Parametric delineation uncertainties contouring (PDUC) modeling on CT scans of prostate cancer patients

PURPOSE

Variability in contouring contributes to large variations in radiation therapy planning and treatment outcomes. The development and testing of tools to automatically detect contouring errors require a source of contours that includes well-understood and realistic errors. The purpose of this work was to develop a simulation algorithm that intentionally injects errors of varying magnitudes into clinically accepted contours and produces realistic contours with different levels of variability.

METHODS

We used a dataset of CT scans from 14 prostate cancer patients with clinician-drawn contours of the regions of interest (ROI) of the prostate, bladder, and rectum. Using our newly developed Parametric Delineation Uncertainties Contouring (PDUC) model, we automatically generated alternative, realistic contours. The PDUC model consists of the contrast-based DU generator and a 3D smoothing layer. The DU generator transforms contours (deformation, contraction, and/or expansion) as a function of image contrast. The generated contours undergo 3D smoothing to obtain a realistic look. After model building, the first batch of auto-generated contours was reviewed. Editing feedback from the reviews was then used in a filtering model for the auto-selection of clinically acceptable (minor-editing) DU contours.

RESULTS

Overall, C values of 5 and 50 consistently produced high proportions of minor-editing contours across all ROI compared to the other C values (0.936 ± $ \pm \;$ 0.111 and 0.552 ± $ \pm \;$ 0.228, respectively). The model performed best on the bladder, which had the highest proportion of minor-editing contours (0.606) of the three ROI. In addition, the classification AUC for the filtering model across all three ROI is 0.724 ± $ \pm \;$ 0.109.

DISCUSSION

The proposed methodology and subsequent results are promising and could have a great impact on treatment planning by generating mathematically simulated alternative structures that are clinically relevant and realistic enough (i.e., similar to clinician-drawn contours) to be used in quality control of radiation therapy.

Water and Meadow Views Both Afford Perceived but Not Performance-Based Attention Restoration: Results From Two Experimental Studies

Attention Restoration Theory proposes that exposure to natural environments helps to restore attention. For sustained attention—the ongoing application of focus to a task, the effect appears to be modest, and the underlying mechanisms of attention restoration remain unclear. Exposure to nature may improve attention performance through many means: modulation of alertness and one’s connection to nature were investigated here, in two separate studies. In both studies, participants performed the Sustained Attention to Response Task (SART) before and immediately after viewing a meadow, ocean, or urban image for 40 s, and then completed the Perceived Restorativeness Scale. In Study 1 (n = 68), an eye-tracker recorded the participants’ tonic pupil diameter during the SARTs, providing a measure of alertness. In Study 2 (n = 186), the effects of connectedness to nature on SART performance and perceived restoration were studied. In both studies, the image viewed was not associated with participants’ sustained attention performance; both nature images were perceived as equally restorative, and more restorative than the urban image. The image viewed was not associated with changes in alertness. Connectedness to nature was not associated with sustained attention performance, but it did moderate the relation between viewing the natural images and perceived restorativeness; participants reporting a higher connection to nature also reported feeling more restored after viewing the nature, but not the urban, images. Dissociation was found between the physiological and behavioral measures and the perceived restorativeness of the images. The results suggest that restoration associated with nature exposure is not associated with modulation of alertness but is associated with connectedness with nature.

Developing a Search Engine for Precision Medicine

Precision medicine focuses on developing new treatments based on an individual's genetic, environmental, and lifestyle profile. While this data-driven approach has led to significant advances, retrieving information specific to a patient's condition has proved challenging for oncologists due to the large volume of data. In this paper, we propose the PRecIsion Medicine Robust Oncology Search Engine (PRIMROSE) for cancer patients that retrieves scientific articles and clinical trials based on a patient's condition, genetic profile, age, and gender. Our search engine utilizes Elasticsearch indexes for information storage and retrieval, and we developed a knowledge graph for query expansion in order to improve recall. Additionally, we experimented with machine learning and learning-to-rank components to the search engine and compared the results of the two approaches. Finally, we developed a front-facing ReactJS website and a REST API for connecting with our search engine. The development of this front-facing website allows for easy access to our system by healthcare providers.

