A Spatio-Temporal Ensemble Deep Learning Architecture for Real-Time Defect Detection during Laser Welding on Low Power Embedded Computing Boards.

In modern production environments, advanced and intelligent process monitoring strategies are required to enable an unambiguous diagnosis of the process situation and thus of the final component quality. In addition, the ability to recognize the current state of product quality in real-time is an important prerequisite for autonomous and self-improving manufacturing systems. To address these needs, this study investigates a novel ensemble deep learning architecture based on convolutional neural networks (CNN), gated recurrent units (GRU) combined with high-performance classification algorithms such as k-nearest neighbors (kNN) and support vector machines (SVM). The architecture uses spatio-temporal features extracted from infrared image sequences to locate critical welding defects including lack of fusion (false friends), sagging, lack of penetration, and geometric deviations of the weld seam. In order to evaluate the proposed architecture, this study investigates a comprehensive scheme based on classical machine learning methods using manual feature extraction and state-of-the-art deep learning algorithms. Optimal hyperparameters for each algorithm are determined by an extensive grid search. Additional work is conducted to investigate the significance of various geometrical, statistical and spatio-temporal features extracted from the keyhole and weld pool regions. The proposed method is finally validated on previously unknown welding trials, achieving the highest detection rates and the most robust weld defect recognition among all classification methods investigated in this work. Ultimately, the ensemble deep neural network is implemented and optimized to operate on low-power embedded computing devices with low latency (1.1 ms), demonstrating sufficient performance for real-time applications.

Communication in high risk ante-natal consultations: a direct observational study of interactions between patients and obstetricians.

BACKGROUND: Effective communication is crucial to any doctor-patient consultation, not least in pregnancy where the outcome affects more than one person. While higher levels of patient participation and shared decision making are recognised as desirable, there is little agreement on how best to achieve this. Most previous research in this area is based on reported data such as interviews or surveys and there is a need for more fine-grained analysis of authentic interaction. This study aimed to identify the discourse characteristics and patterns that exemplify effective communication practices in a high-risk ante-natal clinic. METHODS: We video-recorded 20 consultations in a high-risk ante-natal clinic in a large New Zealand city with patients attending for the first time. Post-consultation interviews were conducted with the 20 patients and 13 obstetricians involved. Discourse analysis of the transcripts and videos of the consultations was conducted, in conjunction with thematic analysis of interview transcripts. RESULTS: Most patients reported high quality communication and high levels of satisfaction; the detailed consultation analysis revealed a range of features likely to have contributed. On the clinician side, these included clear explanations, acknowledgement of the patient's experience, consideration of patient wishes, and realistic and honest answers to patient questions. On the patient side, these included a high level of engagement with technical aspects of events and procedures, and appropriate questioning of obstetricians. CONCLUSIONS: This study has demonstrated the utility of combining direct observation of consultations with data from patient experience interviews to identify specific features of effective communication in routine obstetric ante-natal care. The findings are relevant to improvements needed in obstetric communication identified in the literature, especially in relation to handling psychosocial issues and conveying empathy, and may be useful to inform communication training for obstetricians. The presence of the unborn child may provide an added incentive for parents to develop their own health literacy and to be an active participant in the consultation on behalf of their child. The findings of this study can lay the groundwork for further, more detailed analysis of communication in ante-natal consultations.

Maternal supplementation alone with Lactobacillus rhamnosus HN001 during pregnancy and breastfeeding does not reduce infant eczema.

