OBJECTIVE:Low-cost devices have made obstetric sonography possible in settings where it was previously unfeasible, but ensuring quality and consistency at scale remains a challenge. In the present study, we sought to create a tool to reduce substandard fetal biometry measurement while minimizing care disruption. METHODS:We developed a deep learning artificial intelligence (AI) model to estimate gestational age (GA) in the second and third trimester from fly-to cineloops-brief videos acquired during routine ultrasound biometry-and evaluated its performance in comparison to expert sonographer measurement. We then introduced random error into fetal biometry measurements and analyzed the ability of the AI model to flag grossly inaccurate measurements such as those that might be obtained by a novice. RESULTS:The mean absolute error (MAE) of our model (±standard error) was 3.87 ± 0.07 days, compared to 4.80 ± 0.10 days for expert biometry (difference -0.92 days; 95% CI: -1.10 to -0.76). Based on simulated novice biometry with average absolute error of 7.5%, our model reliably detected cases where novice biometry differed from expert biometry by 10 days or more, with an area under the receiver operating characteristics curve of 0.93 (95% CI: 0.92, 0.95), sensitivity of 81.0% (95% CI: 77.9, 83.8), and specificity of 89.9% (95% CI: 88.1, 91.5). These results held across a range of sensitivity analyses, including where the model was provided suboptimal truncated fly-to cineloops. CONCLUSIONS:Our AI model estimated GA more accurately than expert biometry. Because fly-to cineloop videos can be obtained without any change to sonographer workflow, the model represents a no-cost guardrail that could be incorporated into both low-cost and commercial ultrasound devices to prevent reporting of most gross GA estimation errors.

Temporomandibular joint osteoarthritis (TMJ OA) is a prevalent degenerative disease characterized by chronic pain and impaired jaw function. The complexity of TMJ OA has hindered the development of prognostic tools, posing a significant challenge in timely, patient-specific management. Addressing this gap, our research employs a comprehensive, multidimensional approach to advance TMJ OA prognostication. We conducted a prospective study with 106 subjects, 74 of whom were followed up after 2 to 3 y of conservative treatment. Central to our methodology is the development of an innovative, open-source predictive modeling framework, the Ensemble via Hierarchical Predictions through Nested cross-validation tool (EHPN). This framework synergistically integrates 18 feature selection, statistical, and machine learning methods to yield an accuracy of 0.87, with an area under the ROC curve of 0.72 and an F1 score of 0.82. Our study, beyond technical advancements, emphasizes the global impact of TMJ OA, recognizing its unique demographic occurrence. We highlight key factors influencing TMJ OA progression. Using SHAP analysis, we identified personalized prognostic predictors: lower values of headache, lower back pain, restless sleep, condyle high gray level-GL-run emphasis, articular fossa GL nonuniformity, and long-run low GL emphasis; and higher values of superior joint space, mouth opening, saliva Vascular-endothelium-growth-factor, Matrix-metalloproteinase-7, serum Epithelial-neutrophil-activating-peptide, and age indicate recovery likelihood. Our multidimensional and multimodal EHPN tool enhances clinicians' decision-making, offering a transformative translational infrastructure. The EHPN model stands as a significant contribution to precision medicine, offering a paradigm shift in the management of temporomandibular disorders and potentially influencing broader applications in personalized healthcare.

