Isolation and reconstruction of cardiac mitochondria from SBEM images using a deep learning-based method

Mitochondrial morphological defects are a common feature of diseased cardiac myocytes. However, quantitative assessment of mitochondrial morphology is limited by the time-consuming manual segmentation of electron micrograph (EM) images. To advance understanding of the relation between morphological defects and dysfunction, an efficient morphological reconstruction method is desired to enable isolation and reconstruction of mitochondria from EM images. We propose a new method for isolating and reconstructing single mitochondria from serial block-face scanning EM (SBEM) images. CDeep3M, a cloud-based deep learning network for EM images, was used to segment mitochondrial interior volumes and boundaries. Post-processing was performed using both the predicted interior volume and exterior boundary to isolate and reconstruct individual mitochondria. Series of SBEM images from two separate cardiac myocytes were processed. The highest F1-score was 95% using 50 training datasets, greater than that for previously reported automated methods and comparable to manual segmentations. Accuracy of separation of individual mitochondria was 80% on a pixel basis. A total of 2315 mitochondria in the two series of SBEM images were evaluated with a mean volume of 0.78 µm³. The volume distribution was very broad and skewed; the most frequent mitochondria were 0.04–0.06 µm³, but mitochondria larger than 2.0 µm³ accounted for more than 10% of the total number. The average short-axis length was 0.47 µm. Primarily longitudinal mitochondria (0–30 degrees) were dominant (54%). This new automated segmentation and separation method can help quantitate mitochondrial morphology and improve understanding of myocyte structure–function relationships.     

Automatic deep learning-based pleural effusion classification in lung ultrasound images for respiratory pathology diagnosis

Lung ultrasound (LUS) imaging as a point-of-care diagnostic tool for lung pathologies has been proven superior to X-ray and comparable to CT, enabling earlier and more accurate diagnosis in real-time at the patient’s bedside. The main limitation to widespread use is its dependence on the operator training and experience. COVID-19 lung ultrasound findings predominantly reflect a pneumonitis pattern, with pleural effusion being infrequent. However, pleural effusion is easy to detect and to quantify, therefore it was selected as the subject of this study, which aims to develop an automated system for the interpretation of LUS of pleural effusion. A LUS dataset was collected at the Royal Melbourne Hospital which consisted of 623 videos containing 99,209 2D ultrasound images of 70 patients using a phased array transducer. A standardized protocol was followed that involved scanning six anatomical regions providing complete coverage of the lungs for diagnosis of respiratory pathology. This protocol combined with a deep learning algorithm using a Spatial Transformer Network provides a basis for automatic pathology classification on an image-based level. In this work, the deep learning model was trained using supervised and weakly supervised approaches which used frame- and video-based ground truth labels respectively. The reference was expert clinician image interpretation. Both approaches show comparable accuracy scores on the test set of 92.4% and 91.1%, respectively, not statistically significantly different. However, the video-based labelling approach requires significantly less effort from clinical experts for ground truth labelling.     

Applications of computer vision and machine learning techniques for digitized herbarium specimens: A systematic literature review

Herbaria contain the treasure of millions of specimens that have been preserved for several years for scientific studies. To increase the rate of scientific discoveries, digitization of these specimens is currently ongoing to facilitate the easy access and sharing of data to a wider scientific community. Online digital repositories such as Integrated Digitized Biocollection and the Global Biodiversity Information Facility have already accumulated millions of specimen images yet to be explored. This presents the perfect time to take advantage of the opportunity to automate the identification process and increase the rate of novel discoveries using computer vision (CV) and machine learning (ML) techniques. In this study, a systematic literature review of more than 70 peer-reviewed publications was conducted focusing on the application of computer vision and machine learning techniques to digitized herbarium specimens. The study categorizes the different techniques and applications that are commonly used for digitized herbarium specimens and highlights existing challenges together with their potential solutions. We hope this study will serve as a firm foundation for new researchers in the relevant disciplines and will also be enlightening to both computer science and ecology experts.     

A program system for interactive measurements on digitized cell images.

