Patient-specific placental vessel segmentation with limited data.

A major obstacle in applying machine learning for medical fields is the disparity between the data distribution of the training images and the data encountered in clinics. This phenomenon can be explained by inconsistent acquisition techniques and large variations across the patient spectrum. The result is poor translation of the trained models to the clinic, which limits their implementation in medical practice. Patient-specific trained networks could provide a potential solution. Although patient-specific approaches are usually infeasible because of the expenses associated with on-the-fly labeling, the use of generative adversarial networks enables this approach. This study proposes a patient-specific approach based on generative adversarial networks. In the presented training pipeline, the user trains a patient-specific segmentation network with extremely limited data which is supplemented with artificial samples generated by generative adversarial models. This approach is demonstrated in endoscopic video data captured during fetoscopic laser coagulation, a procedure used for treating twin-to-twin transfusion syndrome by ablating the placental blood vessels. Compared to a standard deep learning segmentation approach, the pipeline was able to achieve an intersection over union score of 0.60 using only 20 annotated images compared to 100 images using a standard approach. Furthermore, training with 20 annotated images without the use of the pipeline achieves an intersection over union score of 0.30, which, therefore, corresponds to a 100% increase in performance when incorporating the pipeline. A pipeline using GANs was used to generate artificial data which supplements the real data, this allows patient-specific training of a segmentation network. We show that artificial images generated using GANs significantly improve performance in vessel segmentation and that training patient-specific models can be a viable solution to bring automated vessel segmentation to the clinic.

Autonomous Magnetic Navigation in Endoscopic Image Mosaics.

Endoscopes navigate within the human body to observe anatomical structures with minimal invasiveness. A major shortcoming of their use is their narrow field-of-view during navigation in large, hollow anatomical regions. Mosaics of endoscopic images can provide surgeons with a map of the tool's environment. This would facilitate procedures, improve their efficiency, and potentially generate better patient outcomes. The emergence of magnetically steered endoscopes opens the way to safer procedures and creates an opportunity to provide robotic assistance both in the generation of the mosaic map and in navigation within this map. This paper proposes methods to autonomously navigate magnetic endoscopes to 1) generate endoscopic image mosaics and 2) use these mosaics as user interfaces to navigate throughout the explored area. These are the first strategies, which allow autonomous magnetic navigation in large, hollow organs during minimally invasive surgeries. The feasibility of these methods is demonstrated experimentally both in vitro and ex vivo in the context of the treatment of twin-to-twin transfusion syndrome. This minimally invasive procedure is performed in utero and necessitates coagulating shared vessels of twin fetuses on the placenta. A mosaic of the vasculature in combination with autonomous navigation has the potential to significantly facilitate this challenging surgery.

Fetal surgery for spina bifida in Zurich: results from 150 cases.

PURPOSE: Over the past 10 years, over 150 fetal spina bifida surgeries were performed at the Zurich Center for Fetal Diagnosis and Therapy. This study looks at surrogates for success and failure of this approach. METHODS: We focused on key outcome parameters including hydrocephalus shunt rate at one year, bladder control at 4, independent ambulation at 3 years, and maternal, fetal, and neonatal complications. RESULTS: From the first 150 patients undergoing fetal surgery for spina bifida, 148 (98.7%) were included in the study. Maternal-fetal surgery was uneventful in 143/148 (97%) cases. Intraoperative problems included resuscitation in 4/148 fetuses (2.7%). 1/148 fetuses (0.7%) died on postoperative day 4. Maternal complications included chorioamniotic membrane separation in 22/148 (15%), lung embolism in 3/148 (2.1%), chorioamnionitis in 2/148 (1.4%), AV-block III and uterine rupture in 1/148 each (0.7%). 1/148 (0.7%) newborn death was recorded. Hindbrain herniation was identified preoperatively in 132/148 (90%) fetuses and resolved completely in 119/132 (90%). At one year, 39/106 (37%) children had required a CSF diversion. At 4 years, 4/34 patients (12%) had normal bladder control. At 3 years, 48/57 (84%) walked independently. CONCLUSION: A majority of patients benefitted from prenatal intervention, in that the shunt rate was lower and the rates of continent and walking patients were higher than reported with postnatal care.

Prenatal Sonographic Head Circumference and Cerebral Ventricle Width Measurements Before and After Open Fetal Myelomeningocele Repair - Prediction of Shunting During the First Year of Life.

PURPOSE: The aim of this study was to describe the sonographic evolution of fetal head circumference (HC) and width of the posterior horn of the lateral ventricle (Vp) after open fetal myelomeningocele (fMMC) repair and to assess whether pre- or postoperative measurements are helpful to predict the need for shunting during the first year of life. PATIENTS & METHODS: All 30 children older than one year by January 2017 who previously had fMMC repair at the Zurich Center for Fetal Diagnosis and Therapy were included. Sonographic evolution of fetal HC and Vp before and after fMMC repair was assessed and compared between the non-shunted (N = 16) and the shunted group (N = 14). ROC curves were generated for the fetal HC Z-score and Vp in order to show their predictive accuracy for the need for shunting until 1 year of age. RESULTS: HC was not an independent factor for predicting shunting. However, the need for shunting was directly dependent on the preoperative Vp as well as the Vp before delivery. A Vp > 10 mm at evaluation for fMMC repair or > 15 mm before delivery identifies 100 % of the infants needing shunt placement at a false-positive rate of 44 % and 25 %, respectively. All fetuses with a Vp > 15 mm at first evaluation received a shunt. CONCLUSION: Fetuses demonstrating a Vp of > 15 mm before in utero MMC repair are extremely likely to develop hydrocephalus requiring a shunt during the first year of life. This compelling piece of evidence must be appropriately integrated into prenatal counseling.