BACKGROUND: In a randomized placebo-controlled trial, we previously found that the probiotic Lactobacillus rhamnosus HN001 (HN001) taken by mothers from 35 weeks of gestation until 6 months post-partum if breastfeeding and their child from birth to age 2 years halved the risk of eczema during the first 2 years of life. We aimed to test whether maternal supplementation alone is sufficient to reduce eczema and compare this to our previous study when both the mother and their child were supplemented. METHODS: In this 2-centre, parallel double-blind, randomized placebo-controlled trial, the same probiotic as in our previous study (HN001, 6 × 109 colony-forming units) was taken daily by mothers from 14-16 weeks of gestation till 6 months post-partum if breastfeeding, but was not given directly to the child. Women were recruited from the same study population as the first study, where they or their partner had a history of treated asthma, eczema or hay fever. RESULTS: Women were randomized to HN001 (N = 212) or placebo (N = 211). Maternal-only HN001 supplementation did not significantly reduce the prevalence of eczema, SCORAD ≥ 10, wheeze or atopic sensitization in the infant by 12 months. This contrasts with the mother and child intervention study, where HN001 was associated with reductions in eczema (hazard ratio (HR): 0.39, 95% CI 0.19-0.79, P = .009) and SCORAD (HR = 0.61, 95% 0.37-1.02). However, differences in the HN001 effect between studies were not significant. HN001 could not be detected in breastmilk from supplemented mothers, and breastmilk TGF-ß/IgA profiles were unchanged. CONCLUSION: Maternal probiotic supplementation without infant supplementation may not be effective for preventing infant eczema.

Severe maternal morbidity due to sepsis: The burden and preventability of disease in New Zealand.

BACKGROUND: Sepsis is a life-threatening systemic condition that appears to be increasing in the obstetric population. Clinical detection can be difficult and may result in increased morbidity via delays in the continuum of patient care. AIMS: To describe the burden of severe maternal morbidity (SMM) caused by sepsis in New Zealand and investigate the potential preventability. METHODS: A multidisciplinary expert review panel was established to review cases of obstetric sepsis admitted to intensive care or high-dependency units over an 18 month span in New Zealand. Cases were then analysed for the characteristics of infection and their preventability. RESULTS: Fifty cases met the inclusion criteria, most commonly due to uterine, respiratory or kidney infection. Fifty per cent (25) of these cases were deemed potentially preventable, predominantly due to delays in diagnosis and treatment. CONCLUSIONS: A high index of suspicion, development of early recognition systems and multi-disciplinary training are recommended to decrease preventable cases of maternal sepsis.

Early pregnancy probiotic supplementation with Lactobacillus rhamnosus HN001 may reduce the prevalence of gestational diabetes mellitus: a randomised controlled trial.

The study aims to assess whether supplementation with the probiotic Lactobacillus rhamnosus HN001 (HN001) can reduce the prevalence of gestational diabetes mellitus (GDM). A double-blind, randomised, placebo-controlled parallel trial was conducted in New Zealand (NZ) (Wellington and Auckland). Pregnant women with a personal or partner history of atopic disease were randomised at 14-16 weeks' gestation to receive HN001 (6×109 colony-forming units) (n 212) or placebo (n 211) daily. GDM at 24-30 weeks was assessed using the definition of the International Association of Diabetes and Pregnancy Study Groups (IADPSG) (fasting plasma glucose ≥5·1 mmol/l, or 1 h post 75 g glucose level at ≥10 mmol/l or at 2 h ≥8·5 mmol/l) and NZ definition (fasting plasma glucose ≥5·5 mmol/l or 2 h post 75 g glucose at ≥9 mmol/l). All analyses were intention-to-treat. A total of 184 (87 %) women took HN001 and 189 (90 %) women took placebo. There was a trend towards lower relative rates (RR) of GDM (IADPSG definition) in the HN001 group, 0·59 (95 % CI 0·32, 1·08) (P=0·08). HN001 was associated with lower rates of GDM in women aged ≥35 years (RR 0·31; 95 % CI 0·12, 0·81, P=0·009) and women with a history of GDM (RR 0·00; 95 % CI 0·00, 0·66, P=0·004). These rates did not differ significantly from those of women without these characteristics. Using the NZ definition, GDM prevalence was significantly lower in the HN001 group, 2·1 % (95 % CI 0·6, 5·2), v. 6·5 % (95 % CI 3·5, 10·9) in the placebo group (P=0·03). HN001 supplementation from 14 to 16 weeks' gestation may reduce GDM prevalence, particularly among older women and those with previous GDM.

