Image processing and machine learning in the morphological analysis of blood cells

Artificial Intelligence-Based Management of Adult Chronic Myeloid Leukemia: Where Are We and Where Are We Going?

Artificial intelligence (AI) is emerging as a discipline capable of providing significant added value in Medicine, in particular in radiomic, imaging analysis, big dataset analysis, and also for generating virtual cohort of patients. However, in coping with chronic myeloid leukemia (CML), considered an easily managed malignancy after the introduction of TKIs which strongly improved the life expectancy of patients, AI is still in its infancy. Noteworthy, the findings of initial trials are intriguing and encouraging, both in terms of performance and adaptability to different contexts in which AI can be applied. Indeed, the improvement of diagnosis and prognosis by leveraging biochemical, biomolecular, imaging, and clinical data can be crucial for the implementation of the personalized medicine paradigm or the streamlining of procedures and services. In this review, we present the state of the art of AI applications in the field of CML, describing the techniques and objectives, and with a general focus that goes beyond Machine Learning (ML), but instead embraces the wider AI field. The present scooping review spans on publications reported in Pubmed from 2003 to 2023, and resulting by searching "chronic myeloid leukemia" and "artificial intelligence". The time frame reflects the real literature production and was not restricted. We also take the opportunity for discussing the main pitfalls and key points to which AI must respond, especially considering the critical role of the 'human' factor, which remains key in this domain.

Blood cell image segmentation and classification: a systematic review

Background

Blood diseases such as leukemia, anemia, lymphoma, and thalassemia are hematological disorders that relate to abnormalities in the morphology and concentration of blood elements, specifically white blood cells (WBC) and red blood cells (RBC). Accurate and efficient diagnosis of these conditions significantly depends on the expertise of hematologists and pathologists. To assist the pathologist in the diagnostic process, there has been growing interest in utilizing computer-aided diagnostic (CAD) techniques, particularly those using medical image processing and machine learning algorithms. Previous surveys in this domain have been narrowly focused, often only addressing specific areas like segmentation or classification but lacking a holistic view like segmentation, classification, feature extraction, dataset utilization, evaluation matrices, etc.

Methodology

This survey aims to provide a comprehensive and systematic review of existing literature and research work in the field of blood image analysis using deep learning techniques. It particularly focuses on medical image processing techniques and deep learning algorithms that excel in the morphological characterization of WBCs and RBCs. The review is structured to cover four main areas: segmentation techniques, classification methodologies, descriptive feature selection, evaluation parameters, and dataset selection for the analysis of WBCs and RBCs.

Results

Our analysis reveals several interesting trends and preferences among researchers. Regarding dataset selection, approximately 50% of research related to WBC segmentation and 60% for RBC segmentation opted for manually obtaining images rather than using a predefined dataset. When it comes to classification, 45% of the previous work on WBCs chose the ALL-IDB dataset, while a significant 73% of researchers focused on RBC classification decided to manually obtain images from medical institutions instead of utilizing predefined datasets. In terms of feature selection for classification, morphological features were the most popular, being chosen in 55% and 80% of studies related to WBC and RBC classification, respectively.

Conclusion

The diagnostic accuracy for blood-related diseases like leukemia, anemia, lymphoma, and thalassemia can be significantly enhanced through the effective use of CAD techniques, which have evolved considerably in recent years. This survey provides a broad and in-depth review of the techniques being employed, from image segmentation to classification, feature selection, utilization of evaluation matrices, and dataset selection. The inconsistency in dataset selection suggests a need for standardized, high-quality datasets to strengthen the diagnostic capabilities of these techniques further. Additionally, the popularity of morphological features indicates that future research could further explore and innovate in this direction.

A machine learning tool for the diagnosis of SARS-CoV-2 infection from hemogram parameters