Using a high precision image scanner and a PDP-8/F minicomputer, we have developed a program system for interactive measurements on microscopic images. By giving simple keyboard commands, the operator can run the image scanner and manipulate the digitized images. The interface between the operator and the microscope-computer system is a Tektronix 4010 graphic terminal. The system allows objects to be isolated and parameters to be calculated from each object, e.g., parameters characterizing shape of the object, irregularity in light transmission over the object, area, integrated light transmission, etc. Objects are isolated and parameters are calculated under complete operator control using interactive computer graphics technique. Calculated parameters may be stored in dedicated data records, which are stored in files for later statistical analysis. The system also includes a statistical evaluation part. Technically, the system consists of a command scanner, which translates commands into internal representation, a parser, which checks the syntax of the commands, and an interpreter, which executes the commands. The system is designed so that new commands can be added easily.     

A learning-based markerless approach for full-body kinematics estimation in-natura from a single image

We present a supervised machine learning approach for markerless estimation of human full-body kinematics for a cyclist from an unconstrained colour image. This approach is motivated by the limitations of existing marker-based approaches restricted by infrastructure, environmental conditions, and obtrusive markers. By using a discriminatively learned mixture-of-parts model, we construct a probabilistic tree representation to model the configuration and appearance of human body joints. During the learning stage, a Structured Support Vector Machine (SSVM) learns body parts appearance and spatial relations. In the testing stage, the learned models are employed to recover body pose via searching in a test image over a pyramid structure. We focus on the movement modality of cycling to demonstrate the efficacy of our approach. In natura estimation of cycling kinematics using images is challenging because of human interaction with a bicycle causing frequent occlusions. We make no assumptions in relation to the kinematic constraints of the model, nor the appearance of the scene. Our technique finds multiple quality hypotheses for the pose. We evaluate the precision of our method on two new datasets using loss functions. Our method achieves a score of 91.1 and 69.3 on mean Probability of Correct Keypoint (PCK) measure and 88.7 and 66.1 on the Average Precision of Keypoints (APK) measure for the frontal and sagittal datasets respectively. We conclude that our method opens new vistas to robust user-interaction free estimation of full body kinematics, a prerequisite to motion analysis.     

基于联邦深度学习的皮肤病智能诊断研究

近年来,基于人工智能的皮肤病智能诊断已经成为智慧医疗领域的热门课题。然而由于单一机构的数据有限,局部训练的神经网络难以满足医疗诊断服务的性能需求,从分散机构中收集数据的集中式学习又存在隐私泄漏的风险。基于上述挑战,本文提出一种基于联邦深度学习的皮肤病智能诊断算法。具体地,对比主流的集中式学习,为了在整合多方数据时防止隐私泄漏,本文引入了联邦学习。各客户端将本地模型发送到中心服务器进行聚合,中心服务器再将聚合得到的全局模型同步到各客户端,实现神经网络模型的训练。进一步,为了解决联邦学习中数据异构性的问题,本文在交叉熵损失的基础上增加了修正项,通过修正项限制本地模型和全局模型间的差异,增加模型对异构数据的关注度,从而减小数据异构对诊断结果的影响。实验结果表明,本文所提的皮肤病智能诊断算法与现有相关方案相比,诊断准确度提高了3%~4%,达到75.9%。     

A software system to record and analyze digitized cell images

This paper describes basic software for digitization and processing of microscopic cell images used at the Department of Clinical Cytology at Uppsala University Hospital. A family of programs running on a PDP-8 minicomputer which is connected to a Leitz Orthoplan microscope with two image scanners, one diode-array scanner and a moving-stage photometer, is used for data collection. The digitized image data is converted by converted by conversion program to IBM compatible format. The data structures for image processing and statistical evaluation on the IBM system are also described. Finally, some experiences from the use of the software in cytology automation are discussed.     

