Craniofacial phenotyping with fetal MRI: a feasibility study of 3D visualisation, segmentation, surface-rendered and physical models.

This study explores the potential of 3D Slice-to-Volume Registration (SVR) motion-corrected fetal MRI for craniofacial assessment, traditionally used only for fetal brain analysis. In addition, we present the first description of an automated pipeline based on 3D Attention UNet trained for 3D fetal MRI craniofacial segmentation, followed by surface refinement. Results of 3D printing of selected models are also presented.Qualitative analysis of multiplanar volumes, based on the SVR output and surface segmentations outputs, were assessed with computer and printed models, using standardised protocols that we developed for evaluating image quality and visibility of diagnostic craniofacial features. A test set of 25, postnatally confirmed, Trisomy 21 fetal cases (24-36 weeks gestational age), revealed that 3D reconstructed T2 SVR images provided 66-100% visibility of relevant craniofacial and head structures in the SVR output, and 20-100% and 60-90% anatomical visibility was seen for the baseline and refined 3D computer surface model outputs respectively. Furthermore, 12 of 25 cases, 48%, of refined surface models demonstrated good or excellent overall quality with a further 9 cases, 36%, demonstrating moderate quality to include facial, scalp and external ears. Additional 3D printing of 12 physical real-size models (20-36 weeks gestational age) revealed good/excellent overall quality in all cases and distinguishable features between healthy control cases and cases with confirmed anomalies, with only minor manual adjustments required before 3D printing.Despite varying image quality and data heterogeneity, 3D T2w SVR reconstructions and models provided sufficient resolution for the subjective characterisation of subtle craniofacial features. We also contributed a publicly accessible online 3D T2w MRI atlas of the fetal head, validated for accurate representation of normal fetal anatomy.Future research will focus on quantitative analysis, optimizing the pipeline, and exploring diagnostic, counselling, and educational applications in fetal craniofacial assessment.

Total and Regional Brain Volumes in Fetuses With Congenital Heart Disease.

BACKGROUND: Congenital heart disease (CHD) is common and is associated with impaired early brain development and neurodevelopmental outcomes, yet the exact mechanisms underlying these associations are unclear. PURPOSE: To utilize MRI data from a cohort of fetuses with CHD as well as typically developing fetuses to test the hypothesis that expected cerebral substrate delivery is associated with total and regional fetal brain volumes. STUDY TYPE: Retrospective case-control study. POPULATION: Three hundred eighty fetuses (188 male), comprising 45 healthy controls and 335 with isolated CHD, scanned between 29 and 37 weeks gestation. Fetuses with CHD were assigned into one of four groups based on expected cerebral substrate delivery. FIELD STRENGTH/SEQUENCE: T2-weighted single-shot fast-spin-echo sequences and a balanced steady-state free precession gradient echo sequence were obtained on a 1.5 T scanner. ASSESSMENT: Images were motion-corrected and reconstructed using an automated slice-to-volume registration reconstruction technique, before undergoing segmentation using an automated pipeline and convolutional neural network that had undergone semi-supervised training. Differences in total, regional brain (cortical gray matter, white matter, deep gray matter, cerebellum, and brainstem) and brain:body volumes were compared between groups. STATISTICAL TESTS: ANOVA was used to test for differences in brain volumes between groups, after accounting for sex and gestational age at scan. PFDR -values <0.05 were considered statistically significant. RESULTS: Total and regional brain volumes were smaller in fetuses where cerebral substrate delivery is reduced. No significant differences were observed in total or regional brain volumes between control fetuses and fetuses with CHD but normal cerebral substrate delivery (all PFDR > 0.12). Severely reduced cerebral substrate delivery is associated with lower brain:body volume ratios. DATA CONCLUSION: Total and regional brain volumes are smaller in fetuses with CHD where there is a reduction in cerebral substrate delivery, but not in those where cerebral substrate delivery is expected to be normal. EVIDENCE LEVEL: 3 TECHNICAL EFFICACY: Stage 3.

Prenatal diagnosis of hypoplastic left heart syndrome on ultrasound using artificial intelligence: How does performance compare to a current screening programme?

