Deep Learning-Driven Transformation: A Novel Approach for Mitigating Batch Effects in Diffusion MRI Beyond Traditional Harmonization.

BACKGROUND: "Batch effect" in MR images, due to vendor-specific features, MR machine generations, and imaging parameters, challenges image quality and hinders deep learning (DL) model generalizability. PURPOSE: We aim to develop a DL model using contrast adjustment and super-resolution to reduce diffusion-weighted images (DWIs) diversity across magnetic field strengths and imaging parameters. STUDY TYPE: Retrospective. SUBJECTS: The DL model was built using an open dataset from one individual. The MR machine identification model was trained and validated on a dataset of 1134 adults (54% females, 46% males), with 1050 subjects showing no DWI abnormalities and 84 with conditions like stroke and tumors. The 21,000 images were divided into 80% for training, 20% for validation, and 3500 for testing. FIELD STRENGTH/SEQUENCE: Seven MR scanners from four manufacturers with 1.5 T and 3 T magnetic field strengths. DWIs were acquired using spin-echo sequences and high-resolution T2WIs using the T2-SPACE sequence. ASSESSMENT: An experienced, board-certified radiologist evaluated the effectiveness of restoring high-resolution T2WI and harmonizing diverse DWI with metrics such as PSNR and SSIM, and the texture and frequency attributes were further analyzed using gray-level co-occurrence matrix and 1-dimensional power spectral density. The model's impact on machine-specific characteristics was gauged through the performance metrics of a ResNet-50 model. Comprehensive statistical tests were employed for statistical robustness, including McNemar's test and the Dice index. RESULTS: Our DL protocol reduced DWI contrast and resolution variation. ResNet-50 model's accuracy decreased from 0.9443 to 0.5786, precision from 0.9442 to 0.6494, recall from 0.9443 to 0.5786, and F1 score from 0.9438 to 0.5587. The t-SNE visualization indicated more consistent image features across multiple MR devices. Autoencoder halved learning iterations; Dice coefficient >0.74 confirmed signal reproducibility in 84 lesions. CONCLUSION: This study presents a DL strategy to mitigate batch effects in diffusion MR images, improving their quality and generalizability. EVIDENCE LEVEL: 3 TECHNICAL EFFICACY: Stage 1.

Automation of a Rule-based Workflow to Estimate Age from Brain MR Imaging of Infants and Children Up to 2 Years Old Using Stacked Deep Learning.

PURPOSE: Myelination-related MR signal changes in white matter are helpful for assessing normal development in infants and children. A rule-based myelination evaluation workflow regarding signal changes on T1-weighted images (T1WIs) and T2-weighted images (T2WIs) has been widely used in radiology. This study aimed to simulate a rule-based workflow using a stacked deep learning model and evaluate age estimation accuracy. METHODS: The age estimation system involved two stacked neural networks: a target network-to extract five myelination-related images from the whole brain, and an age estimation network from extracted T1- and T2WIs separately. A dataset was constructed from 119 children aged below 2 years with two MRI systems. A four-fold cross-validation method was adopted. The correlation coefficient (CC), mean absolute error (MAE), and root mean squared error (RMSE) of the corrected chronological age of full-term birth, as well as the mean difference and the upper and lower limits of 95% agreement, were measured. Generalization performance was assessed using datasets acquired from different MR images. Age estimation was performed in Sturge-Weber syndrome (SWS) cases. RESULTS: There was a strong correlation between estimated age and corrected chronological age (MAE: 0.98 months; RMSE: 1.27 months; and CC: 0.99). The mean difference and standard deviation (SD) were -0.15 and 1.26, respectively, and the upper and lower limits of 95% agreement were 2.33 and -2.63 months. Regarding generalization performance, the performance values on the external dataset were MAE of 1.85 months, RMSE of 2.59 months, and CC of 0.93. Among 13 SWS cases, 7 exceeded the limits of 95% agreement, and a proportional bias of age estimation based on myelination acceleration was exhibited below 12 months of age (P = 0.03). CONCLUSION: Stacked deep learning models automated the rule-based workflow in radiology and achieved highly accurate age estimation in infants and children up to 2 years of age.

