Quantitative phase imaging of living red blood cells combining digital holographic microscopy and deep learning.

Digital holographic microscopy as a non-contacting, non-invasive, and highly accurate measurement technology, is becoming a valuable method for quantitatively investigating cells and tissues. Reconstruction of phases from a digital hologram is a key step in quantitative phase imaging for biological and biomedical research. This study proposes a two-stage deep convolutional neural network named VY-Net, to realize the effective and robust phase reconstruction of living red blood cells. The VY-Net can obtain the phase information of an object directly from a single-shot off-axis digital hologram. We also propose two new indices to evaluate the reconstructed phases. In experiments, the mean of the structural similarity index of reconstructed phases can reach 0.9309, and the mean of the accuracy of reconstructions of reconstructed phases is as high as 91.54%. An unseen phase map of a living human white blood cell is successfully reconstructed by the trained VY-Net, demonstrating its strong generality.

Synchronous Phase-Shifting Interference for High Precision Phase Imaging of Objects Using Common Optics.

Quantitative phase imaging and measurement of surface topography and fluid dynamics for objects, especially for moving objects, is critical in various fields. Although effective, existing synchronous phase-shifting methods may introduce additional phase changes in the light field due to differences in optical paths or need specific optics to implement synchronous phase-shifting, such as the beamsplitter with additional anti-reflective coating and a micro-polarizer array. Therefore, we propose a synchronous phase-shifting method based on the Mach-Zehnder interferometer to tackle these issues in existing methods. The proposed method uses common optics to simultaneously acquire four phase-shifted digital holograms with equal optical paths for object and reference waves. Therefore, it can be used to reconstruct the phase distribution of static and dynamic objects with high precision and high resolution. In the experiment, the theoretical resolution of the proposed system was 1.064 µm while the actual resolution could achieve 1.381 µm, which was confirmed by measuring a phase-only resolution chart. Besides, the dynamic phase imaging of a moving standard object was completed to verify the proposed system's effectiveness. The experimental results show that our proposed method is suitable and promising in dynamic phase imaging and measurement of moving objects using phase-shifting digital holography.

VDE-Net: a two-stage deep learning method for phase unwrapping.

Phase unwrapping is a critical step to obtaining a continuous phase distribution in optical phase measurements and coherent imaging techniques. Traditional phase-unwrapping methods are generally low performance due to significant noise or undersampling. This paper proposes a deep convolutional neural network (DCNN) with a weighted jump-edge attention mechanism, namely, VDE-Net, to realize effective and robust phase unwrapping. Experimental results revealed that the weighted jump-edge attention mechanism, which is first proposed and simple to calculate, is useful for phase unwrapping. The proposed algorithm outperformed other networks or common attention mechanisms. In addition, an unseen wrapped phase image of a living red blood cell (RBC) was successfully unwrapped by the trained VDE-Net, thereby demonstrating its strong generalization capability.

Automatic counting of retinal ganglion cells in the entire mouse retina based on improved YOLOv5.

Glaucoma is characterized by the progressive loss of retinal ganglion cells (RGCs), although the pathogenic mechanism remains largely unknown. To study the mechanism and assess RGC degradation, mouse models are often used to simulate human glaucoma and specific markers are used to label and quantify RGCs. However, manually counting RGCs is time-consuming and prone to distortion due to subjective bias. Furthermore, semi-automated counting methods can produce significant differences due to different parameters, thereby failing objective evaluation. Here, to improve counting accuracy and efficiency, we developed an automated algorithm based on the improved YOLOv5 model, which uses five channels instead of one, with a squeeze-and-excitation block added. The complete number of RGCs in an intact mouse retina was obtained by dividing the retina into small overlapping areas and counting, and then merging the divided areas using a non-maximum suppression algorithm. The automated quantification results showed very strong correlation (mean Pearson correlation coefficient of 0.993) with manual counting. Importantly, the model achieved an average precision of 0.981. Furthermore, the graphics processing unit (GPU) calculation time for each retina was less than 1 min. The developed software has been uploaded online as a free and convenient tool for studies using mouse models of glaucoma, which should help elucidate disease pathogenesis and potential therapeutics.