The Probiotics in Pregnancy Study (PiP Study): rationale and design of a double-blind randomised controlled trial to improve maternal health during pregnancy and prevent infant eczema and allergy.

BACKGROUND: Worldwide there is increasing interest in the manipulation of human gut microbiota by the use of probiotic supplements to modify or prevent a range of communicable and non-communicable diseases. Probiotic interventions administered during pregnancy and breastfeeding offer a unique opportunity to influence a range of important maternal and infant outcomes. The aim of the Probiotics in Pregnancy Study (PiP Study) is to assess if supplementation by the probiotic Lactobacillus rhamnosus HN001 administered to women from early pregnancy and while breastfeeding can reduce the rates of infant eczema and atopic sensitisation at 1 year, and maternal gestational diabetes mellitus, bacterial vaginosis and Group B Streptococcal vaginal colonisation before birth, and depression and anxiety postpartum. METHODS/DESIGN: The PiP Study is a two-centre, randomised, double-blind placebo-controlled trial in Wellington and Auckland, New Zealand. Four hundred pregnant women expecting infants at high risk of allergic disease will be enrolled in the study at 14-16 weeks gestation and randomised to receive either Lactobacillus rhamnosus HN001 (6 × 10(9) colony-forming units per day (cfu/day)) or placebo until delivery and then continuing until 6 months post-partum, if breastfeeding. Primary infant outcomes are the development and severity of eczema and atopic sensitisation in the first year of life. Secondary outcomes are diagnosis of maternal gestational diabetes mellitus, presence of bacterial vaginosis and vaginal carriage of Group B Streptococcus (at 35-37 weeks gestation). Other outcome measures include maternal weight gain, maternal postpartum depression and anxiety, infant birth weight, preterm birth, and rate of caesarean sections. A range of samples including maternal and infant faecal samples, maternal blood samples, cord blood and infant cord tissue samples, breast milk, infant skin swabs and infant buccal swabs will be collected for the investigation of the mechanisms of probiotic action. DISCUSSION: The study will investigate if mother-only supplementation with Lactobacillus rhamnosus HN001 in pregnancy and while breastfeeding can reduce rates of eczema and atopic sensitisation in infants by 1 year, and reduce maternal rates of gestational diabetes mellitus, bacterial vaginosis, vaginal carriage of Group B Streptococcus before birth and maternal depression and anxiety postpartum. TRIAL REGISTRATION: Australian New Zealand Clinical Trials Registration: ACTRN12612000196842. Date Registered: 15/02/12.

Cardiac disease in pregnancy and the multidisciplinary team: the Wellington experience.

AIM: To define the range and severity of cardiac disease in pregnant women in New Zealand, as well as the maternal and neonatal morbidity and mortality compared with the background obstetric population. METHODS: We retrospectively audited pregnant women with cardiac comorbidity seen by a multidisciplinary team at a tertiary referral centre consisting of midwives, cardiologists, obstetricians and anaesthetists in 2016-2017. RESULTS: Seventy-two women were referred to the multidisciplinary team. The most common referral reasons were arrhythmia (n=20, 27.8%), congenital anomalies (n=19, 26.4%) and palpitations (n=10, 13.9%). Fifty-two of these women were found to be at increased risk of morbidity or mortality. A specific delivery plan was devised for 37 of these women (69.8%). There was no serious maternal morbidity or mortality. Instrumental delivery rates were higher for women with cardiac comorbidity than the background obstetric population (19.2% vs 10.8%, p=0.049), however, neonatal admissions were not increased (11.5% compared with 16.5%). CONCLUSION: Multidisciplinary review of obstetric patients with cardiac disease provides an important service to ensure risk modification prior to conception and throughout pregnancy and the puerperium.

The diagnostic role of ventilation/perfusion scans versus computed tomography pulmonary angiography in obstetric patients investigated for pulmonary embolism at Wellington Hospital from 2010 to 2012.