Nondestructive and noncontact evaluation of cellulose nanofiber-reinforced composites using terahertz time-domain spectroscopy.

Cellulose nanofibers (CNFs) can be obtained from natural sources, such as plants and wood fibers. Thermoplastic resin composites reinforced with CNFs exhibit various features, including superior mechanical strength. As the mechanical properties of composites reinforced with CNFs are affected by the amount of fiber addition, it is important to determine the concentration of the CNF filler in the matrix after injection or extrusion molding. We confirmed that there was good linearity between the CNF concentration and terahertz absorption. We could distinguish the difference in the 1%-point-CNF concentration with terahertz time-domain spectroscopy. Furthermore, we estimated the mechanical properties of the CNF nanocomposites using terahertz information.

Gene regulatory network analysis defines transcriptome landscape with alternative splicing of human umbilical vein endothelial cells during replicative senescence.

BACKGROUND: Endothelial cell senescence is the state of permanent cell cycle arrest and plays a critical role in the pathogenesis of age-related diseases. However, a comprehensive understanding of the gene regulatory network, including genome-wide alternative splicing machinery, involved in endothelial cell senescence is lacking. RESULTS: We thoroughly described the transcriptome landscape of replicative senescent human umbilical vein endothelial cells. Genes with high connectivity showing a monotonic expression increase or decrease with the culture period were defined as hub genes in the co-expression network. Computational network analysis of these genes led to the identification of canonical and non-canonical senescence pathways, such as E2F and SIRT2 signaling, which were down-regulated in lipid metabolism, and chromosome organization processes pathways. Additionally, we showed that endothelial cell senescence involves alternative splicing. Importantly, the first and last exon types of splicing, as observed in FLT1 and ACACA, were preferentially altered among the alternatively spliced genes during endothelial senescence. We further identified novel microexons in PRUNE2 and PSAP, each containing 9 nt, which were altered within the specific domain during endothelial senescence. CONCLUSIONS: These findings unveil the comprehensive transcriptome pathway and novel signaling regulated by RNA processing, including gene expression and splicing, in replicative endothelial senescence.

Spectral imaging of pharmaceutical materials with a compact terahertz difference-frequency generation semiconductor source.

Spectral imaging of pharmaceutical material using a compact ultra-broadband (1-4 THz) terahertz semiconductor source was demonstrated. False-color RGB images could be obtained using a simple procedure (calibration free). The ability to distinguish the polymorphism of carbamazepine (CBZ), the hydrate forms of D-(+)-glucose and caffeine, and the crystallinity of nifedipine was demonstrated using the THz DFG source. Crystal forms of pharmaceutical materials can be distinguished using this method.

Clinicopathologic Study of Placentas From Women With a Fontan Circulation.

BACKGROUND: Pregnant women with a Fontan circulation have a high risk of obstetric complications, such as preterm delivery and small for gestational age (SGA), which may be affected by low blood flow to the placenta and hypoxia. This study investigated placental pathology in a Fontan circulation.Methods and Results:Eighteen pregnancies in 11 women with a Fontan circulation were reviewed. Pregnancy outcomes showed 9 miscarriages and 9 live births, with 4 preterm deliveries. Five neonates were SGA (<5th percentile). Eight placentas from live births in 7 women were available for the study. Five placentas had low weight placenta for gestational age, and 7 grossly showed a chronic subchorionic hematoma. Histological examination revealed all placentas had some form of histological hypoxic lesions: maternal vascular malperfusion in 7, fetal vascular malperfusion in 1, and other hypoxia-related lesions in 8. Quantitative analyses, including immunohistochemistry (CD31, CD68, and hypoxia inducible factor-1α antibodies) and Masson's trichrome staining, were also performed and compared with 5 control placentas. Capillary density and the area of fibrosis were significantly greater in placentas from women with a Fontan circulation than in control placentas. CONCLUSIONS: Placentas in a Fontan circulation were characterized by a high frequency of low placental weight, chronic subchorionic hematoma, and constant histological hypoxic changes, which could reflect altered maternal cardiac conditions and lead to poor pregnancy outcomes.

