A Deep Learning Approach for Segmentation of Red Blood Cell Images and Malaria Detection

Clinically applicable optimized periprosthetic joint infection diagnosis via AI based pathology

Periprosthetic joint infection (PJI) is a severe complication after joint replacement surgery that demands precise diagnosis for effective treatment. We enhanced PJI diagnostic accuracy through three steps: (1) developing a self-supervised PJI model with DINO v2 to create a large dataset; (2) comparing multiple intelligent models to identify the best one; and (3) using the optimal model for visual analysis to refine diagnostic practices. The self-supervised model generated 27,724 training samples and achieved a perfect AUC of 1, indicating flawless case differentiation. EfficientNet v2-S outperformed CAMEL2 at the image level, while CAMEL2 was superior at the patient level. By using the weakly supervised PJI model to adjust diagnostic criteria, we reduced the required high-power field diagnoses per slide from five to three. These findings demonstrate AI's potential to improve the accuracy and standardization of PJI pathology and have significant implications for infectious disease diagnostics.

Biophysical profiling of red blood cells from thin-film blood smears using deep learning

Microscopic inspection of thin-film blood smears is widely used to identify red blood cell (RBC) pathologies, including malaria parasitism and hemoglobinopathies, such as sickle cell disease and thalassemia. Emerging research indicates that non-pathologic changes in RBCs can also be detected in images, such as deformability and morphological changes resulting from the storage lesion. In transfusion medicine, cell deformability is a potential biomarker for the quality of donated RBCs. However, a major impediment to the clinical translation of this biomarker is the difficulty associated with performing this measurement. To address this challenge, we developed an approach for biophysical profiling of RBCs based on cell images in thin-film blood smears. We hypothesize that subtle cellular changes are evident in blood smear images, but this information is inaccessible to human expert labellers. To test this hypothesis, we developed a deep learning strategy to analyze Giemsa-stained blood smears to assess the subtle morphologies indicative of RBC deformability and storage-based degradation. Specifically, we prepared thin-film blood smears from 27 RBC samples (9 donors evaluated at 3 storage time points) and imaged them using high-resolution microscopy. Using this dataset, we trained a convolutional neural network to evaluate image-based morphological features related to cell deformability. The prediction of donor deformability is strongly correlated to the microfluidic scores and can be used to categorize images into specific deformability groups with high accuracy. We also used this model to evaluate differences in RBC morphology resulting from cold storage. Together, our results demonstrate that deep learning models can detect subtle cellular morphology differences resulting from deformability and cold storage. This result suggests the potential to assess donor blood quality from thin-film blood smears, which can be acquired ubiquitously in clinical workflows.

Efficient deep learning-based approach for malaria detection using red blood cell smears

Malaria is an extremely malignant disease and is caused by the bites of infected female mosquitoes. This disease is not only infectious among humans, but among animals as well. Malaria causes mild symptoms like fever, headache, sweating and vomiting, and muscle discomfort; severe symptoms include coma, seizures, and kidney failure. The timely identification of malaria parasites is a challenging and chaotic endeavor for health staff. An expert technician examines the schematic blood smears of infected red blood cells through a microscope. The conventional methods for identifying malaria are not efficient. Machine learning approaches are effective for simple classification challenges but not for complex tasks. Furthermore, machine learning involves rigorous feature engineering to train the model and detect patterns in the features. On the other hand, deep learning works well with complex tasks and automatically extracts low and high-level features from the images to detect disease. In this paper, EfficientNet, a deep learning-based approach for detecting Malaria, is proposed that uses red blood cell images. Experiments are carried out and performance comparison is made with pre-trained deep learning models. In addition, k-fold cross-validation is also used to substantiate the results of the proposed approach. Experiments show that the proposed approach is 97.57% accurate in detecting Malaria from red blood cell images and can be beneficial practically for medical healthcare staff.