BACKGROUND: Artificial intelligence (AI) has the potential to improve prenatal detection of congenital heart disease. We analysed the performance of the current national screening programme in detecting hypoplastic left heart syndrome (HLHS) to compare with our own AI model. METHODS: Current screening programme performance was calculated from local and national sources. AI models were trained using four-chamber ultrasound views of the fetal heart, using a ResNet classifier. RESULTS: Estimated current fetal screening programme sensitivity and specificity for HLHS were 94.3% and 99.985%, respectively. Depending on calibration, AI models to detect HLHS were either highly sensitive (sensitivity 100%, specificity 94.0%) or highly specific (sensitivity 93.3%, specificity 100%). Our analysis suggests that our highly sensitive model would generate 45,134 screen positive results for a gain of 14 additional HLHS cases. Our highly specific model would be associated with two fewer detected HLHS cases, and 118 fewer false positives. CONCLUSION: If used independently, our AI model performance is slightly worse than the performance level of the current screening programme in detecting HLHS, and this performance is likely to deteriorate further when used prospectively. This demonstrates that collaboration between humans and AI will be key for effective future clinical use.

Inside the 'imperfect mosaic': Minority ethnic women's qualitative experiences of race and ethnicity during pregnancy, childbirth, and maternity care in the United Kingdom.

BACKGROUND: Persistent, high rates of maternal mortality amongst ethnic minorities is one of the UK's starkest examples of racial disparity. With greater risks of adverse outcomes during maternity care, ethnic minority women are subjected to embedded, structural and systemic discrimination throughout the healthcare service. METHODS: Fourteen semi-structured interviews were undertaken with minority ethnic women who had recent experience of UK maternity care. Data pertaining to ethnicity and race were subject to iterative, inductive coding, and constant comparison through Grounded Theory Analysis to test a previously established theory: The 'Imperfect Mosaic'. ANALYSIS & FINDINGS: A related theory emerged, comprising four themes: 'Stopping Short of Agentic Birth'; 'Silenced and Stigmatised through Tick-Box Care'; 'Anticipating Discrimination and the Need for Advocacy'; and 'Navigating Cultural Differences'. The new theory: Inside the 'Imperfect Mosaic', demonstrates experiences of those who received maternity care which directly mirrors experiences of those who provide care, as seen in the previous theory we set-out to test. However, the current theory is based on more traditional and familiar notions of racial discrimination, rather than the nuanced, subtleties of socio-demographic-based micro-aggressions experienced by healthcare professionals. CONCLUSIONS: Our findings suggest the need for the following actions: Prioritisation of bodily autonomy and agency in perinatal physical and mental healthcare; expand awareness of social and cultural issues (i.e., moral injury; cultural safety) within the NHS; and undertake diversity training and support, and follow-up of translation of the training into practice, across (maternal) health services.

Motion corrected fetal body magnetic resonance imaging provides reliable 3D lung volumes in normal and abnormal fetuses.

OBJECTIVES: To calculate 3D-segmented total lung volume (TLV) in fetuses with thoracic anomalies using deformable slice-to-volume registration (DSVR) with comparison to 2D-manual segmentation. To establish a normogram of TLV calculated by DSVR in healthy control fetuses. METHODS: A pilot study at a single regional fetal medicine referral centre included 16 magnetic resonance imaging (MRI) datasets of fetuses (22-32 weeks gestational age). Diagnosis was CDH (n = 6), CPAM (n = 2), and healthy controls (n = 8). Deformable slice-to-volume registration was used for reconstruction of 3D isotropic (0.85 mm) volumes of the fetal body followed by semi-automated lung segmentation. 3D TLV were compared to traditional 2D-based volumetry. Abnormal cases referenced to a normogram produced from 100 normal fetuses whose TLV was calculated by DSVR only. RESULTS: Deformable slice-to-volume registration-derived TLV values have high correlation with the 2D-based measurements but with a consistently lower volume; bias -1.44 cm3 [95% limits: -2.6 to -0.3] with improved resolution to exclude hilar structures even in cases of motion corruption or very low lung volumes. CONCLUSIONS: Deformable slice-to-volume registration for fetal lung MRI aids analysis of motion corrupted scans and does not suffer from the interpolation error inherent to 2D-segmentation. It increases information content of acquired data in terms of visualising organs in 3D space and quantification of volumes, which may improve counselling and surgical planning.

Exploring a new paradigm for the fetal anomaly ultrasound scan: Artificial intelligence in real time.