Directional seed and pollen dispersal and their separate effects on anisotropy of fine-scale spatial genetic structure among seedlings in a dioecious, wind-pollinated, and wind-dispersed tree species, Cercidiphyllum japonicum.

Prevailing directions of seed and pollen dispersal may induce anisotropy of the fine-scale spatial genetic structure (FSGS), particularly in wind-dispersed and wind-pollinated species. To examine the separate effects of directional seed and pollen dispersal on FSGS, we conducted a population genetics study for a dioecious, wind-pollinated, and wind-dispersed tree species, Cercidiphyllum japonicum Sieb. et Zucc, based on genotypes at five microsatellite loci of 281 adults of a population distributed over a ca. 80 ha along a stream and 755 current-year seedlings. A neighborhood model approach with exponential-power-von Mises functions indicated shorter seed dispersal (mean = 69.1 m) and much longer pollen dispersal (mean = 870.6 m), effects of dispersal directions on the frequencies of seed and pollen dispersal, and the directions with most frequent seed and pollen dispersal (prevailing directions). Furthermore, the distance of effective seed dispersal within the population was estimated to depend on the dispersal direction and be longest at the direction near the prevailing direction. Therefore, patterns of seed and pollen dispersal may be affected by effective wind directions during the period of respective dispersals. Isotropic FSGS and spatial sibling structure analyses indicated a significant FSGS among the seedlings generated by the limited seed dispersal, but anisotropic analysis for the seedlings indicated that the strength of the FSGS varied with directions between individuals and was weakest at a direction near the directions of the most frequent and longest seed dispersal but far from the prevailing direction of pollen dispersal. These results suggest that frequent and long-distance seed dispersal around the prevailing direction weakens the FSGS around the prevailing direction. Therefore, spatially limited but directional seed dispersal would determine the existence and direction of FSGS among the seedlings.

Identification of hub molecules of FUS-ALS by Bayesian gene regulatory network analysis of iPSC model: iBRN.

Fused in sarcoma/translated in liposarcoma (FUS) is a causative gene of amyotrophic lateral sclerosis (ALS). Mutated FUS causes accumulation of DNA damage and cytosolic stress granule (SG) formation, thereby motor neuron (MN) death. However, key molecular aetiology remains unclear. Here, we applied a novel platform technology, iBRN, "Non- biased" Bayesian gene regulatory network analysis based on induced pluripotent stem cell (iPSC)-derived cell model, to elucidate the molecular aetiology using transcriptome of iPSC-derived MNs harboring FUSH517D. iBRN revealed "hub molecules", which strongly influenced transcriptome network, such as miR-125b-5p-TIMELESS axis and PRKDC for the molecular aetiology. Next, we confirmed miR-125b-5p-TIMELESS axis in FUSH517D MNs such that miR-125b-5p regulated several DNA repair-related genes including TIMELESS. In addition, we validated both introduction of miR-125b-5p and knocking down of TIMELESS caused DNA damage in the cell culture model. Furthermore, PRKDC was strongly associated with FUS mis-localization into SGs by DNA damage under impaired DNA-PK activity. Collectively, our iBRN strategy provides the first compelling evidence to elucidate molecular aetiology in neurodegenerative diseases.

DGCR8-dependent efficient pri-miRNA processing of human pri-miR-9-2.