Densitometry from digitized images of X-radiographs: Methodology for measurement of coral skeletal density

Skeletons of massive coral colonies contain annual density bands that are revealed by X-radiography of slices cut along growth axes. These bands allow measurement of skeletal growth parameters such as annual extension rate and annual calcification rate. Such measurements have been important in understanding coral growth, in assessing environmental impacts and in recovering proxy environmental information. Measurements of coral calcification rate from annual density banding require measurements of skeletal density along tracks across skeletal slices and, until now, such density measurements have depended upon specialized and expensive equipment. Here, we describe a straightforward, inexpensive and accurate technique for measuring skeletal density from digitized images of X-radiographs of coral skeletal slices. An aragonitic step-wedge was included in each X-radiograph of a coral slice together with two aluminium bars positioned along the anode-cathode axis. Optical density was measured along tracks across the X-ray images of these different objects. The aragonite step-wedge provided a standard for converting optical density to skeletal density. The aluminium bars were used to correct for the heel effect—a variation in the intensity of the X-ray beam along the anode-cathode axis that would, otherwise, introduce large errors into measurements of skeletal density. Exposure was found to vary from X-radiographs to X-radiograph, necessitating the inclusion of the calibration standards in each X-radiograph of a coral slice. Results obtained using this technique compared well with results obtained by direct gamma densitometry of skeletal slices.     

Prioritizing regionally rare plant species for conservation using herbarium data

Because the conservation of biodiversity occurs under time and resource constraints, it is necessary to prioritize species most deserving of attention. Natural history collections have been identified as a valuable source of information in applied conservation practice, particularly for species-rich taxa like plants. Here, online herbarium information was combined with a novel, straightforward priority setting approach to screen a large list of rare vascular plant species (n = 418) in Saskatchewan, Canada. Data was quantified to develop priority scores (for a given species) using three key criteria: (1) provincial responsibility in species survival, (2) species local population characteristics, and (3) the anthropogenic threats causing species to be rare. The use of a hierarchy of the three criteria, wherein provincial responsibility was assigned the most weight, resulted in the highest ranking for 13 species that exist only in Saskatchewan and no other Canadian province or territory. The list is a first step in identifying species deserving of conservation attention and/or further study, while the method itself was deemed to be highly relevant to conservation managers and decision makers due to its scale adaptability and fairly minimal resource requirements.     

A composite-likelihood approach for detecting directional selection from DNA sequence data

We present a novel composite-likelihood-ratio test (CLRT) for detecting genes and genomic regions that are subject to recurrent natural selection (either positive or negative). The method uses the likelihood functions of Hartl et al. (1994) for inference in a Wright-Fisher genic selection model and corrects for nonindependence among sites by application of coalescent simulations with recombination. Here, we (1) characterize the distribution of the CLRT statistic (Lambda) as a function of the population recombination rate (R=4Ner); (2) explore the effects of bias in estimation of R on the size (type I error) of the CLRT; (3) explore the robustness of the model to population growth, bottlenecks, and migration; (4) explore the power of the CLRT under varying levels of mutation, selection, and recombination; (5) explore the discriminatory power of the test in distinguishing negative selection from population growth; and (6) evaluate the performance of maximum composite-likelihood estimation (MCLE) of the selection coefficient. We find that the test has excellent power to detect weak negative selection and moderate power to detect positive selection. Moreover, the test is quite robust to bias in the estimate of local recombination rate, but not to certain demographic scenarios such as population growth or a recent bottleneck. Last, we demonstrate that the MCLE of the selection parameter has little bias for weak negative selection and has downward bias for positively selected mutations.     

A journey through time: exploring temporal patterns amongst digitized plant specimens from Australia

Online access to species occurrence records has opened new windows into investigating biodiversity patterns across multiple scales. The value of these records for research depends on their spatial, temporal, and taxonomic quality. We assessed temporal patterns in records from the Australasian Virtual Herbarium, asking: (1) How temporally consistent has collecting been across Australia? (2) Which areas of Australia have the most reliable records, in terms of temporal consistency and inventory completeness? (3) Are there temporal trends in the completeness of attribute information associated with records? We undertook a multi-step filtering procedure, then estimated temporal consistency and inventory completeness for sampling units (SUs) of 50?km ×?50?km. We found temporal bias in collecting, with 80% of records collected over the period 1970–1999. South-eastern Australia, the Wet Tropics in north-east Queensland, and parts of Western Australia have received the most consistent sampling effort over time, whereas much of central Australia has had low temporal consistency. Of the SUs, 18% have relatively complete inventories with high temporal consistency in sampling. We also determined that 25% of digitized records had missing attribute information. By identifying areas with low reliability, we can limit erroneous inferences about distribution patterns and identify priority areas for future sampling.     