Accurate prediction of boundaries of high resolution topologically associated domains (TADs) in fruit flies using deep learning

Genomes are organized into self-interacting chromatin regions called topologically associated domains (TADs). A significant number of TAD boundaries are shared across multiple cell types and conserved across species. Disruption of TAD boundaries may affect the expression of nearby genes and could lead to several diseases. Even though detection of TAD boundaries is important and useful, there are experimental challenges in obtaining high resolution TAD locations. Here, we present computational prediction of TAD boundaries from high resolution Hi-C data in fruit flies. By extensive exploration and testing of several deep learning model architectures with hyperparameter optimization, we show that a unique deep learning model consisting of three convolution layers followed by a long short-term-memory layer achieves an accuracy of 96%. This outperforms feature-based models’ accuracy of 91% and an existing method's accuracy of 73–78% based on motif TRAP scores. Our method also detects previously reported motifs such as Beaf-32 that are enriched in TAD boundaries in fruit flies and also several unreported motifs.

Early detection of denervated muscle fibers in hindlimb muscles after sciatic nerve transection in wild type mice and in the G93A mouse model of amyotrophic lateral sclerosis

The cell adhesion molecule N-CAM is localized to the adult neuromuscular junction but is also expressed in the extrajunctional membrane of denervated muscles concurrent with extrajunctional acetylcholine receptors. Here we used N-CAM immunohistochemistry to determine whether we could detect early denervation in hindlimb muscles of the G93A transgenic mouse model of amyotrophic lateral sclerosis (ALS). In denervated wild type mouse muscles, N-CAM immunoreactivity on the sarcolemma of all fiber types and within the sarcoplasm of only type IIA fibers was detected at day 2: approximately 30% of the muscle fibers in cross-section were fully circumscribed by N-CAM immunoreactivity and approximately 25% of fibers were incompletely circumscribed. The proportion of the latter fibers remained constant over the next 8 days as the proportions of the former fibers increased exponentially. Thereafter, fully circumscribed muscle fibers increased to a maximum by 30 days with a concomitant fall in the incompletely circumscribed fibers. Hence, early muscle denervation was detected by the incomplete circumscription of fiber membranes by N-CAM immunoreactivity with full circumscription and intracellular localization indicating more long-term denervation. In the G93A transgenic mouse, rapid denervation of fast-twitch muscles was readily detected by a corresponding proportion of muscle fibers in cross-section with positive N-CAM immunoreactivity. The proportions of incompletely and completely circumscribed muscle fibers corresponded well with the rate of decline in intact motor units and reduced muscle contractile forces. Progressively more fully circumscribed muscle fibers became evident with age. We conclude that the N-CAM immunoreactivity on muscle fiber membranes in muscle cross-sections provides a sensitive means of detecting early muscle fiber denervation.

[Frequency of G6PD deficiency in a group of preschool-aged children in a centrally located area of Cambodia]

The aim of this study was to determine the frequency of G6PD deficiency and assess its impact on morbidity, especially anemia, in preschool-aged children in Cambodia. A total of 151 children including 82 boys and 69 girls from the Kandal province near Phnom Penh were studied. Ages ranged from 8 to 69 months. Blood was collected in EDTA-coated tubes. Blood counts were performed with an ABX Micros 60 system and G6PD in red blood cells was measured with a Roche Cobas Mira Plus system using Gamma reagents. G6PD deficiency was found in 14 cases (13.4% of boys and 4.3% of girls). Deficiency was complete in 7.3% of children and partial in 2%. Anemia defined as hemoglobin concentration less than 110 g/l was detected in 29.1% of children. No case of anemia could be attributed to enzyme deficiency since no sign of hemolysis was observed in any of the three children presenting both conditions. Further study is needed on G6PD deficiency in Cambodia including malaria-endemic areas and on the frequency and severity of jaundice due to enzyme deficiency in newborns.

Antinucleic acid autoantibody responses of normal human origin: antigen specificity and idiotypic characteristics compared to patients with systemic lupus erythematosus and patients with monoclonal IgM

One idiotype family 4.6.3 derived from a human hybridoma IgML monoclonal anti-DNA antibody of normal origin is commonly expressed (approximately 85%) in the serum of patients with systemic lupus erythematosus but not in normal controls, where its level appears to fluctuate with clinical disease activity. In contrast, the IgM released by normal cultured lymphocytes in vitro expresses this idiotype family but does not bind DNA. Both idiotype 4.6.3 and anti-DNA antibodies are frequently detectable in the serum of patients with monoclonal IgM but not IgG. Therefore, it is likely that idiotype 4.6.3 reflects the expression of a common germ line immunoglobulin gene family. Some of the IgM molecules encoded by these germ line immunoglobulin genes also bind DNA.