Monocytes and neutrophils play key roles in the cytokine storm triggered by SARS-CoV-2 infection, which changes their conformation and function. These changes are detectable at the cellular and molecular level and may be different to what is observed in other respiratory infections. Here, we applied machine learning (ML) to develop and validate an algorithm to diagnose COVID-19 using blood parameters. In this retrospective single-center study, 49 hemogram parameters from 12,321 patients with clinical suspicion of COVID-19 and tested by RT-PCR (4239 positive and 8082 negative) were analysed. The dataset was randomly divided into training and validation sets. Blood cell parameters and patient age were used to construct the predictive model with the support vector machine (SVM) tool. The model constructed from the training set (5936 patients) achieved an accuracy for diagnosis of SARS-CoV-2 infection of 0.952 (95% CI: 0.875-0.892). Test sensitivity and specificity was 0.868 and 0.899, respectively, with a positive (PPV) and negative (NPV) predictive value of 0.896 and 0.872, respectively (prevalence 0.50). The validation set model (4964 patients) achieved an accuracy of 0.894 (95% CI: 0.883-0.903). Test sensitivity and specificity was 0.8922 and 0.8951, respectively, with a positive (PPV) and negative (NPV) predictive value of 0.817 and 0.94, respectively (prevalence 0.34). The area under the receiver operating characteristic curve was 0.952 for the algorithm performance. This algorithm may allow to rule out COVID-19 diagnosis with 94% of probability. This represents a great advance for early diagnostic orientation and guiding clinical decisions.

Unbiased image segmentation assessment toolkit for quantitative differentiation of state-of-the-art algorithms and pipelines

BACKGROUND

Image segmentation pipelines are commonly used in microscopy to identify cellular compartments like nucleus and cytoplasm, but there are few standards for comparing segmentation accuracy across pipelines. The process of selecting a segmentation assessment pipeline can seem daunting to researchers due to the number and variety of metrics available for evaluating segmentation quality.

RESULTS

Here we present automated pipelines to obtain a comprehensive set of 69 metrics to evaluate segmented data and propose a selection methodology for models based on quantitative analysis, dimension reduction or unsupervised classification techniques and informed selection criteria.

CONCLUSION

We show that the metrics used here can often be reduced to a small number of metrics that give a more complete understanding of segmentation accuracy, with different groups of metrics providing sensitivity to different types of segmentation error. These tools are delivered as easy to use python libraries, command line tools, Common Workflow Language Tools, and as Web Image Processing Pipeline interactive plugins to ensure a wide range of users can access and use them. We also present how our evaluation methods can be used to observe the changes in segmentations across modern machine learning/deep learning workflows and use cases.

Developments in Image Processing Using Deep Learning and Reinforcement Learning

The growth in the volume of data generated, consumed, and stored, which is estimated to exceed 180 zettabytes in 2025, represents a major challenge both for organizations and for society in general. In addition to being larger, datasets are increasingly complex, bringing new theoretical and computational challenges. Alongside this evolution, data science tools have exploded in popularity over the past two decades due to their myriad of applications when dealing with complex data, their high accuracy, flexible customization, and excellent adaptability. When it comes to images, data analysis presents additional challenges because as the quality of an image increases, which is desirable, so does the volume of data to be processed. Although classic machine learning (ML) techniques are still widely used in different research fields and industries, there has been great interest from the scientific community in the development of new artificial intelligence (AI) techniques. The resurgence of neural networks has boosted remarkable advances in areas such as the understanding and processing of images. In this study, we conducted a comprehensive survey regarding advances in AI design and the optimization solutions proposed to deal with image processing challenges. Despite the good results that have been achieved, there are still many challenges to face in this field of study. In this work, we discuss the main and more recent improvements, applications, and developments when targeting image processing applications, and we propose future research directions in this field of constant and fast evolution.

Automatic generation of artificial images of leukocytes and leukemic cells using generative adversarial networks (syntheticcellgan)

BACKGROUND AND OBJECTIVES

Visual analysis of cell morphology has an important role in the diagnosis of hematological diseases. Morphological cell recognition is a challenge that requires experience and in-depth review by clinical pathologists. Within the new trend of introducing computer-aided diagnostic tools in laboratory medicine, models based on deep learning are being developed for the automatic identification of different types of cells in peripheral blood. In general, well-annotated large image sets are needed to train the models to reach a desired classification performance. This is especially relevant when it comes to discerning between cell images in which morphological differences are subtle and when it comes to low prevalent diseases with the consequent difficulty in collecting cell images. The objective of this work is to develop, train and validate SyntheticCellGAN (SCG), a new system for the automatic generation of artificial images of white blood cells, maintaining morphological characteristics very close to real cells found in practice in clinical laboratories.

METHODS

SCG is designed with two sequential generative adversarial networks. First, a Wasserstein structure is used to transform random noise vectors into low resolution images of basic mononuclear cells. Second, the concept of image-to-image translation is used to build specific models that transform the basic images into high-resolution final images with the realistic morphology of each cell type target: 1) the five groups of normal leukocytes (lymphocytes, monocytes, eosinophils, neutrophils and basophils); 2) atypical promyelocytes and hairy cells, which are two relevant cell types of complex morphology with low abundance in blood smears.

