Scalable Transdiagnostic Early Assessment of Mental Health (STREAM): a study protocol.

INTRODUCTION: Early childhood development forms the foundations for functioning later in life. Thus, accurate monitoring of developmental trajectories is critical. However, such monitoring often relies on time-intensive assessments which necessitate administration by skilled professionals. This difficulty is exacerbated in low-resource settings where such professionals are predominantly concentrated in urban and often private clinics, making them inaccessible to many. This geographic and economic inaccessibility contributes to a significant 'detection gap' where many children who might benefit from support remain undetected. The Scalable Transdiagnostic Early Assessment of Mental Health (STREAM) project aims to bridge this gap by developing an open-source, scalable, tablet-based platform administered by non-specialist workers to assess motor, social and cognitive developmental status. The goal is to deploy STREAM through public health initiatives, maximising opportunities for effective early interventions. METHODS AND ANALYSIS: The STREAM project will enrol and assess 4000 children aged 0-6 years from Malawi (n=2000) and India (n=2000). It integrates three established developmental assessment tools measuring motor, social and cognitive functioning using gamified tasks, observation checklists, parent-report and audio-video recordings. Domain scores for motor, social and cognitive functioning will be developed and assessed for their validity and reliability. These domain scores will then be used to construct age-adjusted developmental reference curves. ETHICS AND DISSEMINATION: Ethical approval has been obtained from local review boards at each site (India: Sangath Institutional Review Board; All India Institute of Medical Science (AIIMS) Ethics Committee; Indian Council of Medical Research-Health Ministry Screening Committee; Malawi: College of Medicine Research and Ethics Committee; Malawi Ministry of Health-Blantyre District Health Office). The study adheres to Good Clinical Practice standards and the ethical guidelines of the 6th (2008) Declaration of Helsinki. Findings from STREAM will be disseminated to participating families, healthcare professionals, policymakers, educators and researchers, at local, national and international levels through meetings, academic journals and conferences.

Using mobile health technology to assess childhood autism in low-resource community settings in India: An innovation to address the detection gap.

LAY ABSTRACT: Autism is diagnosed by highly trained professionals- but most autistic people live in parts of the world that harbour few or no such autism specialists and little autism awareness. So many autistic people go undiagnosed, misdiagnosed, and misunderstood. We designed an app (START) to identify autism and related conditions in such places, in an attempt to address this global gap in access to specialists. START uses computerised games and activities for children and a questionnaire for parents to measure social, sensory, and motor skills. To check whether START can flag undiagnosed children likely to have neurodevelopmental conditions, we tested START with children whose diagnoses already were known: Non-specialist health workers with just a high-school education took START to family homes in poor neighbourhoods of Delhi, India to work with 131 two-to-seven-year-olds. Differences between typically and atypically developing children were highlighted in all three types of skills that START assesses: children with neurodevelopmental conditions preferred looking at geometric patterns rather than social scenes, were fascinated by predictable, repetitive sensory stimuli, and had more trouble with precise hand movements. Parents' responses to surveys further distinguished autistic from non-autistic children. An artificial-intelligence technique combining all these measures demonstrated that START can fairly accurately flag atypically developing children. Health workers and families endorsed START as attractive to most children, understandable to health workers, and adaptable within sometimes chaotic home and family environments. This study provides a proof of principle for START in digital screening of autism and related conditions in community settings.

Quantifying preference for social stimuli in young children using two tasks on a mobile platform.

Children typically prefer to attend to social stimuli (e.g. faces, smiles) over non-social stimuli (e.g. natural scene, household objects). This preference for social stimuli is believed to be an essential building block for later social skills and healthy social development. Preference for social stimuli are typically measured using either passive viewing or instrumental choice paradigms, but not both. Since these paradigms likely tap into different mechanisms, the current study addresses this gap by administering both of these paradigms on an overlapping sample. In this study, we use a preferential looking task and an instrumental choice task to measure preference for social stimuli in 3-9 year old typically developing children. Children spent longer looking at social stimuli in the preferential looking task but did not show a similar preference for social rewards on the instrumental choice task. Task performance in these two paradigms were not correlated. Social skills were found to be positively related to the preference for social rewards on the choice task. This study points to putatively different mechanisms underlying the preference for social stimuli, and highlights the importance of choice of paradigms in measuring this construct.

Quantifying local heterogeneity via morphologic scale: Distinguishing tumoral from stromal regions.