ObjectiveTo evaluate the accuracy of two portable ultrasound machines (PUM) in assessing fetal biometry and estimated gestational age (EGA).MethodsThis was a secondary analysis of data from the Fetal Age Machine Learning Initiative, an observational study of pregnant women in the USA and Zambia. Each participant underwent ultrasound assessment by an experienced sonographer using both a high-specification ultrasound machine (HSUM) and a PUM (Butterfly iQ or Clarius C3) to measure fetal biometry and calculate EGA at each visit. By comparing paired PUM and HSUM scans, we estimated agreement between individual biometry measurements and aggregate gestational age estimates by reporting mean difference, intraclass correlation coefficient (ICC) and Bland-Altman plots, adjusting for trend.ResultsBetween April and December 2021, 818 participants contributed 1386 paired PUM-HSUM ultrasound investigations, of which 991 PUM scans were obtained using the Butterfly iQ device and 395 using the Clarius C3 device. Gestational age at scan ranged from 7 to 38 weeks. Compared with HSUM, the Butterfly iQ PUM had a mean difference of -0.20 (95% CI, -0.60 to 0.20) days in the first trimester and -0.68 (95% CI, -0.93 to -0.44) days in the second/third trimesters. Compared with HSUM, the Clarius C3 PUM had a mean difference of -0.47 (95% CI, -1.11 to 0.18) days in the first trimester and -1.67 (95% CI, -2.10 to -1.25) days in the second/third trimesters. ICCs were 0.989 or greater throughout. Increasing gestational age was associated with increasing error and absolute error in EGA and fetal biometry. Both PUM devices demonstrated a modest trend toward underestimation of EGA with advancing gestational age in second/third-trimester scans, compared with HSUM.ConclusionThe Butterfly iQ and Clarius C3 PUM devices were highly accurate in performing fetal biometry in a diverse population from the USA and Zambia. (c) 2023 International Society of Ultrasound in Obstetrics and Gynecology.

This paper describes SimNorth, an unsupervised learning approach for classifying non-standard fetal ultrasound images. SimNorth utilizes a deep feature learning model with a novel contrastive loss function to project images with similar characteristics closer together in an embedding space while pushing apart those with different image features. We then use non-linear dimensionality reduction via t-SNE and apply standard clustering algorithms such as k-means and dbscan in 2D embedding space to identify clusters containing similar fetal structures. We compare SimNorth to other unsupervised learning techniques (such as Autoencoders, MoCo, and SimCLR) and demonstrate its superior performance based on cluster purity measures.

ABSTRACTObjective This manuscript describes strategies for assessment of precision of several diagnostic artificial intelligence (AI) tools in orthodontics, available open-source image analysis platforms, as well as the use of three-dimensional (3D) surface models and superimpositions.Results The advances described in this manuscript present perspectives on the controversies of whether AI is smarter than clinicians and may replace human clinical decisions. A thorough orthodontic diagnosis requires comprehensive 3D analysis of the interrelationships among the dentition, craniofacial skeleton and soft tissues. Forecasts have indicated that 3D printing technology will provide more than 60% of all dental treatment needs by 2025, and orthodontic companies as well as remote monitoring companies are already using AI technology, being it essential that the clinicians are prepared and knowledgeable with the technology advances now available.Conclusions The AI applications in orthodontics rely on the implementation into diagnostic image records, data analysis for clinical practice and research applications. Continuous training and validation of the AI orthodontic image tools are essential for improving the performance and generalizability of these methods.

ABSTRACTObjectivesTo describe differences in outcomes in pregnancies complicated by polyhydramnios based on whether the diagnosis was made by maximum vertical pocket (MVP), amniotic fluid index (AFI) or both.MethodsThis was a retrospective cohort study examining ultrasound assessment of amniotic fluid in singleton pregnancies, June 2014 to May 2021, with amniotic fluid volume measured at ≥ 20 weeks gestation. The proportion of pregnancies with mild, moderate or severe polyhydramnios diagnosed by (1) MVP, (2) AFI and (3) both MVP and AFI was evaluated. Modified Poisson regression models estimated the relative risk of adverse outcomes for pregnancies with polyhydramnios compared to those with normal fluid. All models were adjusted for potential confounders and analyses stratified by the presence or absence of foetal anomalies.ResultsOf 14 883 pregnancies, 13 557 (91.1%) had both normal AFI and MVP. Polyhydramnios was most frequently diagnosed by MVP (n = 602/1326, 45.5%). All cases diagnosed by either MVP or AFI were mild. Those with polyhydramnios by both MVP and AFI had an increased risk of perinatal mortality (adjusted relative risk [aRR] = 5.94, 95% confidence interval [95% CI] 3.07−11.50), including IUFD (aRR = 5.58, 95% CI 2.81−11.09) and neonatal death (aRR = 13.07, 95% CI 1.72−99.60). Findings were similar when the analysis was stratified by the presence or absence of foetal anomalies.ConclusionsThe use of MVP was associated with a higher likelihood of polyhydramnios diagnosis versus AFI. Polyhydramnios, diagnosed by either MVP or AFI, was mild. Polyhydramnios diagnosed by both measures was associated with an increased risk of perinatal mortality.