In utero Plastic Surgery in Zurich: Successful Use of Distally Pedicled Random Pattern Transposition Flaps for Definitive Skin Closure during Open Fetal Spina Bifida Repair.

BACKGROUND: One of the intraoperative challenges of fetal spina bifida repair is skin closure when there is an extended skin defect. Thus, we examined whether distally pedicled random pattern transposition flaps (TFs) are a valid option to overcome this problem. SUBJECTS AND METHODS: All patients undergoing in utero repair of spina bifida with application of a TF for back skin closure were analyzed focusing on intraoperative flap characteristics and postoperative flap performance. RESULTS: In 30 (70%) of the 43 fetuses a primary skin closure was achieved, in 5 (12%) a skin substitute was used, and in 8 (18%) a TF was applied. Flap raising and insertion was uneventful and perfusion was sufficient in all 8 fetuses (100%). In 3 fetuses (37%) the donor sites were closed primarily, and in 5 (63%) a skin substitute was used for coverage. At birth, 7 flaps were viable and provided robust skin coverage over the center of the former lesion. Complications included a small skin defect with CSF leakage in 1 patient (13%). CONCLUSION: During open fetal spina bifida repair, TFs can be safely and efficaciously used to obtain solid and durable skin coverage over lesions too large to allow conventional primary skin closure.

Fetal Cholelithiasis: Antenatal Diagnosis and Neonatal Follow-Up in a Case of Twin Pregnancy - A Case Report and Review of the Literature.

Fetal cholelithiasis is a rare finding during a third-trimester ultrasound with an average incidence rate of 0.07-1.15%. We report a case of fetal cholelithiasis in twins, observed in a patient with monochorionic diamniotic twin pregnancy hospitalized at our unit for signs of premature labor. We present the outcome of the 2 neonates with a clinical and sonographic follow-up. In addition, we offer a comprehensive review of the literature available to date.

Congenital Lung Lesion: Prenatal Course, Therapy and Predictors of Perinatal Outcome.

Purpose To describe the prenatal course and perinatal outcome, and to define prognostic markers for fetuses with congenital pulmonary airway malformation (CPAM) or bronchopulmonary sequestration (BPS). Materials and Methods A retrospective study was performed at the University Hospital Zurich including pregnancies with either fetal CPAM (n = 26) or BPS (n = 11) between 2000 and 2013. Results Three patients decided for termination of pregnancy. Two intrauterine deaths (CPAM) occurred at 25 weeks. Minimally invasive interventions were performed in 9/37 (24 %) fetuses, post-interventional survival was 8/9 (89 %). Mean gestational age at delivery was 38.1 +/-2.8 and 39.1 +/-2.5 weeks in fetuses with CPAM or BPS, respectively. In fetuses with CPAM the perinatal mortality rate was 4/24 (17 %); the rate of invasive interventions or surgery during the early neonatal period (neonatal morbidity) was 9/22 (41 %). Prenatal diagnosis of hydrothorax and/or increasing cystic volume ratio (CVR) until delivery preceded perinatal death in 3/5 (60 %). Absent mediastinal shift showed a neonatal morbidity rate of 1/8 (13 %) without any perinatal mortality. In fetuses with BPS the perinatal morbidity and mortality were both 1/10 (10 %). Hydrops predicted morbidity and mortality in 100 % of cases. Absent hydrops was followed by uncomplicated perinatal outcome. Conclusion Fetuses with CPAM or BPS have a good outcome under optimal perinatal care including the possibility to perform minimally invasive prenatal interventions. CPAM without mediastinal shift and BPS without hydrops have an excellent prognosis. Hydrothorax, increasing CVR or hydrops indicates a high risk for perinatal morbidity and mortality.

Premiere use of Integra™ artificial skin to close an extensive fetal skin defect during open in utero repair of myelomeningocele.

BACKGROUND: There are fetuses demonstrating very large myelomeningocele lesion which can not be covered with autochothonous skin. MATERIAL AND METHODS: We use Integra™ artificial skin for intrauterine coverage of the back lesion. A reverse latissimus dorsi flap was used postnatally to reinforce the repair site. CONCLUSION: Integra™ appears to be a suitable coverage for large soft tissue defects in utero. Moreover, a postnatal reverse latissimus dorsi flap appears to markedly strengthen tissue coverage over a spinal cord rescued in utero.

Stability considerations of aspartame in the direct analysis of artificial sweeteners in water samples using high-performance liquid chromatography-tandem mass spectrometry (HPLC-MS/MS).

A HPLC-MS/MS method is presented for the simultaneous determination of frequently used artificial sweeteners (ASs) and the main metabolite of aspartame (ASP), diketopiperazine (DKP), in environmental water samples using the direct-injection (DI) technique, thereby achieving limits of quantification (LOQ) of 10 ng L(-1). For a reliable quantification of ASP pH should be adjusted to 4.3 to prevent formation of the metabolite. Acesulfame (ACE), saccharin (SAC), cyclamate (CYC) and sucralose (SUC) were ubiquitously found in water samples. Highest concentrations up to 61 µg L(-1) of ACE were found in wastewater effluents, followed by surface water with concentrations up to 7 µg L(-1), lakes up to 600 ng L(-1) and groundwater and tap water up to 70 ng L(-1). The metabolite DKP was only detected in wastewater up to 200 ng L(-1) and at low detection frequencies.