Novel multi-level optimization-driven 2D/3D matching for reconstructing 3D fetal postures and motion from childbirth MRI during vaginal delivery.

Upper airway segmentation on MR images is a prerequisite step for quantitatively studying the anatomical structure and function of the upper airway and surrounding tissues. However, the complex variability of intensity and shape of anatomical structures and different modes of image acquisition commonly used in this application makes automatic upper airway segmentation challenging. In this paper, we develop and test a comprehensive deep learning-based segmentation system for use on MR images to address this problem. In our study, both static and dynamic MRI data sets are utilized, including 58 axial static 3D MRI studies, 22 mid-retropalatal dynamic 2D MRI studies, 21 mid-retroglossal dynamic 2D MRI studies, 36 mid-sagittal dynamic 2D MRI studies, and 23 isotropic dynamic 3D MRI studies, involving a total of 160 subjects and over 20000 MRI slices. Samples of static and 2D dynamic MRI data sets were randomly divided into training, validation, and test sets by an approximate ratio of 5:2:3. Considering that the variability of annotation data among 3D dynamic MRIs was greater than for other MRI data sets, we increased the ratio of training data for these data to improve the robustness of the model. We designed a unified framework consisting of the following procedures. For static MRI, a generalized region-of-interest (GROI) strategy is applied to localize the partitions of nasal cavity and other portions of upper airway in axial data sets as two separate subobjects. Subsequently, the two subobjects are segmented by two separate 2D U-Nets. The two segmentation results are combined as the whole upper airway structure. The GROI strategy is also applied to other MRI modes. To minimize false-positive and false-negative rates in the segmentation results, we employed a novel loss function based explicitly on these rates to train the segmentation networks. An inter-reader study is conducted to test the performance of our system in comparison to human variability in ground truth (GT) segmentation of these challenging structures. The proposed approach yielded mean Dice coefficients of 0.84±0.03, 0.89±0.13, 0.84±0.07, and 0.86±0.05 for static 3D MRI, mid-retropalatal/mid-retroglossal 2D dynamic MRI, mid-sagittal 2D dynamic MRI, and isotropic dynamic 3D MRI, respectively. The quantitative results show excellent agreement with manual delineation results. The inter-reader study results demonstrate that the segmentation performance of our approach is statistically indistinguishable from manual segmentations considering the inter-reader variability in GT. The proposed method can be utilized for routine upper airway segmentation from static and dynamic MR images with high accuracy and efficiency. The proposed approach has the potential to be employed in other dynamic MRI-related applications, such as lung or heart segmentation.

Strip of the Month: January 2014

Semi-automatic segmentation for 3D motion analysis of the tongue with dynamic MRI

BackgroundMalignant Hepatocellular carcinoma (HCC) is one amongst the foremost widespread cancers within the world. Radiofrequency ablation (RFA) is that the most generally used substitute tool for hepatic carcinoma treatment. Monitoring tumoral response to loco-regional therapy is a vital mission in oncological imaging. Dynamic contrast enhanced MR and recently added subtraction imaging technique improve assessment of the ablated hepatic focal lesions. The aim of this study is that the evaluation of the role of Subtraction MRI within the detection of recurrent or residual tumoral viability after RF ablation.ResultsFifty patients were conducted during this retrospective study, all underwent RFA for 76 hepatic focal lesions and underwent Dynamic MRI study 1 month after ablation. Subtraction imaging was then performed. MRI images were interpreted by two readers who are experienced in hepatic imaging. The primary reader interpreted the standard Dynamic MRI and was blinded to the subtraction sequences; the second reader interpreted both Dynamic and Subtraction MRI images. the primary reader detected 49 resolved cases by dynamic MRI, while the second reader detected residual activity in 6 cases out of 49 via subtraction dynamic MRI (added value). The second reader agreed with first reader in 43 cases (agreement).The first reader detected 27 residual cases by dynamic MRI. The second reader disagreed with first reader in 11 cases appeared resolved out of 27 via subtraction dynamic MRI (added value). The second reader agreed with first reader in residual activity in 16 cases (agreement).The statistical analysis of those results revealed a big additive value of the subtraction imaging to the dynamic MRI (P < 0.001) with moderate degree of agreement between the 2 diagnostic tools (Kappa value = 0.491). This implies that Subtraction MRI significantly improves the reader confidence level within the assessment of treatment response following loco-regional therapies for HCC.ConclusionDynamic Subtraction MRI is a powerful detector for real enhancement in treated HCC lesions after radiofrequency ablation and hence increasing the degree of readers’ confidence and accuracy of treated lesions in follow-up studies. So it's recommended to feature this powerful tool as a routine to any or all Dynamic MRI studies of the Liver.