OBJECTIVE: Advances in artificial intelligence (AI) have demonstrated potential to improve medical diagnosis. We piloted the end-to-end automation of the mid-trimester screening ultrasound scan using AI-enabled tools. METHODS: A prospective method comparison study was conducted. Participants had both standard and AI-assisted US scans performed. The AI tools automated image acquisition, biometric measurement, and report production. A feedback survey captured the sonographers' perceptions of scanning. RESULTS: Twenty-three subjects were studied. The average time saving per scan was 7.62 min (34.7%) with the AI-assisted method (p < 0.0001). There was no difference in reporting time. There were no clinically significant differences in biometric measurements between the two methods. The AI tools saved a satisfactory view in 93% of the cases (four core views only), and 73% for the full 13 views, compared to 98% for both using the manual scan. Survey responses suggest that the AI tools helped sonographers to concentrate on image interpretation by removing disruptive tasks. CONCLUSION: Separating freehand scanning from image capture and measurement resulted in a faster scan and altered workflow. Removing repetitive tasks may allow more attention to be directed identifying fetal malformation. Further work is required to improve the image plane detection algorithm for use in real time.

MRI-Derived Fetal Weight Estimation in the Midpregnancy Fetus: A Method Comparison Study.

OBJECTIVES: The aim of this study was to compare the standard ultrasound (US) estimated fetal weight (EFW) and MRI volume-derived methods for the midtrimester fetus. METHODS: Twenty-five paired US and MRI scans had the EFW calculated (gestational age [GA] range = 20-26 weeks). The intra- and interobserver variability of each method was assessed (2 operators/modality). A small sub-analysis was performed on 5 fetuses who were delivered preterm (mean GA 29 +3 weeks) and compared to the actual birthweight. RESULTS: Two MRI volumetry EFW formulae under-measured compared to US by -10.9% and -14.5% in the midpregnancy fetus (p < 0.001) but had excellent intra- and interobserver agreement (intraclass correlation coefficient = 0.998 and 0.993). In the preterm fetus, the mean relative difference (MRD) between the MRI volume-derived EFW (MRI-EFW) and actual expected birthweight (at the scan GA) was -13.7% (-159.0 g, 95% CI: -341.7 to 23.7 g) and -17.1% (-204.6 g, 95% CI: -380.4 to -28.8 g), for the 2 MRI formulae. The MRD was smaller for US at 5.3% (69.8 g, 95% CI: -34.3 to 173.9). CONCLUSIONS: MRI-EFW results should be interpreted with caution in midpregnancy. Despite excellent observer agreement with MRI volumetry, refinement of the EFW formula is needed in the second trimester, for the small and for the GA and preterm fetus to compensate for lower fetal densities.

Foetal lung volumes in pregnant women who deliver very preterm: a pilot study.

BACKGROUND: Infants born preterm are at increased risk of pulmonary morbidity. The contribution of antenatal factors to impairments in lung structure/function has not been fully elucidated. This study aimed to compare standardized lung volumes from foetuses that were delivered <32 weeks' gestation with foetuses that were delivered >37 weeks. METHODS: Fourteen women who delivered <32 weeks gestation and 56 women who delivered >37 underwent a foetal MRI. Slice-volume reconstruction was then used and the foetal lungs were then segmented using multi-atlas approaches. Body volumes were calculated by manual segmentation and lung:body volume ratios generated. RESULTS: Mean gestation at MRI of the preterm group was 27+2 weeks (SD 2.9, range 20+6-31+3) and control group 25+3 weeks (SD 4.7 range 20+5-31+6). Mean gestation at delivery of the preterm group was 29+2 weeks (SD 2.6, range 22+0-32+0). Lung:body volume ratios and foetal lung volumes were smaller in foetuses that were delivered preterm both with and without preterm premature rupture of membranes compared to those born at term (p < 0.001 in all cases). CONCLUSIONS: Foetuses that were delivered very preterm had reduced lung volumes when standardized for foetal size, irrespective of ruptured membranes. These are novel findings and suggest an antenatal aetiology of insult and possible focus for future preventative therapies.

Automated body organ segmentation, volumetry and population-averaged atlas for 3D motion-corrected T2-weighted fetal body MRI.