Microprocessor complex, including DiGeorge syndrome critical region gene 8 (DGCR8) and DROSHA, recognizes and cleaves primary transcripts of microRNAs (pri-miRNAs) in the maturation of canonical miRNAs. The study of DGCR8 haploinsufficiency reveals that the efficiency of this activity varies for different miRNA species. It is thought that this variation might be associated with the risk of schizophrenia with 22q11 deletion syndrome caused by disruption of the DGCR8 gene. However, the underlying mechanism for varying action of DGCR8 with each miRNA remains largely unknown. Here, we used in vivo monitoring to measure the efficiency of DGCR8-dependent microprocessor activity in cultured cells. We confirmed that this system recapitulates the microprocessor activity of endogenous pri-miRNA with expression of a ratiometric fluorescence reporter. Using this system, we detected mir-9-2 as one of the most efficient targets. We also identified a novel DGCR8-responsive RNA element, which is highly conserved among mammalian species and could be regulated at the epi-transcriptome (RNA modification) level. This unique feature between DGCR8 and pri-miR-9-2 processing may suggest a link to the risk of schizophrenia.

Clinical management of coagulation status and placenta previa in a pregnant woman with Marfan's syndrome after mitral and aortic mechanical heart valve replacement.

Pregnancy with a mechanical heart valve (MHV) is high risk for valve thrombosis because pregnancy is a hyper-coagulation state. In addition, warfarin use during pregnancy is restricted due to its fetotoxicity, and postpartum bleeding may be increased with anticoagulation. In particular, placenta previa under anticoagulation may cause massive postpartum bleeding. The optimal anticoagulation for a pregnant woman with mitral and aortic double MHVs is not known. In addition, suitable techniques for control of bleeding in a case of placenta previa under anticoagulation are not known. Thus, a case of a pregnant woman with mitral and aortic double MHVs and placenta previa is presented. The case was managed without valve thrombosis through precise unfractionated heparin dose adjustment and frequent activated partial thromboplastin time monitoring, along with maintenance of antithrombin levels. Compression sutures were found to be effective for controlling bleeding from the attachment site of placenta previa even under anticoagulation.

Pregnancy outcomes in women with dilated cardiomyopathy: Peripartum cardiovascular events predict post delivery prognosis.

BACKGROUND: The number of pregnant women with dilated cardiomyopathy (DCM) is relatively small, and therefore their prognosis after pregnancy is unknown. This study aims to elucidate pregnancy outcomes among women with DCM, as well as the long-term prognosis after pregnancy. METHODS: Thirty-five pregnancies and deliveries in 30 women, diagnosed with DCM before pregnancy, were retrospectively analyzed. RESULTS: All women had a left ventricular ejection fraction (LVEF) over 30% and belonged to the New York Heart Association (NYHA) class I or II before pregnancy. The mean gestational age at delivery was 36 weeks with 15 (43%) preterm deliveries. Eight pregnancies (23%) were complicated by peripartum cardiac events including 1 ventricular arrhythmia, 6 heart failures, and 1 significant deterioration in LVEF requiring termination of pregnancy. NYHA class II, pre-pregnancy use of angiotensin-converting enzyme inhibitor/angiotensin II receptor blocker/diuretics, elevated brain natriuretic peptide (BNP), and advanced diastolic dysfunction assessed by Doppler echocardiography were defined as risk factors for cardiac events. Although the more severe cases took beta-blockers during pregnancy, the rates of cardiac events and decreasing LVEF did not differ significantly between those taking beta-blockers and those who were not. Values of LVEF decreased by almost 10% after the average 4-year post-delivery follow-up period. The long-term event-free survival was considerably worse among women with peripartum cardiac events than in those without (p<0.0001). CONCLUSIONS: DCM women with pre-pregnancy LVEF over 30% tolerated pregnancy, but the rate of preterm delivery was high. Peripartum cardiovascular events occurred more often in women with NYHA class II, as well as those who received medications before and during pregnancy and showed more elevated BNP and advanced diastolic dysfunction before pregnancy. Beta-blockers likely allowed similar outcomes for DCM patients with lower initial LVEFs. Close monitoring later in life is required, particularly among the women with peripartum cardiac events.

Detection of single human hairs with a terahertz nonlinear quantum cascade laser.