Objective To investigate the value of MRI in the quantitative evaluation of the function of levator ani muscle after vaginal delivery and pelvic prolapse(POP). Methods Sixty-two cases of vaginal delivery primipara were selected as observation group, 40 cases of non pregnancy healthy women in the same period were selected as control group. GE signa HD 1.5 T superconduct MR scanner was used to perform static and dynamic pelvic magnetic resonance imaging (MRI), the imaging manifestations of levator ani muscle in two groups were observed, the anal muscle injury of two groups were determined by the professional doctors in the same group according to the puborectalis musculus image and the iliococcygeus musculus image. According to the clinical history, physical examination and ultrasound or MRI results, the observation group was divided into the POP group (28 cases) and normal pregnancy group (34 cases), and by measuring maternal H line from the resting period to the maximum force, anus muscle fissure area(LHS), levator ani plate angle(LPA), bilateral iliococcygeus angle(ICA) and other indexes, quantitative comparison of MRI measurement parameters of levator ani muscle in the two subgroups was completed. Results The levator ani muscle injury of observation group accounted for 27.42%(17/62), and the control group had no levator ani muscle injury, the difference was significant (P 0.05); Under maximal abdominal pressure, H line in the resting state, LPA, LHS, bilateral ICA of the POP group were higher than those of the normal pregnancy group and the resting state of the POP group, the differences were significant (P<0.05). Conclusions After vaginal delivery, it is easy to cause the damage to the levator ani muscle and increase the risk of POP. The static and dynamic MRI quantitative test can evaluate the function of the muscle function. Key words: Magnetic resonance imaging; Vaginal delivery; Pelvic prolapse; Function of levator ani muscle; Evaluation value

MRI-based pelvimetric measurements as predictors for a successful vaginal breech delivery in the Frankfurt Breech at term cohort (FRABAT)

Most of the articles on fetal movements used today to assess its health relate to the daily biomechanics of the fetus in the absence of voluntary respiratory retention (apnea) in the mother. Therefore, the dominant information about the perceived strength, frequency and nature of fetal movements in late pregnancy is not used to assess the resistance of her fetus to periods of hypoxia that it will experience during vaginal delivery. The article analyzes a review of scientific articles and inventions devoted to the dynamics of fetal movements during voluntary respiratory retention of the mother and aquarium fish during artificial acute hypoxia. The analogy of the biomechanics of the fetus and aquarium fish in the corresponding models of artificial acute hypoxia is shown, taking into account their high and low resistance to hypoxia. It is reported that the presence of reserves of adaptation of aquarium fish and fruits to acute hypoxia is manifested by their complete absence of movements. It has been established that their adaptation and resistance to hypoxia are the higher the longer their period of immobility during acute hypoxia persists. It was found that the period of immobility of the fetus with maternal apnea lasting 30 seconds or more indicates a good resistance of the fetus to intrauterine hypoxia and its readiness to withstand periods of hypoxia awaiting it during vaginal delivery. In turn, the period of immobility of the fetus with maternal apnea lasting less than 10 seconds indicates an excessively low resistance of the fetus to hypoxia and an increased likelihood of stillbirth and/ or encephalopathy with increasing gestation and in the case of delivery by vaginal delivery. It has been shown that with excessively low fetal resistance to hypoxia, a Cesarean section should be planned to prevent stillbirth and perinatal encephalopathy. The review showed that awareness of pregnant women and doctors about fetal biomechanics during maternal apnea can potentially reduce the incidence of stillbirths and encephalopathy in newborns. The essence of the inventions devoted to the assessment of fetal resistance to hypoxia, based on monitoring the biomechanics of the fetus with voluntary respiratory arrest of the mother, is presented.