BACKGROUND Ultrasound is indispensable to gestational age estimation and thus to quality obstetrical care, yet high equipment cost and the need for trained sonographers limit its use in low-resource settings. METHODS From September 2018 through June 2021, we recruited 4695 pregnant volunteers in North Carolina and Zambia and obtained blind ultrasound sweeps (cineloop videos) of the gravid abdomen alongside standard fetal biometry. We trained a neural network to estimate gestational age from the sweeps and, in three test data sets, assessed the performance of the artificial intelligence (AI) model and biometry against previously established gestational age. RESULTS In our main test set, the mean absolute error (MAE) (±SE) was 3.9±0.12 days for the model versus 4.7±0.15 days for biometry (difference, -0.8 days; 95% confidence interval [CI], -1.1 to -0.5; P<0.001). The results were similar in North Carolina (difference, -0.6 days; 95% CI, -0.9 to -0.2) and Zambia (-1.0 days; 95% CI, -1.5 to -0.5). Findings were supported in the test set of women who conceived by in vitro fertilization (MAE of 2.8±0.28 vs. 3.6±0.53 days for the model vs. biometry; difference, -0.8 days; 95% CI, -1.7 to 0.2) and in the set of women from whom sweeps were collected by untrained users with low-cost, battery-powered devices (MAE of 4.9±0.29 vs. 5.4±0.28 days for the model vs. biometry; difference, -0.6; 95% CI, -1.3 to 0.1). CONCLUSIONS When provided blindly obtained ultrasound sweeps of the gravid abdomen, our AI model estimated gestational age with accuracy similar to that of trained sonographers conducting standard fetal biometry. Model performance appears to extend to blind sweeps collected by untrained providers in Zambia using low-cost devices. (Funded by the Bill and Melinda Gates Foundation.).

Osteoarthritis is a chronic disease that affects the temporomandibular joint (TMJ), causing chronic pain and disability. To diagnose patients suffering from this disease before advanced degradation of the bone, we developed a diagnostic tool called TMJOAI. This machine learning based algorithm is capable of classifying the health status TMJ in of patients using 52 clinical, biological and jaw condyle radiomic markers. The TMJOAI includes three parts. the feature preparation, selection and model evaluation. Feature generation includes the choice of radiomic features (condylar trabecular bone or mandibular fossa), the histogram matching of the images prior to the extraction of the radiomic markers, the generation of feature pairwise interaction, etc.; the feature selection are based on the p-values or AUCs of single features using the training data; the model evaluation compares multiple machine learning algorithms (e.g. regression-based, tree-based and boosting algorithms) from 10 times 5-fold cross validation. The best performance was achieved with averaging the predictions of XGBoost and LightGBM models; and the inclusion of 32 additional markers from the mandibular fossa of the joint improved the AUC prediction performance from 0.83 to 0.88. After cross-validation and testing, the tools presented here have been deployed on an open-source, web-based system, making it accessible to clinicians. TMJOAI allows users to add data and automatically train and update the machine learning models, and therefore improve their performance.