RESULTS

The images of the SCG system are evaluated with four experimental tests. In the first test we evaluated the generated images with quantitative metrics for GANs. In the second test, morphological verification of the artificial images is performed by expert clinical pathologists with 100% accuracy. In the third test, two classifiers based on convolutional neural networks (CNN) previously trained with images of real cells are used. Two sets of artificial images of the SCG system are classified with an accuracy of 95.36% and 94%, respectively. In the fourth test, three CNN classifiers are trained with artificial images of the SCG system. Real cells are identified with an accuracy ranging from 87.7% to 100%.

CONCLUSIONS

The SCG system has proven effective in creating images of all normal leukocytes and two low-prevalence cell classes associated with diseases such as acute promyelocyte leukemia and hairy cell leukemia. Once trained, the system requires low computational cost and can help augment high-quality image datasets to improve automatic recognition model training for clinical laboratory practice.

Multicentric evaluation of the variability of digital morphology performances also respect to the reference methods by optical microscopy

INTRODUCTION

Despite the important diagnostic role of peripheral blood morphology, cell classification is subjective. Automated image-processing systems (AIS) provide more accurate and objective morphological evaluation. The aims of this multicenter study were the evaluation of the intra and inter-laboratory variation between different AIS in cell pre-classification and after reclassification, compared with manual optical microscopy, the reference method.

METHODS

Six peripheral blood samples were included in this study, for each sample, 70 May-Grunwald and Giemsa stained PB smears were prepared from each specimen and 10 slides were delivered to the seven laboratories involved. Smears were processed by both optical microscopy (OM) and AIS. In addition, the assessment times of both methods were recorded.

RESULTS

Within-laboratory Reproducibility ranged between 4.76% and 153.78%; between-laboratory Precision ranged between 2.10% and 82.2%, while Total Imprecision ranged between 5.21% and 20.60%. The relative Bland Altman bias ranged between -0.01% and 20.60%. The mean of assessment times were 326 ± 110 s and 191 ± 68 s for AIS post reclassification and OM, respectively.

CONCLUSIONS

AIS can be helpful when the number of cell counted are low and can give advantages in terms of efficiency, objectivity and time saving in the morphological analysis of blood cells. They can also help in the interpretation of some morphological features and can serve as learning and investigation tools.

Burn characterization using object-oriented hyperspectral image classification

This paper presents a new approach based on hyperspectral imaging combined with an object-oriented classification method that allows the generation of burn depth classification maps facilitating easier characterization of burns. Hyperspectral images of 14 patients diagnosed with burns on the upper and lower limbs were acquired using a pushbroom hyperspectral imaging system. The images were analyzed using an object-oriented classification approach that uses objects with specific spectral, textural and spatial attributes as the minimum unit for classifying information. The method performance was evaluated in terms of overall accuracy, sensitivity, precision and specificity computed from the confusion matrix. The results revealed that the approach proposed in this study performed well in differentiating burn classes with a high level of overall accuracy (95.99% ± 0.60%), precision (97.30% ± 2.46%), sensitivity (97.23% ± 3.02%) and specificity (98.02% ± 1.98%). In conclusion, the object-based approach for burns hyperspectral images classification can provide maps that can help surgeons identify with better precision different depths of burn wounds.

Characteristics of Artificial Intelligence Clinical Trials in the Field of Healthcare: A Cross-Sectional Study on ClinicalTrials.gov

Artificial intelligence (AI) has driven innovative transformation in healthcare service patterns, despite a lack of understanding of its performance in clinical practice. We conducted a cross-sectional analysis of AI-related trials in healthcare based on ClinicalTrials.gov, intending to investigate the trial characteristics and AI's development status. Additionally, the Neo4j graph database and visualization technology were employed to construct an AI technology application graph, achieving a visual representation and analysis of research hotspots in healthcare AI. A total of 1725 eligible trials that were registered in ClinicalTrials.gov up to 31 March 2022 were included in this study. The number of trial registrations has dramatically grown each year since 2016. However, the AI-related trials had some design drawbacks and problems with poor-quality result reporting. The proportion of trials with prospective and randomized designs was insufficient, and most studies did not report results upon completion. Currently, most healthcare AI application studies are based on data-driven learning algorithms, covering various disease areas and healthcare scenarios. As few studies have publicly reported results on ClinicalTrials.gov, there is not enough evidence to support an assessment of AI's actual performance. The widespread implementation of AI technology in healthcare still faces many challenges and requires more high-quality prospective clinical validation.