An improved Yolov5s based on transformer backbone network for detection and classification of bronchoalveolar lavage cells

Biological tissue information of the lung, such as cells and proteins, can be obtained from bronchoalveolar lavage fluid (BALF), through which it can be used as a complement to lung biopsy pathology. BALF cells can be confused with each other due to the similarity of their characteristics and differences in the way sections are handled or viewed. This poses a great challenge for cell detection. In this paper, An Improved Yolov5s Based on Transformer Backbone Network for Detection and Classification of BALF Cells is proposed, focusing on the detection of four types of cells in BALF: macrophages, lymphocytes, neutrophils and eosinophils. The network is mainly based on the Yolov5s network and uses Swin Transformer V2 technology in the backbone network to improve cell detection accuracy by obtaining global information; the C3Ghost module (a variant of the Convolutional Neural Network architecture) is used in the neck network to reduce the number of parameters during feature channel fusion and to improve feature expression performance. In addition, embedding intersection over union Loss (EIoU_Loss) was used as a bounding box regression loss function to speed up the bounding box regression rate, resulting in higher accuracy of the algorithm. The experiments showed that our model could achieve mAP of 81.29% and Recall of 80.47%. Compared to the original Yolov5s, the mAP has improved by 3.3% and Recall by 3.67%. We also compared it with Yolov7 and the newly launched Yolov8s. mAP improved by 0.02% and 2.36% over Yolov7 and Yolov8s respectively, while the FPS of our model was higher than both of them, achieving a balance of efficiency and accuracy, further demonstrating the superiority of our model.

An Improved Search and Rescue Algorithm for Global Optimization and Blood Cell Image Segmentation

Image segmentation has been one of the most active research areas in the last decade. The traditional multi-level thresholding techniques are effective for bi-level thresholding because of their resilience, simplicity, accuracy, and low convergence time, but these traditional techniques are not effective in determining the optimal multi-level thresholding for image segmentation. Therefore, an efficient version of the search and rescue optimization algorithm (SAR) based on opposition-based learning (OBL) is proposed in this paper to segment blood-cell images and solve problems of multi-level thresholding. The SAR algorithm is one of the most popular meta-heuristic algorithms (MHs) that mimics humans' exploration behavior during search and rescue operations. The SAR algorithm, which utilizes the OBL technique to enhance the algorithm's ability to jump out of the local optimum and enhance its search efficiency, is termed mSAR. A set of experiments is applied to evaluate the performance of mSAR, solve the problem of multi-level thresholding for image segmentation, and demonstrate the impact of combining the OBL technique with the original SAR for improving solution quality and accelerating convergence speed. The effectiveness of the proposed mSAR is evaluated against other competing algorithms, including the L'evy flight distribution (LFD), Harris hawks optimization (HHO), sine cosine algorithm (SCA), equilibrium optimizer (EO), gravitational search algorithm (GSA), arithmetic optimization algorithm (AOA), and the original SAR. Furthermore, a set of experiments for multi-level thresholding image segmentation is performed to prove the superiority of the proposed mSAR using fuzzy entropy and the Otsu method as two objective functions over a set of benchmark images with different numbers of thresholds based on a set of evaluation matrices. Finally, analysis of the experiments' outcomes indicates that the mSAR algorithm is highly efficient in terms of the quality of the segmented image and feature conservation, compared with the other competing algorithms.

DLBCNet: A Deep Learning Network for Classifying Blood Cells

Background

Blood is responsible for delivering nutrients to various organs, which store important health information about the human body. Therefore, the diagnosis of blood can indirectly help doctors judge a person's physical state. Recently, researchers have applied deep learning (DL) to the automatic analysis of blood cells. However, there are still some deficiencies in these models.

Methods

To cope with these issues, we propose a novel network for the multi-classification of blood cells, which is called DLBCNet. A new specifical model for blood cells (BCGAN) is designed to generate synthetic images. The pre-trained ResNet50 is implemented as the backbone model, which serves as the feature extractor. The extracted features are fed to the proposed ETRN to improve the multi-classification performance of blood cells.

Results

The average accuracy, average sensitivity, average precision, average specificity, and average f1-score of the proposed model are 95.05%, 93.25%, 97.75%, 93.72%, and 95.38%, accordingly.

Conclusions

The performance of the proposed model surpasses other state-of-the-art methods in reported classification results.