Structural fetal body MRI provides true 3D information required for volumetry of fetal organs. However, current clinical and research practice primarily relies on manual slice-wise segmentation of raw T2-weighted stacks, which is time consuming, subject to inter- and intra-observer bias and affected by motion-corruption. Furthermore, there are no existing standard guidelines defining a universal approach to parcellation of fetal organs. This work produces the first parcellation protocol of the fetal body organs for motion-corrected 3D fetal body MRI. It includes 10 organ ROIs relevant to fetal quantitative volumetry studies. We also introduce the first population-averaged T2w MRI atlas of the fetal body. The protocol was used as a basis for training of a neural network for automated organ segmentation. It showed robust performance for different gestational ages. This solution minimises the need for manual editing and significantly reduces time. The general feasibility of the proposed pipeline was also assessed by analysis of organ growth charts created from automated parcellations of 91 normal control 3T MRI datasets that showed expected increase in volumetry during 22-38 weeks gestational age range.

Factors which influence ethnic minority women's participation in maternity research: A systematic review of quantitative and qualitative studies.

BACKGROUND: Women from Black, Asian and mixed ethnicity backgrounds in the UK experience higher rates of maternal and neonatal mortality and morbidity, and report poorer experiences of maternity care. Research is required to understand how to reduce these disparities, however, it is acknowledged these groups of women are under-represented in clinical research. AIM: To investigate factors which influence participation in maternity research for women from an ethnic minority background. METHODS: A systematic review was conducted to examine influencing factors for research participation. MEDLINE/CINHAL/PsycInfo/EMBASE databases were systematically searched in March 2021 and updated in March 2022. Papers were eligible if they explored maternal research participation and identified a woman's ethnicity in the results. No restrictions were placed on methodology. A convergent integrated approach was used to synthesise findings. FINDINGS: A total of 14 papers met the inclusion criteria. Results were divided into eight overarching themes. A personalised approach to recruitment and incorporating culturally sensitive communication and considerations enhanced research participation. Distrust around sharing data, a perception of risk to research participation, and research lacking in personal relevance adversely affected the decision to participate. Large variation existed in the quality of the studies reviewed. CONCLUSIONS: Consideration of a woman's culture and background in the design and the delivery of a maternity research study may facilitate participation, particularly when sampling from a specific population. Further research, informed by women from ethnic minority backgrounds is warranted to develop women-centred recommendations for conducting inclusive maternity research. Prospero registration: www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42021261686.

Exploring the role of a time-efficient MRI assessment of the placenta and fetal brain in uncomplicated pregnancies and these complicated by placental insufficiency.

INTRODUCTION: The development of placenta and fetal brain are intricately linked. Placental insufficiency is related to poor neonatal outcomes with impacts on neurodevelopment. This study sought to investigate whether simultaneous fast assessment of placental and fetal brain oxygenation using MRI T2* relaxometry can play a complementary role to US and Doppler US. METHODS: This study is a retrospective case-control study with uncomplicated pregnancies (n = 99) and cases with placental insufficiency (PI) (n = 49). Participants underwent placental and fetal brain MRI and contemporaneous ultrasound imaging, resulting in quantitative assessment including a combined MRI score called Cerebro-placental-T2*-Ratio (CPTR). This was assessed in comparison with US-derived Cerebro-Placental-Ratio (CPR), placental histopathology, assessed using the Amsterdam criteria [1], and delivery details. RESULTS: Pplacental and fetal brain T2* decreased with increasing gestational age in both low and high risk pregnancies and were corrected for gestational-age alsosignificantly decreased in PI. Both CPR and CPTR score were significantly correlated with gestational age at delivery for the entire cohort. CPTR was, however, also correlated independently with gestational age at delivery in the PI cohort. It furthermore showed a correlation to birth-weight-centile in healthy controls. DISCUSSION: This study indicates that MR analysis of the placenta and brain may play a complementary role in the investigation of fetal development. The additional correlation to birth-weight-centile in controls may suggest a role in the determination of placental health even in healthy controls. To our knowledge, this is the first study assessing quantitatively both placental and fetal brain development over gestation in a large cohort of low and high risk pregnancies. Future larger prospective studies will include additional cohorts.

Artificial intelligence education for radiographers, an evaluation of a UK postgraduate educational intervention using participatory action research: a pilot study.