A safer insecticide for herbarium use

Pyrethrins, the active ingredients of the insecticide pyrethrum, are very effective in killing insects but are quite harmless to mammals. When combined with silica gel to form Drione powder, they can be applied inside herbarium cases for long-term protection against insects. Preliminary tests at the New York Botanical Garden have shown that Drione powder or aerosol spray readily kills anobiid beetles. Drione has potential as a safer alternative to dangerous chemicals that have been used in the past.     

First historical genome of a crop bacterial pathogen from herbarium specimen: Insights into citrus canker emergence

Over the past decade, ancient genomics has been used in the study of various pathogens. In this context, herbarium specimens provide a precious source of dated and preserved DNA material, enabling a better understanding of plant disease emergences and pathogen evolutionary history. We report here the first historical genome of a crop bacterial pathogen, Xanthomonas citri pv. citri (Xci), obtained from an infected herbarium specimen dating back to 1937. Comparing the 1937 genome within a large set of modern genomes, we reconstructed their phylogenetic relationships and estimated evolutionary parameters using Bayesian tip-calibration inferences. The arrival of Xci in the South West Indian Ocean islands was dated to the 19^th century, probably linked to human migrations following slavery abolishment. We also assessed the metagenomic community of the herbarium specimen, showed its authenticity using DNA damage patterns, and investigated its genomic features including functional SNPs and gene content, with a focus on virulence factors.     

A robust deep attention dense convolutional neural network for plant leaf disease identification and classification from smart phone captured real world images

Plant diseases cause significant food loss and hence economic loss around the globe. Therefore, automatic plant disease identification is a primary task to take proper medications for controlling the spread of the diseases. Large variety of plants species and their dissimilar phytopathological symptoms call for the implementation of supervised machine learning techniques for efficient and reliable disease identification and classification. With the development of deep learning strategies, convolutional neural network (CNN) has paved its way for classification of multiple plant diseases by extracting rich features. However, several characteristics of the input images especially captured in real world environment, viz. complex or indistinguishable background, presence of multiple leaves with the diseased leaf, small lesion area, solemnly affect the robustness and accuracy of the CNN modules. Available strategies usually applied standard CNN architectures on the images captured in the laboratory environment and very few have considered practical in-field leaf images for their studies. However, those studies are limited with very limited number of plant species. Therefore, there is need of a robust CNN module which can successfully recognize and classify the dissimilar leaf health conditions of non-identical plants from the in-field RGB images. To achieve the above goal, an attention dense learning (ADL) mechanism is proposed in this article by merging mixed sigmoid attention learning with the basic dense learning process of deep CNN. The basic dense learning process derives new features at higher layer considering all lower layer features and that provides fast and efficient training process. Further, the attention learning process amplifies the learning ability of the dense block by discriminating the meaningful lesion portions of the images from the background areas. Other than adding an extra layer for attention learning, in the proposed ADL block the output features from higher layer dense learning are used as an attention mask to the lower layers. For an effective and fast classification process, five ADL blocks are stacked to build a new CNN architecture named DADCNN-5 for obtaining classification robustness and higher testing accuracy. Initially, the proposed DADCNN-5 module is applied on publicly available extended PlantVillage dataset to classify 38 different health conditions of 14 plant species from 54,305 images. Classification accuracy of 99.93% proves that the proposed CNN module can be used for successful leaf disease identification. Further, the efficacy of the DADCNN-5 model is checked after performing stringent experiments on a new real world plant leaf database, created by the authors. The new leaf database contains 10,851 real-world RGB leaf images of 17 plant species for classifying their 44 distinguished health conditions. Experimental outcomes reveal that the proposed DADCNN-5 outperforms the existing machine learning and standard CNN architectures, and achieved 97.33% accuracy. The obtained sensitivity, specificity and false positive rate values are 96.57%, 99.94% and 0.063% respectively. The module takes approximately 3235 min for training process and achieves 99.86% of training accuracy. Visualization of Class activation mapping (CAM) depicts that DADCNN-5 is able to learn distinguishable features from semantically important regions (i.e. lesion regions) on the leaves. Further, the robustness of the DADCNN-5 is established after experimenting with augmented and noise contaminated images of the practical database.     

