Fast and accurate automated recognition of the dominant cells from fecal images based on Faster R-CNN.
Ontology highlight
ABSTRACT:
SUBMITTER: Zhang J
PROVIDER: S-EPMC8121882 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:Deep learning based models have had great success in object detection, but the state of the art models have not yet been widely applied to biological image data. We apply for the first time an object detection model previously used on natural images to identify cells and recognize their stages in brightfield microscopy images of malaria-infected blood. Many micro-organisms like malaria parasites are still studied by expert manual inspection and hand counting. This type of object detection task is challenging due to factors like variations in cell shape, density, and color, and uncertainty of some cell classes. In addition, annotated data useful for training is scarce, and the class distribution is inherently highly imbalanced due to the dominance of uninfected red blood cells. We use Faster Region-based Convolutional Neural Network (Faster R-CNN), one of the top performing object detection models in recent years, pre-trained on ImageNet but fine tuned with our data, and compare it to a baseline, which is based on a traditional approach consisting of cell segmentation, extraction of several single-cell features, and classification using random forests. To conduct our initial study, we collect and label a dataset of 1300 fields of view consisting of around 100,000 individual cells. We demonstrate that Faster R-CNN outperforms our baseline and put the results in context of human performance.
| S-EPMC8691760 | biostudies-literature
Project description:BACKGROUND:There are many different types of microRNAs (miRNAs) and elucidating their functions is still under intensive research. A fundamental step in functional annotation of a new miRNA is to classify it into characterized miRNA families, such as those in Rfam and miRBase. With the accumulation of annotated miRNAs, it becomes possible to use deep learning-based models to classify different types of miRNAs. In this work, we investigate several key issues associated with successful application of deep learning models for miRNA classification. First, as secondary structure conservation is a prominent feature for noncoding RNAs including miRNAs, we examine whether secondary structure-based encoding improves classification accuracy. Second, as there are many more non-miRNA sequences than miRNAs, instead of assigning a negative class for all non-miRNA sequences, we test whether using softmax output can distinguish in-distribution and out-of-distribution samples. Finally, we investigate whether deep learning models can correctly classify sequences from small miRNA families. RESULTS:We present our trained convolutional neural network (CNN) models for classifying miRNAs using different types of feature learning and encoding methods. In the first method, we explicitly encode the predicted secondary structure in a matrix. In the second method, we use only the primary sequence information and one-hot encoding matrix. In addition, in order to reject sequences that should not be classified into targeted miRNA families, we use a threshold derived from softmax layer to exclude out-of-distribution sequences, which is an important feature to make this model useful for real transcriptomic data. The comparison with the state-of-the-art ncRNA classification tools such as Infernal shows that our method can achieve comparable sensitivity and accuracy while being significantly faster. CONCLUSION:Automatic feature learning in CNN can lead to better classification accuracy and sensitivity for miRNA classification and annotation. The trained models and also associated codes are freely available at https://github.com/HubertTang/DeepMir.
| S-EPMC6933638 | biostudies-literature
Project description:In this study, we present initial efforts for a new speech recognition approach aimed at producing different input images for convolutional neural network (CNN)-based speech recognition. We explored the potential of the tympanic membrane (eardrum)-inspired viscoelastic membrane-type diaphragms to deliver audio visualization images using a cross-recurrence plot (CRP). These images were formed by the two phase-shifted vibration responses of viscoelastic diaphragms. We expect this technique to replace the fast Fourier transform (FFT) spectrum currently used for speech recognition. Herein, we report that the new creation method of color images enabled by combining two phase-shifted vibration responses of viscoelastic diaphragms with CRP shows a lower computation burden and a promising potential alternative way to STFT (conventional spectrogram) when the image resolution (pixel size) is below critical resolution.