To accelerate dynamic MR applications using infimal convolution of total generalized variation functionals (ICTGV) as spatio-temporal regularization for image reconstruction. ICTGV comprises a new image prior tailored to dynamic data that achieves regularization via optimal local balancing between spatial and temporal regularity. Here it is applied for the first time to the reconstruction of dynamic MRI data. CINE and perfusion scans were investigated to study the influence of time dependent morphology and temporal contrast changes. ICTGV regularized reconstruction from subsampled MR data is formulated as a convex optimization problem. Global solutions are obtained by employing a duality based non-smooth optimization algorithm. The reconstruction error remains on a low level with acceleration factors up to 16 for both CINE and dynamic contrast-enhanced MRI data. The GPU implementation of the algorithm suites clinical demands by reducing reconstruction times of one dataset to less than 4 min. ICTGV based dynamic magnetic resonance imaging reconstruction allows for vast undersampling and therefore enables for very high spatial and temporal resolutions, spatial coverage and reduced scan time. With the proposed distinction of model and regularization parameters it offers a new and robust method of flexible decomposition into components with different degrees of temporal regularity. Magn Reson Med 78:142-155, 2017. © 2016 International Society for Magnetic Resonance in Medicine.

Semi-automated bone tracking in dynamic CINE MRI during controlled knee motion.

A model-based approach to reconstruction of 3D human arm motion from a monocular image sequence taken under orthographic projection is presented. The reconstruction is divided into two stages. First, a 2D shape model is used to track the arm silhouettes and second-order curves are used to model the arm based on an iteratively reweighted least square method. As a result, 2D stick figures are extracted. In the second stage, the stick figures are backprojected into the scene. 3D postures are reconstructed using the constraints of a 3D kinematic model of the human arm. The motion of the arm is then derived as a transition between the arm postures. Applications of these results are foreseen in the analysis of human motion patterns.

Pose estimation is crucial for ensuring passenger safety and better user experiences in semi- and fully autonomous vehicles. Traditional methods relying on pose estimation from regular color images face significant challenges due to a lack of three-dimensional (3D) information and the sensitivity to occlusion and lighting conditions. Depth images, which are invariant to lighting issues and provide 3D information about the scene, offer a promising alternative. However, there is a lack of strong work in 3D pose estimation from such images due to the time-consuming process of annotating depth images with 3D postures. In this paper, we present a novel approach to 3D human posture estimation using depth and infrared (IR) images. Our method leverages a three-stage fine-tuning process involving simulation data, approximated data, and a limited set of manually annotated samples. This approach allows us to effectively train a model capable of accurate 3D pose estimation with a median error of under 10 cm across all joints, using fewer than 100 manually annotated samples. To the best of our knowledge, this is the first work focusing on vehicle occupant posture detection utilizing only depth and IR data. Our results demonstrate the feasibility and efficacy of this approach, paving the way for enhanced passenger safety in autonomous vehicle systems.