INTRODUCTION: The notion of local scale was introduced to characterize varying levels of image detail so that localized image processing tasks could be performed while simultaneously yielding a globally optimal result. In this paper, we have presented the methodological framework for a novel locally adaptive scale definition, morphologic scale (MS), which is different from extant local scale definitions in that it attempts to characterize local heterogeneity as opposed to local homogeneity. METHODS: At every point of interest, the MS is determined as a series of radial paths extending outward in the direction of least resistance, navigating around obstructions. Each pixel can then be directly compared to other points of interest via a rotationally invariant quantitative feature descriptor, determined by the application of Fourier descriptors to the collection of these paths. RESULTS: OUR GOAL IS TO DISTINGUISH TUMOR AND STROMAL TISSUE CLASSES IN THE CONTEXT OF FOUR DIFFERENT DIGITIZED PATHOLOGY DATASETS: prostate tissue microarrays (TMAs) stained with hematoxylin and eosin (HE) (44 images) and TMAs stained with only hematoxylin (H) (44 images), slide mounts of ovarian H (60 images), and HE breast cancer (51 images) histology images. Classification performance over 50 cross-validation runs using a Bayesian classifier produced mean areas under the curve of 0.88 ± 0.01 (prostate HE), 0.87 ± 0.02 (prostate H), 0.88 ± 0.01 (ovarian H), and 0.80 ± 0.01 (breast HE). CONCLUSION: For each dataset listed in Table 3, we randomly selected 100 points per image, and using the procedure described in Experiment 1, we attempted to separate them as belonging to stroma or epithelium.

High-throughput biomarker segmentation on ovarian cancer tissue microarrays via hierarchical normalized cuts.

We present a system for accurately quantifying the presence and extent of stain on account of a vascular biomarker on tissue microarrays. We demonstrate our flexible, robust, accurate, and high-throughput minimally supervised segmentation algorithm, termed hierarchical normalized cuts (HNCuts) for the specific problem of quantifying extent of vascular staining on ovarian cancer tissue microarrays. The high-throughput aspect of HNCut is driven by the use of a hierarchically represented data structure that allows us to merge two powerful image segmentation algorithms-a frequency weighted mean shift and the normalized cuts algorithm. HNCuts rapidly traverses a hierarchical pyramid, generated from the input image at various color resolutions, enabling the rapid analysis of large images (e.g., a 1500 × 1500 sized image under 6 s on a standard 2.8-GHz desktop PC). HNCut is easily generalizable to other problem domains and only requires specification of a few representative pixels (swatch) from the object of interest in order to segment the target class. Across ten runs, the HNCut algorithm was found to have average true positive, false positive, and false negative rates (on a per pixel basis) of 82%, 34%, and 18%, in terms of overlap, when evaluated with respect to a pathologist annotated ground truth of the target region of interest. By comparison, a popular supervised classifier (probabilistic boosting trees) was only able to marginally improve on the true positive and false negative rates (84% and 14%) at the expense of a higher false positive rate (73%), with an additional computation time of 62% compared to HNCut. We also compared our scheme against a k-means clustering approach, which both the HNCut and PBT schemes were able to outperform. Our success in accurately quantifying the extent of vascular stain on ovarian cancer TMAs suggests that HNCut could be a very powerful tool in digital pathology and bioinformatics applications where it could be used to facilitate computer-assisted prognostic predictions of disease outcome.

A high-throughput active contour scheme for segmentation of histopathological imagery.

In this paper a minimally interactive high-throughput system which employs a color gradient based active contour model for rapid and accurate segmentation of multiple target objects on very large images is presented. While geodesic active contours (GAC) have become very popular tools for image segmentation, they tend to be sensitive to model initialization. A second limitation of GAC models is that the edge detector function typically involves use of gray scale gradients; color images usually being converted to gray scale, prior to gradient computation. For color images, however, the gray scale gradient image results in broken edges and weak boundaries, since the other channels are not exploited in the gradient computation. To cope with these limitations, we present a new GAC model that is driven by an accurate and rapid object initialization scheme; hierarchical normalized cuts (HNCut). HNCut draws its strength from the integration of two powerful segmentation strategies-mean shift clustering and normalized cuts. HNCut involves first defining a color swatch (typically a few pixels) from the object of interest. A multi-scale, mean shift coupled normalized cuts algorithm then rapidly yields an initial accurate detection of all objects in the scene corresponding to the colors in the swatch. This detection result provides the initial contour for a GAC model. The edge-detector function of the GAC model employs a local structure tensor based color gradient, obtained by calculating the local min/max variations contributed from each color channel. We show that the color gradient based edge-detector function results in more prominent boundaries compared to the classical gray scale gradient based function. By integrating the HNCut initialization scheme with color gradient based GAC (CGAC), HNCut-CGAC embodies five unique and novel attributes: (1) efficiency in segmenting multiple target structures; (2) the ability to segment multiple objects from very large images; (3) minimal human interaction; (4) accuracy; and (5) reproducibility. A quantitative and qualitative comparison of the HNCut-CGAC model against other state of the art active contour schemes (including a Hybrid Active Contour model (Paragios-Deriche) and a region-based AC model (Rousson-Deriche)), across 196 digitized prostate histopathology images, suggests that HNCut-CGAC is able to outperform state of the art hybrid and region based AC techniques. Our results show that HNCut-CGAC is computationally efficient and may be easily applied to a variety of different problems and applications.

