Auto-segmentation of thoracic brachial plexuses for radiation therapy planning.

Recently, deep learning networks have achieved considerable success in segmenting organs in medical images. Several methods have used volumetric information with deep networks to achieve segmentation accuracy. However, these networks suffer from interference, risk of overfitting, and low accuracy as a result of artifacts, in the case of very challenging objects like the brachial plexuses. In this paper, to address these issues, we synergize the strengths of high-level human knowledge (i.e., natural intelligence (NI)) with deep learning (i.e., artificial intelligence (AI)) for recognition and delineation of the thoracic brachial plexuses (BPs) in computed tomography (CT) images. We formulate an anatomy-guided deep learning hybrid intelligence approach for segmenting thoracic right and left brachial plexuses consisting of 2 key stages. In the first stage (AAR-R), objects are recognized based on a previously created fuzzy anatomy model of the body region with its key organs relevant for the task at hand wherein high-level human anatomic knowledge is precisely codified. The second stage (DL-D) uses information from AAR-R to limit the search region to just where each object is most likely to reside and performs encoder-decoder delineation in slices. The proposed method is tested on a dataset that consists of 125 images of the thorax acquired for radiation therapy planning of tumors in the thorax and achieves a Dice coefficient of 0.659.

Project ECHO telementoring program in epilepsy for school nurses.

OBJECTIVES: Many children with epilepsy experience seizures at school. School nurses must have the clinical expertise to deliver high-quality, safe care for students with epilepsy. However, in some regions of the U.S. access to interactive, epilepsy evidence-based education programs is limited. The objective of this project was to assess the feasibility of adapting the Epilepsy Foundation's (EFs) school nurse education program to the ECHO model and evaluate its impact on school nurse knowledge and self-efficacy in managing epilepsy in students with seizures and program satisfaction. METHODS: The EFs educational program for school nurses was adapted to the ECHO model and delivered by a team of interdisciplinary epilepsy specialists via videoconferencing. Retrospective post-program surveys were administered at program completion. Data from 32 participants with complete post-program surveys were used for the analysis of knowledge and confidence. Descriptive statistics and the sign test were conducted. RESULTS: Participants were 166 school nurses from 13 states. The majority had > 15 years of school nurse experience and served schools in suburban or rural areas. Improvements in knowledge and confidence were reported on most survey items. The highest improvements in self-reported knowledge and confidence were in psychosocial aspects of care, comorbidities, and recognition of nonepileptic events. Program satisfaction was rated as high by over 90% of participants. CONCLUSIONS: Telementoring using the ECHO methodology is a feasible modality to educate and link epilepsy specialists and providers with school nurses nationwide. Findings suggest that attending the MSS ECHO provided an educational and meaningful learning experience. The gains in knowledge and confidence in psychosocial aspects of epilepsy care and comorbidities highlight the importance of the inclusion of this content in educational programs.

Object recognition in medical images via anatomy-guided deep learning.

PURPOSE: Despite advances in deep learning, robust medical image segmentation in the presence of artifacts, pathology, and other imaging shortcomings has remained a challenge. In this paper, we demonstrate that by synergistically marrying the unmatched strengths of high-level human knowledge (i.e., natural intelligence (NI)) with the capabilities of deep learning (DL) networks (i.e., artificial intelligence (AI)) in garnering intricate details, these challenges can be significantly overcome. Focusing on the object recognition task, we formulate an anatomy-guided deep learning object recognition approach named AAR-DL which combines an advanced anatomy-modeling strategy, model-based non-deep-learning object recognition, and deep learning object detection networks to achieve expert human-like performance. METHODS: The AAR-DL approach consists of 4 key modules wherein prior knowledge (NI) is made use of judiciously at every stage. In the first module AAR-R, objects are recognized based on a previously created fuzzy anatomy model of the body region with all its organs following the automatic anatomy recognition (AAR) approach wherein high-level human anatomic knowledge is precisely codified. This module is purely model-based with no DL involvement. Although the AAR-R operation lacks accuracy, it is robust to artifacts and deviations (much like NI), and provides the much-needed anatomic guidance in the form of rough regions-of-interest (ROIs) for the following DL modules. The 2nd module DL-R makes use of the ROI information to limit the search region to just where each object is most likely to reside and performs DL-based detection of the 2D bounding boxes (BBs) in slices. The 2D BBs hug the shape of the 3D object much better than 3D BBs and their detection is feasible only due to anatomy guidance from AAR-R. In the 3rd module, the AAR model is deformed via the found 2D BBs providing refined model information which now embodies both NI and AI decisions. The refined AAR model more actively guides the 4th refined DL-R module to perform final object detection via DL. Anatomy knowledge is made use of in designing the DL networks wherein spatially sparse objects and non-sparse objects are handled differently to provide the required level of attention for each. RESULTS: Utilizing 150 thoracic and 225 head and neck (H&N) computed tomography (CT) data sets of cancer patients undergoing routine radiation therapy planning, the recognition performance of the AAR-DL approach is evaluated on 10 thoracic and 16 H&N organs in comparison to pure model-based approach (AAR-R) and pure DL approach without anatomy guidance. Recognition accuracy is assessed via location error/ centroid distance error, scale or size error, and wall distance error. The results demonstrate how the errors are gradually and systematically reduced from the 1st module to the 4th module as high-level knowledge is infused via NI at various stages into the processing pipeline. This improvement is especially dramatic for sparse and artifact-prone challenging objects, achieving a location error over all objects of 4.4 mm and 4.3 mm for the two body regions, respectively. The pure DL approach failed on several very challenging sparse objects while AAR-DL achieved accurate recognition, almost matching human performance, showing the importance of anatomy guidance for robust operation. Anatomy guidance also reduces the time required for training DL networks considerably. CONCLUSIONS: (i) High-level anatomy guidance improves recognition performance of DL methods. (ii) This improvement is especially noteworthy for spatially sparse, low-contrast, inconspicuous, and artifact-prone objects. (iii) Once anatomy guidance is provided, 3D objects can be detected much more accurately via 2D BBs than 3D BBs and the 2D BBs represent object containment with much more specificity. (iv) Anatomy guidance brings stability and robustness to DL approaches for object localization. (v) The training time can be greatly reduced by making use of anatomy guidance.