We report the demonstration of imaging of a single human hair with a terahertz quantum cascade laser (THz-QCL) source based on intracavity difference-frequency generation. A single human hair whose diameter was about 100 µm was detected using the THz-QCL source operating at 240 K, of which the THz beam had a linear polarization. The results show that the THz image of a human hair clearly depends on the polarization direction of the THz beam. The THz QCL sources that are capable of room temperature operation will be useful for detection of small foreign objects like human hairs.

Appropriate delivery method for cardiac disease pregnancy based on noninvasive cardiac monitoring.

Background There are numerous significant physiological changes occurring in circulation during labor. To detect these rapid hemodynamic changes, invasive and intermittent measurement techniques are not reliable. To suggest a suitable delivery method for pregnancy with cardiac disease, this study analyzed how each delivery method influences cardiac function using a noninvasive and continuous measurement technique. Methods A prospective study was accomplished at the National Cerebral and Cardiovascular Center in Japan from October 1, 2014, to November 30, 2018. The classification of the healthy heart pregnant women was according to the delivery method: vaginal delivery (VD) without epidural anesthesia, VD with epidural anesthesia, and caesarean section (CS). The hemodynamic parameters cardiac index (CI), stroke volume index (SI), and heart rate (HR) were evaluated regularly throughout delivery by noninvasive electrical cardiometry monitor. Results Ten cases were examined for each group. CI and HR were significantly increased before VD, while the increase in CI and HR was mild in the epidural group in comparison to the nonepidural group. SI was increased toward the delivery in the epidural group, and it was constant in the nonepidural group. However, there was no alteration in the level of outcomes of the two groups. In CS, SI increased and HR decreased before delivery. After delivery, SI continued to increase, while HR did not change but CI increased. Conclusion In VD, the increase in venous circulation according to the autotransfusion is managed by increasing HR. By epidural anesthesia, the increase in HR was suppressed and SI was increased. However, as epidural anesthesia increases the vascular capacity, the level of SI outcome was comparable. In CS, the HR was decreased because of the spinal anesthesia and the SI was increased because of many factors like hydration. As there are many factors to control in CS, VD with epidural anesthesia will be the first preference for most cardiac patients.

Analysis of the nucleocytoplasmic shuttling RNA-binding protein HNRNPU using optimized HITS-CLIP method.

RNA-binding proteins (RBPs) control many types of post-transcriptional regulation, including mRNA splicing, mRNA stability, and translational efficiency, by directly binding to their target RNAs and their mutation and dysfunction are often associated with several human neurological diseases and tumorigenesis. Crosslinking immunoprecipitation (CLIP), coupled with high-throughput sequencing (HITS-CLIP), is a powerful technique for investigating the molecular mechanisms underlying disease pathogenesis by comprehensive identification of RBP target sequences at the transcriptome level. However, HITS-CLIP protocol is still required for some optimization due to experimental complication, low efficiency and time-consuming, whose library has to be generated from very small amounts of RNAs. Here we improved a more efficient, rapid, and reproducible CLIP method by optimizing BrdU-CLIP. Our protocol produced a 10-fold greater yield of pre-amplified CLIP library, which resulted in a low duplicate rate of CLIP-tag reads because the number of PCR cycles required for library amplification was reduced. Variance of the yields was also reduced, and the experimental period was shortened by 2 days. Using this, we validated IL-6 expression by a nuclear RBP, HNRNPU, which directly binds the 3'-UTR of IL-6 mRNA in HeLa cells. Importantly, this interaction was only observed in the cytoplasmic fraction, suggesting a role of cytoplasmic HNRNPU in mRNA stability control. This optimized method enables us to accurately identify target genes and provides a snapshot of the protein-RNA interactions of nucleocytoplasmic shuttling RBPs.

Terahertz optical properties of wood-plastic composites.

Terahertz (THz) optical materials containing polymeric materials have been useful for terahertz technologies. We investigated the THz optical properties of wood-plastic composites (WPCs), which are composed of polystyrene and wood powder, and their suitability as THz optical materials. We found that the refractive indexes and absorption coefficients of the WPCs increased with increasing wood powder content. WPCs are inexpensive and have tunable THz optical properties.