End-to-End Deep Learning-Based Cells Detection in Microscopic Leucorrhea Images.

Vaginitis is a prevalent gynecologic disease that threatens millions of women's health. Although microscopic examination of vaginal discharge is an effective method to identify vaginal infections, manual analysis of microscopic leucorrhea images is extremely time-consuming and labor-intensive. To automate the detection and identification of visible components in microscopic leucorrhea images for early-stage diagnosis of vaginitis, we propose a novel end-to-end deep learning-based cells detection framework using attention-based detection with transformers (DETR) architecture. The transfer learning was applied to speed up the network convergence while maintaining the lowest annotation cost. To address the issue of detection performance degradation caused by class imbalance, the weighted sampler with on-the-fly data augmentation module was integrated into the detection pipeline. Additionally, the multi-head attention mechanism and the bipartite matching loss system of the DETR model perform well in identifying partially overlapping cells in real-time. With our proposed method, the pipeline achieved a mean average precision (mAP) of 86.00% and the average precision (AP) of epithelium, leukocyte, pyocyte, mildew, and erythrocyte was 96.76, 83.50, 74.20, 89.66, and 88.80%, respectively. The average test time for a microscopic leucorrhea image is approximately 72.3 ms. Currently, this cell detection method represents state-of-the-art performance.

A Data-Efficient Framework for the Identification of Vaginitis Based on Deep Learning.

Vaginitis is a gynecological disease affecting the health of millions of women all over the world. The traditional diagnosis of vaginitis is based on manual microscopy, which is time-consuming and tedious. The deep learning method offers a fast and reliable solution for an automatic early diagnosis of vaginitis. However, deep neural networks require massive well-annotated data. Manual annotation of microscopic images is highly cost extensive because it not only is a time-consuming process but also needs highly trained people (doctors, pathologists, or technicians). Most existing active learning approaches are not applicable in microscopic images due to the nature of complex backgrounds and numerous formed elements. To address the problem of high cost of labeling microscopic images, we present a data-efficient framework for the identification of vaginitis based on transfer learning and active learning strategies. The proposed informative sample selection strategy selected the minimal training subset, and then the pretrained convolutional neural network (CNN) was fine-tuned on the selected subset. The experiment results show that the proposed pipeline can save 37.5% annotation cost while maintaining competitive performance. The proposed promising novel framework can significantly save the annotation cost and has the potential of extending widely to other microscopic imaging applications, such as blood microscopic image analysis.

Morphological components detection for super-depth-of-field bio-micrograph based on deep learning.

Accompanied with the clinical routine examination demand increase sharply, the efficiency and accuracy are the first priority. However, automatic classification and localization of cells in microscopic images in super depth of Field (SDoF) system remains great challenges. In this paper, we advance an object detection algorithm for cells in the SDoF micrograph based on Retinanet model. Compared with the current mainstream algorithm, the mean average precision (mAP) index is significantly improved. In the experiment of leucorrhea samples and fecal samples, mAP indexes are 83.1% and 88.1%, respectively, with an average increase of 10%. The object detection model proposed in this paper can be applied to feces and leucorrhea detection equipment, and significantly improve the detection efficiency and accuracy.

SDoF-Net: Super Depth of Field Network for Cell Detection in Leucorrhea Micrograph.

Accompanied with the rapid increase of the demand for routine examination of leucorrhea, efficiency and accuracy become the primary task. However, in super depth of field (SDoF) system, the problem of automatic detection and localization of cells in leucorrhea micro-images is still a big challenge. The changing of the relative position between the cell center and focus plane of microscope lead to variable cell morphological structure in the two-dimensional image, which is an important reason for the low accuracy of current deep learning target detection algorithms. In this paper, an object detection method based on Retinanet in state of super depth of field is proposed, which can achieve high precision detecting of leucorrhea components by the SDoF feature aggregation module. Compared with the current mainstream algorithms, the mean average accuracy (mAP) index has been improved significantly, the mAP index is 82.7% for SDoF module and 83.0% for SDoF+ module, with an average increase of more than 10%. These improved features can significantly improve the efficiency and accuracy of the algorithm. The algorithm proposed in this paper can be integrated into the leucorrhea automatic detection system.