An optimized features selection approach based on Manta Ray Foraging Optimization (MRFO) method for parasite malaria classification

Malaria is a serious and lethal disease that has been reported by the World Health Organization (WHO), with an estimated 219 million new cases and 435,000 deaths globally. The most frequent malaria detection method relies mainly on the specialists who examine the samples under a microscope. Therefore, a computerized malaria diagnosis system is required. In this article, malaria cell segmentation and classification methods are proposed. The malaria cells are segmented using a color-based k-mean clustering approach on the selected number of clusters. After segmentation, deep features are extracted using pre-trained models such as efficient-net-b0 and shuffle-net, and the best features are selected using the Manta-Ray Foraging Optimization (MRFO) method. Two experiments are performed for classification using 10-fold cross-validation, the first experiment is based on the best features selected from the pre-trained models individually, while the second experiment is performed based on the selection of best features from the fusion of extracted features using both pre-trained models. The proposed method provided an accuracy of 99.2% for classification using the linear kernel of the SVM classifier. An empirical study demonstrates that the fused features vector results are better as compared to the individual best-selected features vector and the existing latest methods published so far.

A Radiomics-Based Machine Learning Model for Prediction of Tumor Mutational Burden in Lower-Grade Gliomas

Simple Summary

Lower-grade glioma (LGG) is a kind of center nervous system neoplasm that arises from the glial cells. Lower-grade glioma patients have a median survival time in the range of 1.5–8 years based on the tumor genotypes. In term of epidemiology, most of the lower-grade glioma patients are diagnosed at young adult of age, which led to an early age of death. For exact diagnosis and effective treatment, a pathological result from biopsy sample is required. However, it is long turnaround time. In this study, using pre-operative magnetic resonance images, we developed a non-invasive model to classify tumor mutational burden (TMB), a prognostic factor of treatment response in lower-grade glioma patients, with an accuracy of 0.7936. To our knowledge, our study represents the best model for classification of TMB in LGG patients at present.

Abstract

Glioma is a Center Nervous System (CNS) neoplasm that arises from the glial cells. In a new scheme category of the World Health Organization 2016, lower-grade gliomas (LGGs) are grade II and III gliomas. Following the discovery of suppression of negative immune regulation, immunotherapy is a promising effective treatment method for lower-grade glioma patients. However, the therapy is not effective for all types of LGGs, and tumor mutational burden (TMB) has been shown to be a potential biomarker for the susceptibility and prognosis of immunotherapy in lower-grade glioma patients. Hence, predicting TMB benefits brain cancer patients. In this study, we investigated the correlation between MRI (magnetic resonance imaging)-based radiomic features and TMB in LGG by applying machine learning methods. Six machine learning classifiers were examined on the features extracted from the genetic algorithm. Subsequently, a light gradient boosting machine (LightGBM) succeeded in selecting 11 radiomics signatures for TMB classification. Our LightGBM model resulted in high accuracy of 0.7936, and reached a balance between sensitivity and specificity, achieving 0.76 and 0.8107, respectively. To our knowledge, our study represents the best model for classification of TMB in LGG patients at present.

A content-based image retrieval system for the diagnosis of lymphoma using blood micrographs: An incorporation of deep learning with a traditional learning approach

Lymphomas, or cancers of the lymphatic system, account for around half of all blood cancers diagnosed each year. Lymphoma is a condition that is difficult to diagnose, and accurate diagnosis is critical for effective treatment. Manual microscopic analysis of blood cells requires the involvement of medical experts, whose precision is dependent on their abilities, and it takes time. This paper describes a content-based image retrieval system that uses deep learning-based feature extraction and a traditional learning method for feature reduction to retrieve similar images from a database to aid early/initial lymphoma diagnosis. The proposed algorithm employs a pre-trained network called ResNet-101 to extract image features required to distinguish four types of cells: lymphoma cells, blasts, lymphocytes, and other cells. The issue of class imbalance is resolved by over-sampling the training data followed by data augmentation. Deep learning features are extracted using the activations of the feature layer in the pre-trained net, then dimensionality reduction techniques are used to select discriminant features for the image retrieval system. Euclidean distance is used as the similarity measure to retrieve similar images from the database. The experimentation uses a microscopic blood image dataset with 1673 leukocytes of the categories blasts, lymphoma, lymphocytes, and other cells. The proposed algorithm achieves 98.74% precision in lymphoma cell classification and 99.22% precision @10 for lymphoma cell image retrieval. Experimental findings confirm our approach's practicability and effectiveness. Extended studies endorse the idea of using the prescribed system in actual medical applications, helping doctors diagnose lymphoma, dramatically reducing human resource requirements.