Backgound: Perinatal brain development undergoes dramatically changing in human life. Multiple insults in perinatal period such as hypoxia or and ischemia, infection and inflammation often cause different severity and patterns of brain injury such as ischemic injury or hemorrahgic injury. To explore the epidemiological rule and correlated risk factors of premature infant with brain injury we used MRI findings to define the diagnosis of brain injury. Methods: The study was prospectively performed on 358 preterm infants from 2008 to 2011at the neonatal intensive care unit of Shengjing Hospital of China Medical University. Detailed Clinical data of all subjects were recorded including: gender, gestation age, birth weight, and the following risk factors: precipiate labor, pregnancy-induced hypertension, fetal distress, multiple pregnancy, prenatal infection, placental abruption, placental previa, gestational diabetes mellitus, prenatal seroid administration, magnesium sulfate, resuscitation history, circulatory disorder, vaginal delivery, early-onset sepsis, mechanical ventilation, blood gas analysis(metabolic acidosis, hyperkalemia, hyponatremia and hypocalcemia). Conventional MRI and diffuse-weighted imaging (DWI) were performed iin 358 preterm infants using 3.0 Tesla MRI scanner. The infants were sedated for imaging with chloral hadrate (50mg/kg). Seriously ill intubated newborn infants were monitored by clinician during scanning and hand-ventilated, and pulse oximetry and electrocardiography were monitored during the procedure. The study protocol were approved by the ethic committee of our university and consented by their parents. All MRI scans were assessed by two radiologists who blinded to neonatal clinical data. The premature infants with brain injury were classified into two groups according to MRI findings: hemorrahgic brain injury and non-hemorrahgic injury (white matter damage). Statistic analyses were performed using SPSS (version 11.5). All data were described as mean±standard deviation. Students't test was used to evaluate the difference in numerical variables. The significance of the difference between premature infants with and without brain injury was tested using the Chi-square test. A logistic regression analysis was performed on factors which found to be significant in univariate analysis. A probability value of P<0.05 was considered statistically significant. Results: There were 128 premature infants with hemorrahgic brain injury (70.7%), and 96 with WMD (53.0%); 43 premature infants with hemorrahgic brain injury complicated with WMD. (23.7%). 21 risk factors were analyzed. Univariate analysis: precipitate labor (X 2=5.295, P=0.021), twin (X2 =4.576, P=0.003), prenatal infection (X2=7.922, P=0.005), circulatory disorder (X2 =5.710, P=0.0017), vaginal delivery (X2=53.624, P=0.000), metabolic acidosis (X2 =13.594, P=0.001), hyponatremia (X 2 =11.691, P=0.001), hypocalcemia (X2 =12.805, P=0.000) for premature infants with hemorrahgic brain were considered statistically significant; prenatal infation (X 2 =5.628, P=0.018), gestational diabetes mellitus (X 2=14.944, P=0.001), magnesium sulfate (X2 =9.248, P=0.002), resuscitation (X 2=6.362, P=0.012), circulatory disorder (X 2=6.341, P=0.012), vaginal delivery (X2 =17.029, P=0.000), metabolic acidosis (X2 =14.944, P=0.001), hyponatremia (X2 =20.242, P=0.000), hypocalcemia (X2 =32.595, P=0.000) for premature infants with WMD were considered statistically significant. Logistic regression analysis: prenatal infection (OR=4.738, 95% CI: 1.201, 18.685, P<0.05), vaginal delivery (OR=9.191, 95% CI: 4.699, 17.979, P<0.05), hyponatremia (OR=3.331, 95% CI: 1.506, 7.366, P<0.05) and hypocalcemia (OR=3.162, 95% CI: 1.325, 7.545, P<0.05) were risk factors for premature infants with hemorrhagic brain injury. Maternal diabetes mellitus(OR=5.211, 95% CI: 1.272, 21.341, P<0.05), vaginal delivery (or=3.078, 95% CI: 1.824, 5.194, P<0.05), hyponatremia (OR=3.331, 95% CI: 1.506, 7.366, P<0.05) and hypocalcemia (OR=4.713, 95% CI: 2.412,9.209, P<0.05) were risk factors for premature infants with WMD. While the prenatal use of magnesium sulfate (OR=0.375, 95% CI: 1.183, 0.766, P<0.05) was an its protective factor. In addition, we did not find the WMD prevalence difference between the premature infants with gestation age less than 34 weeks and the late preterm infants, even the severe diffusive WMD was not unusually found in the late premature infants. Early widespread and diffusive WMD on MRI-DWI often evolved into cystic PVL. Conclusions: Intrauterine exposure including prenatal mother suffering from infection, diabetes mellitus, pattern of delivery and postnatal electrolytes disturbance may contribute to premature brain injury. Prenatal magnesium administration could protect brain from insults. Hierarchy WMD in the premature infants may depend on the severity of insults, but not only on the brain developmental maturation.