AIM: To develop best practice clinical guidelines for the use of ventilation/perfusion (V/Q) scanning and computed tomography perfusion angiography (CTPA) in pregnancy and the postpartum period. METHOD: Retrospective analysis of the clinical findings and radiologic investigation for pulmonary embolism (PE) in obstetric women at Wellington Hospital from 2010 to 2012. RESULTS: Fifty-four women were investigated for PE with a V/Q scan or CTPA, including 29 antenatal women and 25 postnatal women. Eleven (37.9%) antenatal women had V/Q scans and 18 (62%) had CTPAs. Five (20%) postnatal women had V/Q scans, 19 (76%) had CTPAs and one (4%) had a V/Q scan followed by a CTPA. Three of the 54 women (5.6%) had a positive radiologic finding of PE (two by V/Q scan and one by CTPA). Four (22.2%) antenatal women and 5 (25%) postnatal women had a diagnosis made on CTPA, which was not seen on chest x-ray. CONCLUSION: This audit found that clinicians varied in their investigation of cases suspected of PE. We have proposed a clinical pathway for the investigation of PE in pregnancy and the postpartum period.

Outcomes in HrHPV-positive women with low grade cervical smears and normal or low grade initial colposcopy results.

AIM: To determine outcomes in HrHPV-positive women with low grade cervical smears and normal or low grade initial colposcopy biopsy results in a cohort of women over a 2-year follow-up period. BACKGROUND: The revised National Cervical Screening (NCS) programme guidelines in New Zealand were implemented in October 2009. The guidelines state that women 30 years and older should undergo reflex HPV testing. If this test confirms the presence of HrHPV, women are to be referred for a colposcopic assessment. The guidelines do not mention what the follow-up period should be of women with HrHPV and normal or low grade abnormalities at colposcopy/biopsy. METHOD: In this study we followed up women 30 years and older referred to Wellington Hospital from 1/10/2009 to 1/10/2011 with a LGSIL or ASC-US smear and positive HrHPV test. Those with a normal or low grade biopsy result were followed over a 2-year period to determine outcomes. RESULTS: Our study found that 4% of women with initial normal biopsy results and 15.2% with initial low grade results had progressed to high grade (CIN 2/3/invasion) over a 2-year follow-up period. During the same time period, 68% of women with an initial normal biopsy and 61% with a low grade biopsy had a normal colposcopy after 2 years. Twenty-eight percent of women with normal and 24% of those with initial with low grade biopsy continued to have LG abnormalities at 2 years of follow-up. CONCLUSION: Women 30 years and older who are HrHPV-positive and have low grade abnormalities at colposcopic biopsy may be followed up with a 12-month cervical smear rather than repeat colposcopy as the risk of progression to a high grade abnormality is low.

Is obstetrics and gynaecology going the distance?

AIMS: To determine how distance learning should be delivered to trainee interns in New Zealand, and compare these methods with the views of teachers. BACKGROUND: Many sixth-year medical students (trainee interns) in Obstetrics and Gynaecology (O&G) are placed in provincial centres. Feedback surveys suggested that these students felt disadvantaged using distance education compared with students at the main base. Currently distance education is delivered using traditional methods (textbooks and DVDs) and 'Web 1.0' software (e-mails and static material on blackboard™). METHOD: Using a mix of structured interviews and a survey we asked how staff and students would prefer to access distance education, how familiar they were with 'Web 2.0' software (wikis, podcasts, blogs and social networking) and what is required to make distance education work well. RESULTS: 'Web 2.0' technologies were used by students socially but not academically, except for Wikipedia. Both staff and students felt competent using 'Web 1.0' (the internet), but many students did not access blackboard™ (a course management and content system) for information, citing the limited material there as a reason. Both groups highly valued face-to-face contact, and distant students were frustrated by the inability to attend tutorials given at the base medical school. CONCLUSIONS: A blended learning delivery comprising videoconferencing combined with a better use of blackboard™ would meet the learning preferences of the students. Universities should consider investing in and encouraging videoconferencing and a better use of current technology, rather than implementing 'Web 2.0' technology.