| S-EPMC10115895 | biostudies-literature
Project description:ObjectiveTo propose a method for mitigate the impact of anomaly points (such as dust, bubbles, scratches on the chip surface, and minor indentations) in images on the results of digital droplet PCR (ddPCR) detection to achieve high-throughput, stable, and accurate detection.MethodsWe propose a Filter Faster R-CNN ddPCR detection model, which employs Faster R-CNN to generate droplet prediction boxes followed by removing the anomalies within the positive droplet prediction boxes using an outlier filtering module (Filter). Using a plasmid carrying a norovirus fragment as the template, we established a ddPCR dataset for model training (2462 instances, 78.56%) and testing (672 instances, 21.44%). Ablation experiments were performed to test the effectiveness of 3 filtering branches of the Filter for anomaly removal on the validation dataset. Comparative experiments with other ddPCR droplet detection models and absolute quantification experiments of ddPCR were conducted to assess the performance of the Filter Faster R-CNN model.ResultsIn low-dust and dusty environments, the Filter Faster R-CNN model achieved detection accuracies of 98.23% and 88.35% for positive droplets, respectively, with composite F1 scores reaching 99.15% and 99.14%, obviously superior to the other models. The introduction of the filtering module significantly enhanced the positive accuracy of the model in dusty environments. In the absolute quantification experiments, a regression line was plotted using the results from commercial flow cytometry equipment as the standard concentration. The results show a regression line slope of 1.0005, an intercept of -0.025, and a determination coefficient of 0.9997, indicating high consistency between the two results.ConclusionThe ddPCR detection technique using the Filter Faster R-CNN model provides a robust detection method for ddPCR under various environmental conditions.
| S-EPMC10954537 | biostudies-literature
Project description:Smart surveillance systems are used to monitor specific areas, such as homes, buildings, and borders, and these systems can effectively detect any threats. In this work, we investigate the design of low-cost multiunit surveillance systems that can control numerous surveillance cameras to track multiple objects (i.e., people, cars, and guns) and promptly detect human activity in real time using low computational systems, such as compact or single board computers. Deep learning techniques are employed to detect certain objects to surveil homes/buildings and recognize suspicious and vital events to ensure that the system can alarm officers of relevant events, such as stranger intrusions, the presence of guns, suspicious movements, and identified fugitives. The proposed model is tested on two computational systems, specifically, a single board computer (Raspberry Pi) with the Raspbian OS and a compact computer (Intel NUC) with the Windows OS. In both systems, we employ components, such as a camera to stream real-time video and an ultrasonic sensor to alarm personnel of threats when movement is detected in restricted areas or near walls. The system program is coded in Python, and a convolutional neural network (CNN) is used to perform recognition. The program is optimized by using a foreground object detection algorithm to improve recognition in terms of both accuracy and speed. The saliency algorithm is used to slice certain required objects from scenes, such as humans, cars, and airplanes. In this regard, two saliency algorithms, based on local and global patch saliency detection are considered. We develop a system that combines two saliency approaches and recognizes the features extracted using these saliency techniques with a conventional neural network. The field results demonstrate a significant improvement in detection, ranging between 34% and 99.9% for different situations. The low percentage is related to the presence of unclear objects or activities that are different from those involving humans. However, even in the case of low accuracy, recognition and threat identification are performed with an accuracy of 100% in approximately 0.7 s, even when using computer systems with relatively weak hardware specifications, such as a single board computer (Raspberry Pi). These results prove that the proposed system can be practically used to design a low-cost and intelligent security and tracking system.
| S-EPMC7959643 | biostudies-literature
Project description:The emergence of the novel coronavirus pneumonia (COVID-19) pandemic at the end of 2019 led to worldwide chaos. However, the world breathed a sigh of relief when a few countries announced the development of a vaccine and gradually began to distribute it. Nevertheless, the emergence of another wave of this pandemic returned us to the starting point. At present, early detection of infected people is the paramount concern of both specialists and health researchers. This paper proposes a method to detect infected patients through chest x-ray images by using the large dataset available online for COVID-19 (COVIDx), which consists of 2128 X-ray images of COVID-19 cases, 8,066 normal cases, and 5,575 cases of pneumonia. A hybrid algorithm is applied to improve image quality before undertaking neural network training. This algorithm combines two different noise-reduction filters in the image, followed by a contrast enhancement algorithm. To detect COVID-19, we propose a novel convolution neural network (CNN) architecture called KL-MOB (COVID-19 detection network based on the MobileNet structure). The performance of KL-MOB is boosted by adding the Kullback-Leibler (KL) divergence loss function when trained from scratch. The KL divergence loss function is adopted for content-based image retrieval and fine-grained classification to improve the quality of image representation. The results are impressive: the overall benchmark accuracy, sensitivity, specificity, and precision are 98.7%, 98.32%, 98.82% and 98.37%, respectively. These promising results should help other researchers develop innovative methods to aid specialists. The tremendous potential of the method proposed herein can also be used to detect COVID-19 quickly and safely in patients throughout the world.