Fetal MRI and prenatal diagnosis of congenital heart defects

The aim of this study was to compare pelvic floor anatomy and laxity at rest and on straining (Valsalva's maneuver) using dynamic ultrafast MR imaging in women who were continent versus those with stress incontinence differing in obstetric history. Thirty continent women were divided into three equal groups (nulliparous, previous cesarean delivery, previous vaginal delivery) and compared with 10 women with stress-incontinence with a history of at least one vaginal delivery. MR imaging of the pelvic floor at rest and on maximal strain was performed, using axial T2-weighted fast spin-echo images followed by sagittal ultrafast T2-weighted single-shot fast spin-echo sequences. Mean population age (age range, 22-45 years; mean +/- SD, 36 +/- 5.4 years), was similar in the four groups, as was parity in the three parous groups. Mean distances between the bladder floor and pubococcygeal line at rest did not differ between the four groups. On straining, bladder floor descent was 1.1 +/- 0.9, 1.0 +/- 1.1, and 1.9 +/- 0.9 cm in continent nulliparous, cesarean delivery, and vaginal delivery women, respectively, versus 3.2 +/- 1.0 cm in incontinent women (p = 0.0005). Cervical descent was greater in incontinent versus nulliparous women (p = 0.0019). Bladder floor descent was greater in the continent vaginal delivery group than in continent cesarean delivery control patients (p = 0.04). In patients with stress incontinence, symptoms did not correlate with amplitude of descent. The right levator muscle was thinner overall than the left, regardless of frequency direction (p = 0.001). Ultrafast MR imaging using the T2-weighted single-shot fast spin-echo sequence allows dynamic evaluation of the pelvic compartments at maximal strain with no need for contrast medium. Pelvic floor laxity and supporting fascia abnormalities were most common in patients with stress incontinence followed by continent women with a history of vaginal delivery. The results are therefore compatible with the hypothesis of vaginal delivery as a contributory factor to stress incontinence in older parous women.

To follow a prospective cohort of women during their first term pregnancy to elucidate the nature and timing of changes to the pelvic floor during pregnancy and after vaginal delivery. Enrolled subjects were evaluated at four time points with dynamic MRI, POP-Q examinations, and validated symptom questionnaires. The four assessments occurred during the first trimester (ePG), late third trimester (lPG), within a week after vaginal delivery (ePP), and three months postpartum (lPP). Two-dimensional T1-weighted MRI measurements included bladder descent and area of the levator hiatus at rest and during Valsalva maneuvers. Sample size of ten subjects was calculated for a power of 0.8 to detect a 20% change in bladder position with p < 0.05. Comparative statistical tests were used for parametric and non-parametric data, respectively. Twelve subjects completed the study. At lPP, the bladder descent was increased (p = 0.03) at rest and with Valsalva compared to ePG. Levator hiatus area did not differ (p = 0.63) between time points at rest or with Valsalva. Median POP stage increased (p = 0.001) to 1.5 at lPP. Mean genital hiatus increased (p = 0.0003) at each time point. Higher scores were recorded on the UDI-6 (p < 0.001) and the PFDI-20 (p = 0.003) questionnaires at lPG and ePP, but returned to ePG levels by lPP. Anatomic changes measured by dynamic MRI and POP-Q examinations demonstrate significant descent at 3 months postpartum. However, these anatomic changes did not significantly correlate with changes in symptoms.