Chlamydia trachomatous is an infectious ocular condition that can cause the eyelid to turn inward so that one or more eyelashes touch the eyeball, a condition call trachomatous trichiasis (TT), which can lead to blindness. Community-based screeners are used in rural areas to identify patients with TT, who can then be referred for proper medical care. Having automatic methods to detect TT will reduce the amount of time required to train screeners and improve accuracy of detection. This paper proposes a method to automatically identify regions of an eye and identify TT, using photographs taken with smartphones in the field. The attention-based gated deep learning networks in combination with a regionidentification network can identify TT with an accuracy of 91%, sensitivity of 92% and specificity of 87%, showing that these methods have the potential to be deployed in the field.

ABSTRACT Background Ultrasound is indispensable to gestational age estimation, and thus to quality obstetric care, yet high equipment cost and need for trained sonographers limit its use in low-resource settings. Methods From September 2018 through June 2021, we recruited 4,695 pregnant volunteers in North Carolina and Zambia and obtained blind ultrasound sweeps (cineloops) of the gravid abdomen alongside standard fetal biometry. We trained a neural network to estimate gestational age from the sweeps and, in three test sets, assessed performance of the model and biometry against previously established gestational age. Results In our main test set, model mean absolute error (MAE) was 3.9 days (standard error [SE] 0.12) vs. 4.7 days (SE 0.15) for biometry (difference -0.8 days; 95% CI -1.1, -0.5; p<0.001). Results were similar in North Carolina (difference -0.6 days, 95% CI -0.9, -0.2) and Zambia (−1.0 days, 95% CI -1.5, -0.5). Findings were supported in the test set of women who conceived by in vitro fertilization (model MAE 2.8 days [SE 0.28] vs. 3.6 days [SE 0.53] for biometry; difference -0.8 days, 95% CI -1.7, 0.2), and in the set of women from whom sweeps were collected by untrained users with low-cost, battery-powered devices (model MAE 4.9 days [SE 0.29] vs. 5.4 days [SE 0.28] for biometry; difference -0.6, 95% CI -1.3, 0.1). Conclusions Our model estimated gestational age more accurately from blindly obtained ultrasound sweeps than did trained sonographers performing fetal biometry. These results presage a future where all pregnant people – not just those in rich countries – can access the diagnostic benefits of sonography.

Accurate assessment of fetal gestational age (GA) is critical to the clinical management of pregnancy. Industrialized countries rely upon obstetric ultrasound (US) to make this estimate. In low- and middle- income countries, automatic measurement of fetal structures using a low-cost obstetric US may assist in establishing GA without the need for skilled sonographers. In this report, we leverage a large database of obstetric US images acquired, stored and annotated by expert sonographers to train algorithms to classify, segment, and measure several fetal structures: biparietal diameter (BPD), head circumference (HC), crown rump length (CRL), abdominal circumference (AC), and femur length (FL). We present a technique for generating raw images suitable for model training by removing caliper and text annotation and describe a fully automated pipeline for image classification, segmentation, and structure measurement to estimate the GA. The resulting framework achieves an average accuracy of 93% in classification tasks, a mean Intersection over Union accuracy of 0.91 during segmentation tasks, and a mean measurement error of 1.89 centimeters, finally leading to a 1.4 day mean average error in the predicted GA compared to expert sonographer GA estimate using the Hadlock equation.

Studies show that cracked teeth are the third most common cause for tooth loss in industrialized countries. If detected early and accurately, patients can retain their teeth for a longer time. Most cracks are not detected early because of the discontinuous symptoms and lack of good diagnostic tools. Currently used imaging modalities like Cone Beam Computed Tomography (CBCT) and intraoral radiography often have low sensitivity and do not show cracks clearly. This paper introduces a novel method that can detect, quantify, and localize cracks automatically in high resolution CBCT (hr-CBCT) scans of teeth using steerable wavelets and learning methods. These initial results were created using hr-CBCT scans of a set of healthy teeth and of teeth with simulated longitudinal cracks. The cracks were simulated using multiple orientations. The crack detection was trained on the most significant wavelet coefficients at each scale using a bagged classifier of Support Vector Machines. Our results show high discriminative specificity and sensitivity of this method. The framework aims to be automatic, reproducible, and open-source. Future work will focus on the clinical validation of the proposed techniques on different types of cracks ex-vivo. We believe that this work will ultimately lead to improved tracking and detection of cracks allowing for longer lasting healthy teeth.