Combining natural and artificial intelligence for robust automatic anatomy segmentation: Application in neck and thorax auto-contouring.

BACKGROUND: Automatic segmentation of 3D objects in computed tomography (CT) is challenging. Current methods, based mainly on artificial intelligence (AI) and end-to-end deep learning (DL) networks, are weak in garnering high-level anatomic information, which leads to compromised efficiency and robustness. This can be overcome by incorporating natural intelligence (NI) into AI methods via computational models of human anatomic knowledge. PURPOSE: We formulate a hybrid intelligence (HI) approach that integrates the complementary strengths of NI and AI for organ segmentation in CT images and illustrate performance in the application of radiation therapy (RT) planning via multisite clinical evaluation. METHODS: The system employs five modules: (i) body region recognition, which automatically trims a given image to a precisely defined target body region; (ii) NI-based automatic anatomy recognition object recognition (AAR-R), which performs object recognition in the trimmed image without DL and outputs a localized fuzzy model for each object; (iii) DL-based recognition (DL-R), which refines the coarse recognition results of AAR-R and outputs a stack of 2D bounding boxes (BBs) for each object; (iv) model morphing (MM), which deforms the AAR-R fuzzy model of each object guided by the BBs output by DL-R; and (v) DL-based delineation (DL-D), which employs the object containment information provided by MM to delineate each object. NI from (ii), AI from (i), (iii), and (v), and their combination from (iv) facilitate the HI system. RESULTS: The HI system was tested on 26 organs in neck and thorax body regions on CT images obtained prospectively from 464 patients in a study involving four RT centers. Data sets from one separate independent institution involving 125 patients were employed in training/model building for each of the two body regions, whereas 104 and 110 data sets from the 4 RT centers were utilized for testing on neck and thorax, respectively. In the testing data sets, 83% of the images had limitations such as streak artifacts, poor contrast, shape distortion, pathology, or implants. The contours output by the HI system were compared to contours drawn in clinical practice at the four RT centers by utilizing an independently established ground-truth set of contours as reference. Three sets of measures were employed: accuracy via Dice coefficient (DC) and Hausdorff boundary distance (HD), subjective clinical acceptability via a blinded reader study, and efficiency by measuring human time saved in contouring by the HI system. Overall, the HI system achieved a mean DC of 0.78 and 0.87 and a mean HD of 2.22 and 4.53 mm for neck and thorax, respectively. It significantly outperformed clinical contouring in accuracy and saved overall 70% of human time over clinical contouring time, whereas acceptability scores varied significantly from site to site for both auto-contours and clinically drawn contours. CONCLUSIONS: The HI system is observed to behave like an expert human in robustness in the contouring task but vastly more efficiently. It seems to use NI help where image information alone will not suffice to decide, first for the correct localization of the object and then for the precise delineation of the boundary.

Current state of the union of epilepsy care in the United States: Antiepileptic drugs - An introduction to the Connectors Project.