Automatic Semicircular Canal Segmentation of CT Volumes Using Improved 3D U-Net with Attention Mechanism.

The vestibular system is the sensory apparatus that helps the body maintain its postural equilibrium, and semicircular canal is an important organ of the vestibular system. The semicircular canals are three membranous tubes, each forming approximately two-thirds of a circle with a diameter of approximately 6.5 mm, and segmenting them accurately is of great benefit for auxiliary diagnosis, surgery, and treatment of vestibular disease. However, the semicircular canal has small volume, which accounts for less than 1% of the overall computed tomography image. Doctors have to annotate the image in a slice-by-slice manner, which is time-consuming and labor-intensive. To solve this problem, we propose a novel 3D convolutional neural network based on 3D U-Net to automatically segment the semicircular canal. We added the spatial attention mechanism of 3D spatial squeeze and excitation modules, as well as channel attention mechanism of 3D global attention upsample modules to improve the network performance. Our network achieved an average dice coefficient of 92.5% on the test dataset, which shows competitive performance in semicircular canals segmentation task.

A Comprehensive Study of Data Augmentation Strategies for Prostate Cancer Detection in Diffusion-Weighted MRI Using Convolutional Neural Networks.

Data augmentation refers to a group of techniques whose goal is to battle limited amount of available data to improve model generalization and push sample distribution toward the true distribution. While different augmentation strategies and their combinations have been investigated for various computer vision tasks in the context of deep learning, a specific work in the domain of medical imaging is rare and to the best of our knowledge, there has been no dedicated work on exploring the effects of various augmentation methods on the performance of deep learning models in prostate cancer detection. In this work, we have statically applied five most frequently used augmentation techniques (random rotation, horizontal flip, vertical flip, random crop, and translation) to prostate diffusion-weighted magnetic resonance imaging training dataset of 217 patients separately and evaluated the effect of each method on the accuracy of prostate cancer detection. The augmentation algorithms were applied independently to each data channel and a shallow as well as a deep convolutional neural network (CNN) was trained on the five augmented sets separately. We used area under receiver operating characteristic (ROC) curve (AUC) to evaluate the performance of the trained CNNs on a separate test set of 95 patients, using a validation set of 102 patients for finetuning. The shallow network outperformed the deep network with the best 2D slice-based AUC of 0.85 obtained by the rotation method.

A Transfer Learning-Based Active Learning Framework for Brain Tumor Classification.

Brain tumor is one of the leading causes of cancer-related death globally among children and adults. Precise classification of brain tumor grade (low-grade and high-grade glioma) at an early stage plays a key role in successful prognosis and treatment planning. With recent advances in deep learning, artificial intelligence-enabled brain tumor grading systems can assist radiologists in the interpretation of medical images within seconds. The performance of deep learning techniques is, however, highly depended on the size of the annotated dataset. It is extremely challenging to label a large quantity of medical images, given the complexity and volume of medical data. In this work, we propose a novel transfer learning-based active learning framework to reduce the annotation cost while maintaining stability and robustness of the model performance for brain tumor classification. In this retrospective research, we employed a 2D slice-based approach to train and fine-tune our model on the magnetic resonance imaging (MRI) training dataset of 203 patients and a validation dataset of 66 patients which was used as the baseline. With our proposed method, the model achieved area under receiver operating characteristic (ROC) curve (AUC) of 82.89% on a separate test dataset of 66 patients, which was 2.92% higher than the baseline AUC while saving at least 40% of labeling cost. In order to further examine the robustness of our method, we created a balanced dataset, which underwent the same procedure. The model achieved AUC of 82% compared with AUC of 78.48% for the baseline, which reassures the robustness and stability of our proposed transfer learning augmented with active learning framework while significantly reducing the size of training data.

Fast and accurate automated recognition of the dominant cells from fecal images based on Faster R-CNN.