BACKGROUND: Artificial intelligence (AI)-enabled applications are increasingly being used in providing healthcare services, such as medical imaging support. Sufficient and appropriate education for medical imaging professionals is required for successful AI adoption. Although, currently, there are AI training programmes for radiologists, formal AI education for radiographers is lacking. Therefore, this study aimed to evaluate and discuss a postgraduate-level module on AI developed in the UK for radiographers. METHODOLOGY: A participatory action research methodology was applied, with participants recruited from the first cohort of students enrolled in this module and faculty members. Data were collected using online, semi-structured, individual interviews and focus group discussions. Textual data were processed using data-driven thematic analysis. RESULTS: Seven students and six faculty members participated in this evaluation. Results can be summarised in the following four themes: a. participants' professional and educational backgrounds influenced their experiences, b. participants found the learning experience meaningful concerning module design, organisation, and pedagogical approaches, c. some module design and delivery aspects were identified as barriers to learning, and d. participants suggested how the ideal AI course could look like based on their experiences. CONCLUSIONS: The findings of our work show that an AI module can assist educators/academics in developing similar AI education provisions for radiographers and other medical imaging and radiation sciences professionals. A blended learning delivery format, combined with customisable and contextualised content, using an interprofessional faculty approach is recommended for future similar courses.

3D T2w fetal body MRI: automated organ volumetry, growth charts and population-averaged atlas.

Structural fetal body MRI provides true 3D information required for volumetry of fetal organs. However, current clinical and research practice primarily relies on manual slice-wise segmentation of raw T2-weighted stacks, which is time consuming, subject to inter- and intra-observer bias and affected by motion-corruption. Furthermore, there are no existing standard guidelines defining a universal approach to parcellation of fetal organs. This work produces the first parcellation protocol of the fetal body organs for motion-corrected 3D fetal body MRI. It includes 10 organ ROIs relevant to fetal quantitative volumetry studies. We also introduce the first population-averaged T2w MRI atlas of the fetal body. The protocol was used as a basis for training of a neural network for automated organ segmentation. It showed robust performance for different gestational ages. This solution minimises the need for manual editing and significantly reduces time. The general feasibility of the proposed pipeline was also assessed by analysis of organ growth charts created from automated parcellations of 91 normal control 3T MRI datasets that showed expected increase in volumetry during 22-38 weeks gestational age range. In addition, the results of comparison between 60 normal and 12 fetal growth restriction datasets revealed significant differences in organ volumes.

Placenta segmentation in ultrasound imaging: Addressing sources of uncertainty and limited field-of-view.

Automatic segmentation of the placenta in fetal ultrasound (US) is challenging due to the (i) high diversity of placenta appearance, (ii) the restricted quality in US resulting in highly variable reference annotations, and (iii) the limited field-of-view of US prohibiting whole placenta assessment at late gestation. In this work, we address these three challenges with a multi-task learning approach that combines the classification of placental location (e.g., anterior, posterior) and semantic placenta segmentation in a single convolutional neural network. Through the classification task the model can learn from larger and more diverse datasets while improving the accuracy of the segmentation task in particular in limited training set conditions. With this approach we investigate the variability in annotations from multiple raters and show that our automatic segmentations (Dice of 0.86 for anterior and 0.83 for posterior placentas) achieve human-level performance as compared to intra- and inter-observer variability. Lastly, our approach can deliver whole placenta segmentation using a multi-view US acquisition pipeline consisting of three stages: multi-probe image acquisition, image fusion and image segmentation. This results in high quality segmentation of larger structures such as the placenta in US with reduced image artifacts which are beyond the field-of-view of single probes.

Fast fetal head compounding from multi-view 3D ultrasound.

The diagnostic value of ultrasound images may be limited by the presence of artefacts, notably acoustic shadows, lack of contrast and localised signal dropout. Some of these artefacts are dependent on probe orientation and scan technique, with each image giving a distinct, partial view of the imaged anatomy. In this work, we propose a novel method to fuse the partially imaged fetal head anatomy, acquired from numerous views, into a single coherent 3D volume of the full anatomy. Firstly, a stream of freehand 3D US images is acquired using a single probe, capturing as many different views of the head as possible. The imaged anatomy at each time-point is then independently aligned to a canonical pose using a recurrent spatial transformer network, making our approach robust to fast fetal and probe motion. Secondly, images are fused by averaging only the most consistent and salient features from all images, producing a more detailed compounding, while minimising artefacts. We evaluated our method quantitatively and qualitatively, using image quality metrics and expert ratings, yielding state of the art performance in terms of image quality and robustness to misalignments. Being online, fast and fully automated, our method shows promise for clinical use and deployment as a real-time tool in the fetal screening clinic, where it may enable unparallelled insight into the shape and structure of the face, skull and brain.

