A convolutional neural network for segmentation of yeast cells without manual training annotations.

MOTIVATION: Single-cell time-lapse microscopy is a ubiquitous tool for studying the dynamics of complex cellular processes. While imaging can be automated to generate very large volumes of data, the processing of the resulting movies to extract high-quality single-cell information remains a challenging task. The development of software tools that automatically identify and track cells is essential for realizing the full potential of time-lapse microscopy data. Convolutional neural networks (CNNs) are ideally suited for such applications, but require great amounts of manually annotated data for training, a time-consuming and tedious process. RESULTS: We developed a new approach to CNN training for yeast cell segmentation based on synthetic data and present (i) a software tool for the generation of synthetic images mimicking brightfield images of budding yeast cells and (ii) a convolutional neural network (Mask R-CNN) for yeast segmentation that was trained on a fully synthetic dataset. The Mask R-CNN performed excellently on segmenting actual microscopy images of budding yeast cells, and a density-based spatial clustering algorithm (DBSCAN) was able to track the detected cells across the frames of microscopy movies. Our synthetic data creation tool completely bypassed the laborious generation of manually annotated training datasets, and can be easily adjusted to produce images with many different features. The incorporation of synthetic data creation into the development pipeline of CNN-based tools for budding yeast microscopy is a critical step toward the generation of more powerful, widely applicable and user-friendly image processing tools for this microorganism. AVAILABILITY AND IMPLEMENTATION: The synthetic data generation code can be found at https://github.com/prhbrt/synthetic-yeast-cells. The Mask R-CNN as well as the tuning and benchmarking scripts can be found at https://github.com/ymzayek/yeastcells-detection-maskrcnn. We also provide Google Colab scripts that reproduce all the results of this work. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Measuring decline in white matter integrity after systemic treatment for breast cancer: omitting skeletonization enhances sensitivity.

Chemotherapy for non-central nervous system cancers is associated with abnormalities in brain structure and function. Diffusion tensor imaging (DTI) allows for studying in vivo microstructural changes in brain white matter. Tract-based spatial statistics (TBSS) is a widely used processing pipeline in which DTI data are typically normalized to a generic DTI template and then 'skeletonized' to compensate for misregistration effects. However, this approach greatly reduces the overall white matter volume that is subjected to statistical analysis, leading to information loss. Here, we present a re-analysis of longitudinal data previously analyzed with standard TBSS (Menning et al., BIB 2018, 324-334). For our current approach, we constructed a pipeline with an optimized registration method in Advanced Normalization Tools (ANTs) where DTI data are registered to a study-specific, high-resolution T1 template and the skeletonization step is omitted. In a head to head comparison, we show that with our novel approach breast cancer survivors who had received chemotherapy plus or minus endocrine therapy (BC + SYST, n = 26) showed a global decline in overall FA that was not present in breast cancer survivors who did not receive systemic therapy (BC-SYST, n = 23) or women without a cancer diagnosis (no cancer controls, NC, n = 30). With the standard TBSS approach we did not find any group differences. Moreover, voxel-based analysis for our novel pipeline showed a widespread decline in FA in the BC + SYST compared to the NC group. Interestingly, the BC-SYST group also showed a decline in FA compared to the NC group, although in much less voxels. These results were not found with the standard TBSS approach. We demonstrate that a modified processing pipeline makes DTI data more sensitive to detecting changes in white matter integrity in non-CNS cancer patients after treatment, particularly chemotherapy.

Chronic Health Conditions and Neurocognitive Function in Aging Survivors of Childhood Cancer: A Report from the Childhood Cancer Survivor Study.

Background: Neurocognitive impairment in survivors of childhood cancer may be associated with direct neurotoxicity, as well as indirect effects of systemic health complications. We evaluated associations among treatment exposures, chronic health conditions, and neurocognitive outcomes in adult survivors of childhood cancer. Methods: Participants included 5507 adult survivors of childhood cancer (47.1% male; mean [SD] age = 31.8 [7.6] years at evaluation; 23.1 [4.5] years postdiagnosis) in the Childhood Cancer Survivor Study who completed a self-report measure of neurocognitive function. Cardiac, pulmonary, and endocrine chronic health conditions were graded using the National Cancer Institute Common Terminology Criteria for Adverse Events (version 4.03). Structural equation modeling was used to examine a priori hypothesized causal pathways among cancer treatment, subsequent chronic health conditions, and neurocognitive outcomes. Multivariable models were used to estimate relative risk for associations of treatments and chronic conditions on neurocognitive function. All statistical tests were two-sided. Results: One-third of survivors with a grade 2 or higher chronic condition reported impairments in task efficiency and memory. In addition to direct effects of cranial radiation, path analyses and multivariable models demonstrated direct effects of cardiopulmonary (ß = 0.10, P = .002; relative risk [RR] = 1.27, 95% confidence interval [CI] = 1.12 to 1.44) and endocrine (ß = 0.07, P = .04; RR = 1.14, 95% CI = 1.02 to 1.28) conditions on impaired task efficiency. We identified similar effects of cardiopulmonary condition on memory (P = .01) and emotional regulation (P = .01). Thoracic radiation was associated with impaired task efficiency (P = .01) and emotional regulation (P = .01) through endocrine morbidity. Conclusions: Non-neurotoxic exposures, such as thoracic radiation, can adversely impact survivors' neurocognitive function through chronic conditions. Management of chronic diseases may mitigate neurocognitive outcomes among aging survivors of childhood cancer.

Impact of sleep, fatigue, and systemic inflammation on neurocognitive and behavioral outcomes in long-term survivors of childhood acute lymphoblastic leukemia.

BACKGROUND: Long-term survivors of childhood acute lymphoblastic leukemia (ALL) are at risk for neurocognitive impairment, which may be associated with fatigue, sleep problems, systemic inflammation, and oxidative stress. We examined these associations among survivors of childhood ALL treated with chemotherapy only. METHODS: Survivors of childhood ALL (male, n = 35 and female, n = 35; mean age, 14.3 years [standard deviation, 4.7 years] and mean years from diagnosis, 7.4 years [standard deviation, 1.9 years]) completed neurocognitive testing, behavioral ratings, and reported sleep quality and fatigue symptoms 5 years after diagnosis. Serum was collected concurrently and assayed for interleukin (IL)-1ß and IL-6, tumor necrosis factor α (TNF-α), high-sensitivity C-reactive protein (hsCRP), malondialdehyde, myeloperoxidase, and oxidized low-density lipoprotein. General linear modeling was used to assess associations among biomarkers and functional outcomes, adjusting for age and stratified by sex. RESULTS: Survivors performed worse than population norms on executive function and processing speed and reported more behavioral problems (P < .05 adjusted for multiple comparison). In female survivors, fatigue was associated with poor executive function (r = 0.41; P = .02), processing speed (r = 0.56; P < .001), and attention (r = 0.36-0.55; P < .05). Female survivors with frequent nighttime awakening displayed more inattention (P = .01), hyperactivity (P = .03), and aggression (P = .01). Worse executive function, processing speed, and behavioral symptoms were observed in female survivors with higher levels of IL-6, IL-1ß, and hsCRP (P < .05). Male survivors with high levels of TNF-α demonstrated worse organization (P = .03), but no significant associations between neurocognitive outcomes and sleep/fatigue measures were observed. CONCLUSION: Neurocognitive function in female survivors of childhood ALL appears more susceptible to the effects of sleep disturbance and fatigue. Systemic inflammation may play a role in neurocognitive impairment and behavioral symptoms. Cancer 2017;123:3410-9. © 2017 American Cancer Society.