baseline in performance of classification was observed with the ensemble learning framework. After removing the CXRs with detected noisy labels, 8 out of 14 abnormalities improved significantly on CXR14. The suggested framework achieved AUC score of 0.827 on CXR14. CONCLUSION: The methods of this study boost the classification on CXR with awareness of the label noise. Expanded experimental results show that all of them were able to improve multi-label thoracic abnormality classification performance, respectively. A new state-of-the-art is achieved in this study.

Generative adversarial network-based post-processed image super-resolution technology for accelerating brain MRI: comparison with compressed sensing.

BACKGROUND: It is unclear whether deep-learning-based super-resolution technology (SR) or compressed sensing technology (CS) can accelerate magnetic resonance imaging (MRI) . PURPOSE: To compare SR accelerated images with CS images regarding the image similarity to reference 2D- and 3D gradient-echo sequence (GRE) brain MRI. MATERIAL AND METHODS: We prospectively acquired 1.3× and 2.0× faster 2D and 3D GRE images of 20 volunteers from the reference time by reducing the matrix size or increasing the CS factor. For SR, we trained the generative adversarial network (GAN), upscaling the low-resolution images to the reference images with twofold cross-validation. We compared the structural similarity (SSIM) index of accelerated images to the reference image. The rate of incorrect answers of a radiologist discriminating faster and reference image was used as a subjective image similarity (ISM) index. RESULTS: The SR demonstrated significantly higher SSIM than the CS (SSIM=0.9993-0.999 vs. 0.9947-0.9986; P < 0.001). In 2D GRE, it was challenging to discriminate the SR image from the reference image, compared to the CS (ISM index 40% vs. 17.5% in 1.3×; P = 0.039 and 17.5% vs. 2.5% in 2.0×; P = 0.034). In 3D GRE, the CS revealed a significantly higher ISM index than the SR (22.5% vs. 2.5%; P = 0.011) in 2.0 × faster images. However, the ISM index was identical for the 2.0× CS and 1.3× SR (22.5% vs. 27.5%; P = 0.62) with comparable time costs. CONCLUSION: The GAN-based SR outperformed CS in image similarity with 2D GRE for MRI acceleration. In addition, CS was more advantageous in 3D GRE than SR.