| S-EPMC8459788 | biostudies-literature
Project description:Detailed measurement of cell phenotype information from digital fluorescence images has the potential to greatly advance biomedicine in various disciplines such as patient diagnostics or drug screening. Yet, the complexity of cell conformations presents a major barrier preventing effective determination of cell boundaries, and introduces measurement error that propagates throughout subsequent assessment of cellular parameters and statistical analysis. State-of-the-art image segmentation techniques that require user-interaction, prolonged computation time and specialized training cannot adequately provide the support for high content platforms, which often sacrifice resolution to foster the speedy collection of massive amounts of cellular data. This work introduces a strategy that allows us to rapidly obtain accurate cell boundaries from digital fluorescent images in an automated format. Hence, this new method has broad applicability to promote biotechnology.
| S-EPMC3721074 | biostudies-literature
Project description:The automated generation of radiology reports provides X-rays and has tremendous potential to enhance the clinical diagnosis of diseases in patients. A new research direction is gaining increasing attention that involves the use of hybrid approaches based on natural language processing and computer vision techniques to create auto medical report generation systems. The auto report generator, producing radiology reports, will significantly reduce the burden on doctors and assist them in writing manual reports. Because the sensitivity of chest X-ray (CXR) findings provided by existing techniques not adequately accurate, producing comprehensive explanations for medical photographs remains a difficult task. A novel approach to address this issue was proposed, based on the continuous integration of convolutional neural networks and long short-term memory for detecting diseases, followed by the attention mechanism for sequence generation based on these diseases. Experimental results obtained by using the Indiana University CXR and MIMIC-CXR datasets showed that the proposed model attained the current state-of-the-art efficiency as opposed to other solutions of the baseline. BLEU-1, BLEU-2, BLEU-3, and BLEU-4 were used as the evaluation metrics.
| S-EPMC8736265 | biostudies-literature
Project description:The most significant barrier to success in human activity recognition is extracting and selecting the right features. In traditional methods, the features are chosen by humans, which requires the user to have expert knowledge or to do a large amount of empirical study. Newly developed deep learning technology can automatically extract and select features. Among the various deep learning methods, convolutional neural networks (CNNs) have the advantages of local dependency and scale invariance and are suitable for temporal data such as accelerometer (ACC) signals. In this paper, we propose an efficient human activity recognition method, namely Iss2Image (Inertial sensor signal to Image), a novel encoding technique for transforming an inertial sensor signal into an image with minimum distortion and a CNN model for image-based activity classification. Iss2Image converts real number values from the X, Y, and Z axes into three color channels to precisely infer correlations among successive sensor signal values in three different dimensions. We experimentally evaluated our method using several well-known datasets and our own dataset collected from a smartphone and smartwatch. The proposed method shows higher accuracy than other state-of-the-art approaches on the tested datasets.
| S-EPMC6263516 | biostudies-other
Project description:Measurements of morphometrical parameters on i.e., fish larvae are useful for assessing the quality and condition of the specimen in environmental research or optimal growth in the cultivation industry. Manually acquiring morphometrical parameters from microscopy images can be time consuming and tedious, this can be a limiting factor when acquiring samples for an experiment. Mask R-CNN, an instance segmentation neural network architecture, has been implemented for finding outlines on parts of interest on fish larvae (Atlantic cod, Gadus morhua). Using classical machine vision techniques on the outlines makes it is possible to acquire morphometrics such as area, diameter, length, and height. The combination of these techniques is providing accurate-, consistent-, and high-volume information on the morphometrics of small organisms, making it possible to sample more data for morphometric analysis.•Capabilities to automatically analyse a set of microscopy images in approximately 2-3 seconds per image, with results that have a high degree of accuracy when compared to morphometrics acquired manually by an expert.•Can be implemented on other species of ichthyoplankton or zooplankton and has successfully been tested on ballan wrasse, zebrafish, lumpsucker and calanoid copepods.
| S-EPMC8666706 | biostudies-literature