Fetal magnetic resonance imaging (MRI) enhances the diagnosis of congenital body anomalies.

AIMS: We sought to assess variability and concordance between fetal MRI and ultrasound (USS) in the evaluation of fetal body abnormalities. METHODS: All fetal body anomalies reported on F-MRI within the iFIND database (http://www.ifindproject.com) were included. Differences in findings regarding anomalies on contemporaneous USS were explored. Three clinical specialists evaluated each case independently, and the anomaly severity was graded: as "insignificant" to "lethal". The value of MRI in alteration of either antenatal or postnatal care was established. RESULTS: Fifty-four cases were identified consisting of 5 healthy controls, 37 with USS-identified body anomalies, and 12 with known CNS or cardiac anomalies. In fetuses with a known body anomaly, information on the MRI was relevant to change the clinical course in 59% of cases. There was also an incidental detection rate of 7% in fetuses with known cardiac or CNS anomalies, or 1.5% of normal control, although these were rarely clinically relevant. Importantly, fetuses undergoing MRI for cardiac concerns did have major anomalies that were missed (one case of oesophageal atresia and two cases of ARM). CONCLUSIONS: In cases where fetal anomalies are suspected, F-MRI is a valuable means of further characterizing anomalies and may detect additional anomalies in fetuses with recognized cardiac or CNS anomalies. In fetuses with a recognized body anomaly, more than half of those scanned by MRI had information available which changed clinical management. Importantly there were also incidental findings in healthy control fetuses, so the management of these needs to be recognized in fetal MRI research. LEVEL OF EVIDENCE: II, Prospective cohort study.

An insight into the current perceptions of UK radiographers on the future impact of AI on the profession: A cross-sectional survey.

INTRODUCTION: As a profession, radiographers have always been keen on adapting and integrating new technologies. The increasing integration of artificial intelligence (AI) into clinical practice in the last five years has been met with scepticism by some, who predict the demise of the profession, whilst others suggest a bright future with AI, full of opportunities and synergies. Post COVID-19 pandemic need for economic recovery and a backlog of medical imaging and reporting may accelerate the adoption of AI. It is therefore timely to appreciate practitioners' perceptions of AI used in clinical practice and their perception of the short-term impact on the profession. AIM: This study aims to explore the perceptions of AI in the UK radiography workforce and to investigate its current AI applications and future technological expectations of radiographers. METHODS: An online survey (QualtricsⓇ) was created by a team of radiography AI experts. The survey was disseminated via social media and professional networks in the UK. Demographic information and perceptions of the impact of AI on several aspects of the radiography profession were gathered, including the current use of AI in practice, future expectations and the perceived impact of AI on the profession. RESULTS: 411 responses were collected (80% diagnostic radiographers (DR); 20% therapeutic radiographers (TR)). Awareness of AI used in clinical practice is low, with DR respondents suggesting AI will have the most value/potential in cross sectional imaging and image reporting. TR responses linked AI as having most value in treatment planning, contouring, and image acquisition/matching. Respondents felt that AI will impact radiographers' daily work (DR, 79.6%; TR, 88.9%) by standardising some aspects of patient care and technical factors of radiography practice. A mixed response about impact on careers was reported. CONCLUSIONS: Respondents were unsure about the ways in which AI is currently used in practice and how AI will impact on careers in the future. It was felt that AI integration will lead to increased job opportunities to contribute to decision making as an end user. Job security was not identified as a cause for concern.

Assessment of the fetal thymus gland: Comparing MRI-acquired thymus volumes with 2D ultrasound measurements.