A heuristic approach for detecting RNA H-type pseudoknots

MOTIVATION: RNA H-type pseudoknots are ubiquitous pseudoknots that are found in almost all classes of RNA and thought to play very important roles in a variety of biological processes. Detection of these RNA H-type pseudoknots can improve our understanding of RNA structures and their associated functions. However, the currently existing programs for detecting such RNA H-type pseudoknots are still time consuming and sometimes even ineffective. Therefore, efficient and effective tools for detecting the RNA H-type pseudoknots are needed. RESULTS: In this paper, we have adopted a heuristic approach to develop a novel tool, called HPknotter, for efficiently and accurately detecting H-type pseudoknots in an RNA sequence. In addition, we have demonstrated the applicability and effectiveness of HPknotter by testing on some sequences with known H-type pseudoknots. Our approach can be easily extended and applied to other classes of more general pseudoknots. AVAILABILITY: The web server of our HPknotter is available for online analysis at http://bioalgorithm.life.nctu.edu.tw/HPKNOTTER/ CONTACT: cllu@mail.nctu.edu.tw, chiu@cc.nctu.edu.tw     

A deep learning approach for converting prompt gamma images to proton dose distributions: A Monte Carlo simulation study

PurposeIn proton therapy, imaging prompt gamma (PG) rays has the potential to verify proton dose (PD) distribution. Despite the fact that there is a strong correlation between the gamma-ray emission and PD, they are still different in terms of the distribution and the Bragg peak (BP) position. In this work, we investigated the feasibility of using a deep learning approach to convert PG images to PD distributions.MethodsWe designed the Monte Carlo simulations using 20 digital brain phantoms irradiated with a 100-MeV proton pencil beam. Each phantom was used to simulate 200 pairs of PG images and PD distributions. A convolutional neural network based on the U-net architecture was trained to predict PD distributions from PG images.ResultsOur simulation results show that the pseudo PD distributions derived from the corresponding PG images agree well with the simulated ground truths. The mean of the BP position errors from each phantom was less than 0.4 mm. We also found that 2000 pairs of PG images and dose distributions would be sufficient to train the U-net. Moreover, the trained network could be deployed on the unseen data (i.e. different beam sizes, proton energies and real patient CT data).ConclusionsOur simulation study has shown the feasibility of predicting PD distributions from PG images using a deep learning approach, but the reliable prediction of PD distributions requires high-quality PG images. Image-degrading factors such as low counts and limited spatial resolution need to be considered in order to obtain high-quality PG images.     

Extraction of DNA from milligram amounts of fresh,herbarium and mummified plant tissues

Summary We have developed a DNA extraction procedure for milligram amounts of plant tissue. Yields ranged from 0.3–200 nanograms of DNA per milligram of tissue. The factors affecting yield are discussed. Fresh tissue, as well as herbarium specimens (22–118 years old) and mummified seeds and embryos (500 to greater than 44 600 years old) were used. All tissues attempted (57 types from 29 species) yielded measurable amounts of DNA. In no case tested was inhibition observed for restriction enzymes BamHI or EcoRI.     

Columnar cactus recognition in aerial images using a deep learning approach

Tehuacán-Cuicatlán Valley is a semi-arid zone in the south of Mexico. It was inscribed in the World Heritage List by the UNESCO in 2018. This unique area has wide biodiversity including several endemic plants. Unfortunately, human activity is constantly affecting the area. A way to preserve a protected area is to carry out autonomous surveillance of the area. A first step to reach this autonomy is to automatically detect and recognize elements in the area. In this work, we present a deep learning based approach for columnar cactus recognition, specifically, the Neobuxbaumia tetetzo species, endemic of the Valley. An image dataset was generated for this study by our research team, containing more than 10,000 image examples. The proposed approach uses this dataset to train a modified LeNet-5 Convolutional Neural Network. Experimental results have shown a high recognition accuracy, 0.95 for the validation set, validating the use of the approach for columnar cactus recognition.