Identification of a selective DDX3X inhibitor with newly developed quantitative high-throughput RNA helicase assays.

The DEAD-box family of RNA helicases plays essential roles in both transcriptional and translational mRNA degradation; they unwind short double-stranded RNA by breaking the RNA-RNA interactions. Two DEAD-box RNA helicases, eukaryotic translation initiation factor 4A3 (eIF4A3) and DEAD-box helicase 3 (DDX3X), show high homology in the ATP-binding region and are considered key molecules for cancer progression. Several small molecules that target eIF4A3 and DDX3X have been reported to inhibit cancer cell growth; however, more potent compounds are required for cancer therapeutics, and there is a critical need for high-throughput assays to screen for RNA helicase inhibitors. In this study, we developed novel fluorescence resonance energy transfer-based high-throughput RNA helicase assays for eIF4A3 and DDX3X. Using these assays, we identified several eIF4A3 allosteric inhibitors whose inhibitory effect on eIF4A3 ATPase showed a strong correlation with inhibitory effect on helicase activity. From 102 compounds that exhibited eIF4A3 ATPase inhibition, we identified a selective DDX3X inhibitor, C1, which showed stronger inhibition of DDX3X than of eIF4A3. Small-molecule helicase inhibitors can be valuable for clarifying the molecular machinery of DEAD-box RNA helicases. The high-throughput quantitative assays established here should facilitate the evaluation of the helicase inhibitory activity of compounds.

Environment-dependent introgression from Quercus dentata to a coastal ecotype of Quercus mongolica var. crispula in northern Japan.

Introgression from one species in a specific environment to another may facilitate colonization of the environment by the recipient species. However, such environment-dependent introgression has been clarified in limited plant taxa. In northern Japan, there are two interfertile oak species: Quercus dentata (Qd) in coastal areas and Q. mongolica var. crispula (Qc) in inland areas. However, at higher latitudes where Qd is rare, a coastal Qc ecotype with Qd-like traits is distributed in the coastal areas. We distinguished inland Qc, coastal Qc, and coastal Qd populations based on genome-wide genotypes and multitrait phenotypes and verified introgression from coastal Qd to coastal Qc using reduced library sequencing. Genotypes and phenotypes differed among the populations, and coastal Qc was intermediate between inland Qc and coastal Qd. The ABBA-BABA test showed introgression from coastal Qd to coastal Qc. In coastal Qc, we found various stages of introgression after the first generation of backcross but detected no genomic regions where introgression was enhanced. Overall, we show evidence for introgression from a coastal species to an ecotype of an inland species, which has colonized the coastal environment. It remains unclear whether introgressed alleles are selected in the coastal environment.

Spatial variation in bird pollination and its mitigating effects on the genetic diversity of pollen pools accepted by Camellia japonica trees within a population at a landscape level.

Bird pollination can vary spatially in response to spatial fluctuations in flowering even within plant populations. In this study, we examined the hypothesis that the spatial variation in bird pollination may induce mitigating effects, which maintains or increases genetic diversity of pollen pools at local sites with low flowering densities. To test this hypothesis, we analyzed the landscape-level genetic effects within a population of Camellia japonica on the pollen pools accepted by individuals in two reproductive years by using genotypes at eight microsatellite loci of 1323 seeds from 19 seed parents. Regression analyses using the quadratic models of correlated paternity between pollen pools against spatial distances between the seed-parent pairs revealed not only local pollination but also some amount of long-distance pollen dispersal. The genetic diversity of pollen pools accepted by seed parents tended to be negatively related to the densities of flowering individuals near the seed parents during winter (when the effective pollination of C. japonica is mediated mostly by Zosterops japonica). We show that the low density of flowering individuals may induce the expansion of the foraging areas of Z. japonica and consequently increase the genetic diversity of pollen pools. This spatial variation in bird pollination may induce the mitigating effects on the C. japonica population. The comparisons between the two study years indicate that the overall pattern of bird pollination and the genetic effects described here, including the mitigating effects, may be stable over time.