Hierarchical normalized cuts: unsupervised segmentation of vascular biomarkers from ovarian cancer tissue microarrays.

Research has shown that tumor vascular markers (TVMs) may serve as potential OCa biomarkers for prognosis prediction. One such TVM is ESM-1, which can be visualized by staining ovarian Tissue Microarrays (TMA) with an antibody to ESM-1. The ability to quickly and quantitatively estimate vascular stained regions may yield an image based metric linked to disease survival and outcome. Automated segmentation of the vascular stained regions on the TMAs, however, is hindered by the presence of spuriously stained false positive regions. In this paper, we present a general, robust and efficient unsupervised segmentation algorithm, termed Hierarchical Normalized Cuts (HNCut), and show its application in precisely quantifying the presence and extent of a TVM on OCa TMAs. The strength of HNCut is in the use of a hierarchically represented data structure that bridges the mean shift (MS) and the normalized cuts (NCut) algorithms. This allows HNCut to efficiently traverse a pyramid of the input image at various color resolutions, efficiently and accurately segmenting the object class of interest (in this case ESM-1 vascular stained regions) by simply annotating half a dozen pixels belonging to the target class. Quantitative and qualitative analysis of our results, using 100 pathologist annotated samples across multiple studies, prove the superiority of our method (sensitivity 81%, Positive predictive value (PPV), 80%) versus a popular supervised learning technique, Probabilistic Boosting Trees (sensitivity, PPV of 76% and 66%).

Arterial pulse system: modern methods for traditional Indian medicine.

Ayurveda is one of the most comprehensive healing systems in the world and has classified the body system according to the theory of Tridosha to overcome ailments. Diagnosis similar to the traditional pulse-based method requires a system of clean input signals, and extensive experiments for obtaining classification features. In this paper we briefly describe our system of generating pulse waveforms and use various feature detecting methods to show that an arterial pulse contains typical physiological properties. The beat-to-beat variability is captured using a complex B-spline mother wavelet based peak detection algorithm. We also capture--to our knowledge for the first time--the self-similarity in the physiological signal, and quantifiable chaotic behavior using recurrence plot structures.

Nadi Tarangini: a pulse based diagnostic system.

Ayurveda is a traditional medicine and natural healing system in India. Nadi-Nidan (pulse-based diagnosis) is a prominent method in Ayurveda, and is known to dictate all the salient features of a human body. In this paper, we provide details of our procedure for obtaining the complete spectrum of the nadi pulses as a time series. The system Nadi Tarangini1 contains a diaphragm element equipped with strain gauge, a transmitter cum amplifier, and a digitizer for quantifying analog signal. The system acquires the data with 16-bit accuracy with practically no external electronic or interfering noise. Prior systems for obtaining the nadi pulses have been few and far between, when compared to systems such as ECG. The waveforms obtained with our system have been compared with these other similar equipment developed earlier, and is shown to contain more details. The pulse waveform is also shown to have the desirable variations with respect to age of patients, and the pressure applied at the sensing element. The system is being evaluated by Ayurvedic practitioners as a computer-aided diagnostic tool.

Fast image transforms using diophantine methods.

Many image transformations in computer vision and graphics involve a pipeline when an initial integer image is processed with floating point computations for purposes of symbolic information. Traditionally, in the interests of time, the floating point computation is approximated by integer computations where the integerization process requires a guess of an integer. Examples of this phenomenon include the discretization interval of rho and theta in the accumulator array in classical Hough transform, and in geometric manipulation of images (e.g., rotation, where a new grid is overlaid on the image). The result of incorrect discretization is a poor quality visual image, or worse, hampers measurements of critical parameters such as density or length in high fidelity machine vision. Correction techniques include, at best, anti-aliasing methods, or more commonly, a "kludge" to cleanup. In this paper, we present a method that uses the theory of basis reduction in Diophantine approximations; the method outperforms prior integer based computation without sacrificing accuracy (subject to machine epsilon).