OBJECTIVES: The fetal thymus gland has been shown to involute in response to intrauterine infection, and therefore could be used as a non-invasive marker of fetal compartment infection. The objective of this study was to evaluate how accurately 2D ultrasound-derived measurements of the fetal thymus reflect the 3D volume of the gland derived from motion corrected MRI images. STUDY DESIGN: A retrospective study was performed using paired ultrasound and MRI datasets from the iFIND project (http://www.ifindproject.com). To obtain 3D volumetry of the thymus gland, T2-weighted single shot turbo spin echo (ssTSE) sequences of the fetal thorax were acquired. Thymus volumes were manually segmented from deformable slice-to-volume reconstructed images. To obtain 2D ultrasound measurements, previously stored fetal cine loops were used and measurements obtained at the 3-vessel-view (3VV) and 3-vessel-trachea view (3VT): anterior-posterior diameter (APD), intrathoracic diameter (ITD), transverse diameter (TD), perimeter and 3-vessel-edge (3VE). Inter-observer and intra-observer reliability (ICC) was calculated for both MRI and ultrasound measurements. Pearson correlation coefficients (PCC) were used to compare 2D-parameters with acceptable ICC to TV. RESULTS: 38 participants were identified. Adequate visualisation was possible on 37 MRI scans and 31 ultrasound scans. Of the 30 datasets where both MRI and ultrasound data were available, MRI had good interobserver reliability (ICC 0.964) and all ultrasound 3VV 2D-parameters and 3VT 3VE had acceptable ICC (>0.75). Four 2D parameters were reflective of the 3D thymus volume: 3VV TD r = 0.540 (P = 0.002); 3VV perimeter r = 0.446 (P = 0.013); 3VV APD r = 0.435 (P = 0.110) and 3VT TD r = 0.544 (P = 0.002). CONCLUSIONS: MRI appeared superior to ultrasound for visualization of the thymus gland and reproducibility of measurements. Three 2D US parameters, 3VV TD, perimeter and 3VT APD, correlated well with TV. Therefore, these represent a more accurate reflection of the true size of the gland than other 2D measurements, where MRI is not available.

Towards Standardized Acquisition with a Dual-probe Ultrasound Robot for Fetal Imaging.

Standardized acquisitions and diagnoses using robots and AI would potentially increase the general usability and reliability of medical ultrasound. Working towards this prospect, this paper presents the recent developments of a standardized acquisition workflow using a novel dual-probe ultrasound robot, for a project known as intelligent Fetal Imaging and Diagnosis (iFIND). The workflow includes an abdominal surface mapping step to obtain a non-parametric spline surface, a rule-based end-point calculation method to position each individual joint, and a motor synchronization method to achieve a smooth motion towards a target point. The design and implementation of the robot are first presented in this paper and the proposed workflow is then explained in detail with simulation and volunteer experiments performed and analyzed. The closed-form analytical solution to the specific motion planning problem has demonstrated a reliable performance controlling the robot to move towards the expected scanning areas and the calculated proximity of the robot to the surface shows that the robot maintains a safe distance while moving around the abdomen. The volunteer study has successfully demonstrated the reliable working and controllability of the robot in terms of acquiring desired ultrasound views. Our future work will focus on improving the motion planning, and on integrating the proposed standardized acquisition workflow with newly- developed ultrasound image processing methods to obtain diagnostic results in an accurate and consistent way.

Mutual Information-Based Disentangled Neural Networks for Classifying Unseen Categories in Different Domains: Application to Fetal Ultrasound Imaging.

Deep neural networks exhibit limited generalizability across images with different entangled domain features and categorical features. Learning generalizable features that can form universal categorical decision boundaries across domains is an interesting and difficult challenge. This problem occurs frequently in medical imaging applications when attempts are made to deploy and improve deep learning models across different image acquisition devices, across acquisition parameters or if some classes are unavailable in new training databases. To address this problem, we propose Mutual Information-based Disentangled Neural Networks (MIDNet), which extract generalizable categorical features to transfer knowledge to unseen categories in a target domain. The proposed MIDNet adopts a semi-supervised learning paradigm to alleviate the dependency on labeled data. This is important for real-world applications where data annotation is time-consuming, costly and requires training and expertise. We extensively evaluate the proposed method on fetal ultrasound datasets for two different image classification tasks where domain features are respectively defined by shadow artifacts and image acquisition devices. Experimental results show that the proposed method outperforms the state-of-the-art on the classification of unseen categories in a target domain with sparsely labeled training data.