Fecal samples can easily be collected and are representative of a person's current health state; therefore, the demand for routine fecal examination has increased sharply. However, manual operation may pollute the samples, and low efficiency limits the general examination speed; therefore, automatic analysis is needed. Nevertheless, recognition exhaustion time and accuracy remain major challenges in automatic testing. Here, we introduce a fast and efficient cell-detection algorithm based on the Faster-R-CNN technique: the Resnet-152 convolutional neural network architecture. Additionally, a region proposal network and a network combined with principal component analysis are proposed for cell location and recognition in microscopic images. Our algorithm achieved a mean average precision of 84% and a 723 ms detection time per sample for 40,560 fecal images. Thus, this approach may provide a solid theoretical basis for real-time detection in routine clinical examinations while accelerating the process to satisfy increasing demand.

Automatic classification of cells in microscopic fecal images using convolutional neural networks.

The analysis of fecal-type components for clinical diagnosis is important. The main examination involves the counting of red blood cells (RBCs), white blood cells (WBCs), and molds under the microscopic. With the development of machine vision, some vision-based detection schemes have been proposed. However, these methods have a single target for detection, with low detection efficiency and low accuracy. We proposed an algorithm to identify the visible image of fecal composition based on intelligent deep learning. The algorithm mainly includes region proposal and candidate recognition. In the process of segmentation, we proposed a morphology extraction algorithm in a complex background. As for the candidate recognition, we proposed a new convolutional neural network (CNN) architecture based on Inception-v3 and principal component analysis (PCA). This method achieves high-average Precision of 90.7%, which is better than the other mainstream CNN models. Finally, the images within the rectangle marks were obtained. The total time for detection of an image was roughly 1200 ms. The algorithm proposed in the present paper can be integrated into an automatic fecal detection system.

Trichomonas Detection in Leucorrhea Based on VIBE Method.

Trichomonas examination is one of the important items in the leucorrhea routine detection. And it cannot be recognized by still images because of the unstable morphology and unfixed focal location caused by motion characteristic. We proposed an improved VIBE algorithm. 6 videos (totally 1414 frames) are collected for testing. In order to compare the effects of the algorithms, we segment each frame artificially as ground truth. Experiments show that percentage of correct classification (PCC) achieves 88%. The proposed improved method can effectively suppress the false detection caused by the formed components such as epithelial cells in the leucorrhea microscopic image and the missed detection caused by the background model update during the movement. At the same time, improvements can effectively suppress smear and ghost areas. The algorithm proposed in this paper can be integrated into the leucorrhea automatic detection system.

Leukocyte recognition in human fecal samples using texture features.

Unlike urine or blood samples with a single background, human fecal samples contain large amounts of food debris, amorphous particles, and undigested plant cells. It is difficult to segment such impurities when mixed with leukocytes. Cell degradation results in ambiguous nuclei, incompleteness of the cell membrane, and a changeable cell morphology, which are difficult to recognize. Aiming at the segmentation problem, a threshold segmentation method combining an inscribed circle and circumscribed circle is proposed to effectively remove the adhesion impurities with a segmentation accuracy reaching 97.6%. For the identification problem, five texture features (i.e., LBP-uniform, Gabor, HOG, GLCM, and Haar) were extracted and classified using four kinds of classifiers (support vector machine (SVM), artificial neural network, AdaBoost, and random forest). The experimental results show that using a histogram of oriented gradient features with an SVM classifier can achieve precision of 88.46% and recall of 88.72%.

Automatic detection of trichomonads based on an improved Kalman background reconstruction algorithm.

Automatic detection of trichomonads in leukorrhea provides important information for evaluating gynecological diseases. Traditional manual microscopy, which depends on the operator's expertise and subjective factors, has high false-positive rates (i.e., low specificity) and low efficiency. To date, there are many detection methods for biological cells based on morphological characteristics. However, the morphology of trichomonads changes, and its size is not fixed; moreover, they are similar to human leukocytes. Therefore, it is difficult to classify trichomonads based on morphological characteristics. In this study, a moving object detection method based on an improved Kalman background reconstruction algorithm is proposed to detect trichomonads automatically, considering the dynamic characteristics of trichomonads at room temperature. The experimental results show that the trichomonads can be accurately identified, and the phenomena of tailing and ghosts are eliminated. Furthermore, this algorithm easily adapts to continuous or sudden changes in light, focal length variation, and the impact of lens shift, and it has good robustness and only a moderate amount of calculation burden.