PURPOSE: How antiepileptic drugs (AEDs) are used in the United States (US) is one proxy public health indicator for the current state of epilepsy management. The use of phenytoin, other older AEDs, and newer AEDs may act as an indicator for the quality of epilepsy practice in addition to the current American Academy of Neurology quality measures. Data on AED used by states and populations can help identify which public health interventions are necessary to improve the status of epilepsy care. The Connectors Project, a collaboration between the Epilepsy Foundation and UCB Pharma, is a multiyear project designed to improve epilepsy awareness and management in underserved communities. The objective of the first phase of the Connectors Project was to assess geographic variation in epilepsy care and identify locations in need of improved epilepsy care by initially evaluating AED use in the US. METHODS: A retrospective cross-sectional administrative claim analysis was conducted using the QuintilesIMS™ database which included US longitudinal retail prescription and office medical claims data. Patients with a confirmed epilepsy diagnosis who were prescribed AEDs were identified. Patients with an AED prescription over a 3-year period from January 2013 to December 2015 were included if they had an epilepsy diagnosis in the 2-year period before their first AED prescription in the reporting period. The percentages of patients initially prescribed phenytoin, other older AEDs (carbamazepine and valproate), and newer AEDs (eslicarbazepine, lacosamide, lamotrigine, levetiracetam, oxcarbazepine, perampanel, topiramate) were calculated and stratified by US state and Washington, DC. Patients were considered newly treated if they had an epilepsy diagnosis code and had not received an epilepsy drug in the 1-year period preceding the first AED prescription in the reporting period. Data are reported using the moving annual total ending December 2015. RESULTS: Approximately 2.5 million US patients with epilepsy and their AED prescriptions were identified from 2013 to 2015. Predictably, states with the largest population had the highest number of patients with epilepsy who were prescribed an AED, including California, Texas, Florida, and New York. Regions with the highest total proportion of phenytoin use with a low proportion of newer AED use were Mississippi (24.4% and 53.1%, respectively) and Washington, DC (24.7% and 58.1%). Montana had the lowest proportion of phenytoin use with the highest proportion of newer AED use (7.9% and 70.4%). Among newly treated patients (N=237,347), Hawaii (39.1%) and Alaska (38.8%) had the highest percentage of phenytoin use compared with all other states. Idaho (86.1%) and Montana (84.4%) had the highest proportion of newer AED use. Washington, DC (50.9%) and Hawaii (60.9%) had the lowest proportion of patients treated with newer AEDs. North Dakota (29.6%) and Washington, DC (27.9%) had the highest rates of other older AEDs use. CONCLUSIONS: A substantial proportion of newly treated US patients with epilepsy are underserved regarding newer AED use with Mississippi and Washington, DC having the highest proportion of phenytoin use relative to newer AED use. Understanding the socioeconomic and demographic barriers for these observations is essential in planning interventions to improve the quality of life and care for patients with epilepsy, including newly treated patients. These data provide a baseline to target educational and clinical interventions for improving the quality of US epilepsy care.

PelvicSim--a computational-experimental system for biomechanical evaluation of female pelvic floor organ disorders and associated minimally invasive interventions.

In this paper we describe the prototype of a new computational simulation system, PelvicSim. This system is being developed to simulate the in vivo biomechanics of the female pelvic floor organ system with the intent to provide clinical researchers, medical device designers with a virtual environment to understand the various biomechanical pathologies occurring in the pelvic floor. This information can then be used to develop new reconstructive surgical techniques, or design non surgical/surgical devices for the treatment of urinary incontinence and pelvic organ prolapse. In this paper, we provide the initial results from the development of the PelvicSim modules which combine in vivo sensing experiments, Ultrasound and MRI imaging datasets, and an inverse finite element modeling technology based on hyperviscoelastic constitutive modeling of the pelvic floor organs and tissues.

Probing the influence of stereoelectronic effects on the biophysical properties of oligonucleotides: comprehensive analysis of the RNA affinity, nuclease resistance, and crystal structure of ten 2'-O-ribonucleic acid modifications.

The syntheses of 10 new RNA 2'-O-modifications, their incorporation into oligonucleotides, and an evaluation of their properties such as RNA affinity and nuclease resistance relevant to antisense activity are presented. All modifications combined with the natural phosphate backbone lead to significant gains in terms of the stability of hybridization to RNA relative to the first-generation DNA phosphorothioates (PS-DNA). The nuclease resistance afforded in particular by the 2'-O-modifications carrying a positive charge surpasses that of PS-DNA. However, small electronegative 2'-O-substituents, while enhancing the RNA affinity, do not sufficiently protect against degradation by nucleases. Similarly, oligonucleotides containing 3'-terminal residues modified with the relatively large 2'-O-[2-(benzyloxy)ethyl] substituent are rapidly degraded by exonucleases, proving wrong the assumption that steric bulk will generally improve protection against nuclease digestion. To analyze the factors that contribute to the enhanced RNA affinity and nuclease resistance we determined crystal structures of self-complementary A-form DNA decamer duplexes containing single 2'-O-modified thymidines per strand. Conformational preorganization of substituents, favorable electrostatic interactions between substituent and sugar-phosphate backbone, and a stable water structure in the vicinity of the 2'-O-modification all appear to contribute to the improved RNA affinity. Close association of positively charged substituents and phosphate groups was observed in the structures with modifications that protect most effectively against nucleases. The promising properties exhibited by some of the analyzed 2'-O-modifications may warrant a more detailed evaluation of their potential for in vivo antisense applications. Chemical modification of RNA can also be expected to significantly improve the efficacy of small interfering RNAs (siRNA). Therefore, the 2'-O-modifications introduced here may benefit the development of RNAi therapeutics.