Automatic segmentation of blood cells from microscopic slides: A comparative analysis

With the recent developments in deep learning, automatic cell segmentation from images of microscopic examination slides seems to be a solved problem as recent methods have achieved comparable results on existing benchmark datasets. However, most of the existing cell segmentation benchmark datasets either contain a single cell type, few instances of the cells, not publicly available. Therefore, it is unclear whether the performance improvements can generalize on more diverse datasets. In this paper, we present a large and diverse cell segmentation dataset BBBC041Seg¹, which consists both of uninfected cells (i.e., red blood cells/RBCs, leukocytes) and infected cells (i.e., gametocytes, rings, trophozoites, and schizonts). Additionally, all cell types do not have equal instances, which encourages researchers to develop algorithms for learning from imbalanced classes in a few shot learning paradigm. Furthermore, we conduct a comparative study using both classical rule-based and recent deep learning state-of-the-art (SOTA) methods for automatic cell segmentation and provide them as strong baselines. We believe the introduction of BBBC041Seg will promote future research towards clinically applicable cell segmentation methods from microscopic examinations, which can be later used for downstream tasks such as detecting hematological diseases (i.e., malaria).

Automatic identification of malaria and other red blood cell inclusions using convolutional neural networks

Malaria is a serious disease responsible for thousands of deaths each year. Many efforts have been made to aid in the diagnosis of malaria using machine learning techniques, but to date, the presence of other elements that may interfere with the recognition of malaria has not been considered. We have developed the first deep learning model using convolutional neural networks capable of differentiating malaria-infected red blood cells from not only normal erythrocytes but also erythrocytes with other types of inclusions. 6415 images of red blood cells were segmented from digital images of 53 peripheral blood smears using thresholding and watershed transformation techniques. These images were used to train a VGG-16 architecture using transfer learning. Using an independent test set of 23 smears, this model was 99.5% accurate in classifying malaria parasites and other red blood cell inclusions. This model also exhibited sensitivity and specificity values of 100% and 91.7%, respectively, classifying a complete smear as infected or not infected. Our model represents a promising advance for automation in the identification of malaria-infected patients. The differentiation between malaria parasites and other red blood cell inclusions demonstrates the potential utility of our model in a real work environment.

Experimental evaluation of deep learning method in reticulocyte enumeration in peripheral blood

INTRODUCTION

Reticulocytes (RET) are immature red blood cells, and RET enumeration in peripheral blood has important clinical value in diagnosis, treatment efficacy observation, and prognosis of anemic diseases. For RET enumeration, flow cytometric reference method has shown to be more precise than the manual method by light microscopy. However, flow cytometric method generates occasionally spurious RET counts in some situations. The manual method, which is subjective, imprecise, and tedious, currently remains as an accepted reference method. As a result, there is a need for manual method to be more objective, precise, and rapid.

METHODS

40 EDTA-K2 anticoagulated whole blood samples were randomly selected for the study. 784 microscopic images were taken from blood slides as dataset, and all mature RBCs and RETs in these images were located and labeled by experienced experts. Then, we leverage a Faster R-CNN deep neural network to train a RET detection model and evaluate the model.

RESULTS

Both the recall and precision rate of the model are more than 97%, and average analysis time of a single image is 0.21 seconds.

CONCLUSION

The deep learning method shows outstanding performance including high accuracy and fast speed. The experimental results show that the deep learning method holds the potential to act as a rapid computer-aid method for manual RET enumeration for cytological examiners.

A comparative analysis of deep learning architectures on high variation malaria parasite classification dataset

Malaria, one of the leading causes of death in underdeveloped countries, is primarily diagnosed using microscopy. Computer-aided diagnosis of malaria is a challenging task owing to the fine-grained variability in the appearance of some uninfected and infected class. In this paper, we transform a malaria parasite object detection dataset into a classification dataset, making it the largest malaria classification dataset (63,645 cells), and evaluate the performance of several state-of-the-art deep neural network architectures pretrained on both natural and medical images on this new dataset. We provide detailed insights into the variation of the dataset and qualitative analysis of the results produced by the best models. We also evaluate the models using an independent test set to demonstrate the model's ability to generalize in different domains. Finally, we demonstrate the effect of conditional image synthesis on malaria parasite detection. We provide detailed insights into the influence of synthetic images for the class imbalance problem in the malaria diagnosis context.