SlicerSALT is an open-source platform for disseminating state-of-the-art methods for performing statistical shape analysis. These methods are developed as 3D Slicer extensions to take advantage of its powerful underlying libraries. SlicerSALT itself is a heavily customized 3D Slicer package that is designed to be easy to use for shape analysis researchers. The packaged methods include powerful techniques for creating and visualizing shape representations as well as performing various types of analysis.

The outcome of cranial vault reconstruction for the surgical treatment of craniosynostosis heavily depends on the surgeon’s expertise because of the lack of an objective target shape. We introduce a surface-based diffeomorphic registration framework to create the optimal post-surgical cranial shape during craniosynostosis treatment. Our framework estimates and labels where each bone piece needs to be cut using a reference template. Then, it calculates how much each bone piece needs to be translated and in which direction, using the closest normal shape from a multi-atlas as a reference. With our locally affine approach, the method also allows for bone bending, modeling independently the transformation of each bone piece while ensuring the consistency of the global transformation. We evaluated the optimal plan for 15 patients with metopic craniosynostosis. Our results showed that the automated surgical planning creates cranial shapes with a reduction in cranial malformations of 51.43% and curvature discrepancies of 35.09%, which are the two indices proposed in the literature to quantify cranial deformities objectively. In addition, the cranial shapes created were within healthy ranges.

Craniosynostosis is a congenital malformation of the infant skull typically treated via corrective surgery. To accurately quantify the extent of deformation and identify the optimal correction strategy, the patient-specific skull model extracted from a pre-surgical computed tomography (CT) image needs to be registered to an atlas of head CT images representative of normal subjects. Here, the authors present a robust multi-stage, multi-resolution registration pipeline to map a patient-specific CT image to the atlas space of normal CT images. The proposed registration pipeline first performs an initial optimisation at very low resolution to yield a good initial alignment that is subsequently refined at high resolution. They demonstrate the robustness of the proposed method by evaluating its performance on 560 head CT images of 320 normal subjects and 240 craniosynostosis patients and show a success rate of 92.8 and 94.2%, respectively. Their method achieved a mean surface-to-surface distance between the patient and template skull of <2.5 mm in the targeted skull region across both the normal subjects and patients.

Classification procedure aims at finding regions ofthe classes in the feature space. There are several algorithms proposed with supervised and unsupervised strategies for classification in literature. This paper goes on to propose a supervised method of classification using Information Slicing. Information lies in the feature space of the data to be classified. Training stage consists of slicing of the information by continuous partitioning of the feature space to find pure regions for classes from training data set. Classification then becomes a find-and-assign problem in the feature space for the input data. It has been shown here that this method of classification works well on data which have highly overlapping regions amongst the classes. This classification method is further applied for object classification in satellite imagery, where objects are homogeneous regions in images of considerably high resolution. Identification of homogeneous regions is done by a graph based segmentation algorithm which searches for segments having high intra region pixel similarity and high inter-region pixel dissimilarity. Feature vectors for each of these objects are calculated and then these feature vectors are given as an input to the Information Slicing classifier. Ground truth however is needed for the training stage of the classifier which has been generated here by manual intervention. Experiments have been done on four classes and results show that the method of information slicing works well on selective classes.

Classification methods in general use a linear or piecewise linear boundary to define the separation between the classes in the feature space. This results into some error when the classes are not linearly separable from each other. In fact when one class is surrounded by another class, things become more complicated. Instead, if a non-linear boundary between the classes is created or areas where the class lie are marked, then classification becomes an easier task. The paper here deals with an approach of feature space partitioning wherein the feature space is partitioned continuously till some predefined resolution. At the end of the procedure, an area which a particular class occupies is defined and these areas are further utilized for assigning incoming data points to respective classes.