Evaluation of two semi-supervised learning methods and their combination for automatic classification of bone marrow cells

Differential bone marrow (BM) cell counting is an important test for the diagnosis of various hematological diseases. However, it is difficult to accurately classify BM cells due to non-uniformity and the lack of reproducibility of differential counting. Therefore, automatic classification systems have been developed in which deep learning is used. These systems requires large and accurately labeled datasets for training. To overcome this, we used semi-supervised learning (SSL), in which learning proceeds while labeling. We used three methods: self-training (ST), active learning (AL), and a combination of these methods, and attempted to automatically classify 16 types of BM cell images. ST involves data verification, as in AL, before adding them to the training dataset (confirmed self-training: CST). After 25 rounds of CST, AL, and CST + AL, the initial number of training data increased from 425 to 40,518; 3682; and 47,843, respectively. Accuracies for the test data of 50 images for each cell type were 0.944, 0.941, and 0.976, respectively. Data added with CST or AL showed some imbalances between classes, while CST + AL exhibited fewer imbalances. We suggest that CST + AL, when combined with two SSL methods, is efficient in increasing training data for the development of automatic BM cells classification systems.

Discrete protein metric (DPM): A new image similarity metric to calculate accuracy of deep learning-generated cell focal adhesion predictions

Understanding cell behaviors can provide new knowledge on the development of different pathologies. Focal adhesion (FA) sites are important sub-cellular structures that are involved in these processes. To better facilitate the study of FA sites, deep learning (DL) can be used to predict FA site morphology based on limited microscopic datasets (e.g., cell membrane images). However, calculating the accuracy score of these predictions can be challenging due to the discrete/point pattern like nature of FA sites. In the present work, a new image similarity metric, discrete protein metric (DPM), was developed to calculate FA prediction accuracy. This metric measures differences in distribution (d), shape/size (s), and angle (a) of FA sites between predicted and ground truth microscopy images. Performance of the DPM was evaluated by comparing it to three other commonly used image similarity metrics: Pearson correlation coefficient (PCC), feature similarity index (FSIM), and Intersection over Union (IoU). A sensitivity analysis was performed by comparing changes in each metric value due to quantifiable changes in FA site location, number, aspect ratio, area, or orientation. Furthermore, accuracy score of DL-generated predictions was calculated using all four metrics to compare their ability to capture variation across samples. Results showed better sensitivity and range of variation for DPM compared to the other metrics tested. Most importantly, DPM had the ability to determine which FA predictions were quantitatively more accurate and consistent with qualitative assessments. The proposed DPM hence provides a method to validate DL-generated FA predictions and has the potential to be used for investigation of other sub-cellular protein aggregates relevant to cell biology.

Morphometric evaluation of two-pronucleus zygote images using image-processing techniques

Identifying embryos with a high potential for implementation remains a challenge in in vitro fertilization (IVF) cycles. Despite progress in IVF treatment, only a minority of generated embryos has the ability to implant. Another drawback of this practice is the high frequency of multiple pregnancies. This problem leads to economic and health problems. Therefore, the transfer of a single embryo with high implantation potential is the ideal strategy. Morphometric evaluation of two-pronucleus zygote images is a helpful technique when aiming to transfer a single embryo with a high implantation potential. In this study, an automated zygote morphometric evaluation algorithm, called the zygote morphology evaluation (ZME) algorithm, was created to analyze the zygote and provide morphological measurements. The first and most crucial step of the ZME algorithm is the noise reduction step, which was first applied to zygote images. After that, the proposed algorithm detects different parts of the zygote that are indicators of embryo viability and normality, that is the oolemma, perivitelline space, zona pellucida, and nucleolar precursor bodies (NPBs). In addition, a novel dataset was prepared for this task. This dataset consisted of 703 human zygote images, and called the human zygote morphometric evaluation dataset (HZME-DS). Our experimental results in the HZME-DS showed that the ZME algorithm was able to achieve 79.58% average accuracy in identifying the oolemma region, 79.40% average accuracy in determining the perivitelline space, and 79.72% accuracy in identifying the zona pellucida. To calculate the accuracy of identifying NPBs, the proposed algorithm uses Recall and Precision measures, and their harmonic average (F1 measure) reached values of 81.14% and 79.53%, respectively. These encouraging results for our proposed method, which is an automatic and very fast method, showed that the ZME algorithm could help embryologists to evaluate the best zygotes in real time and the best embryos subsequently.

A Review of Artificial Intelligence Applications in Hematology Management: Current Practices and Future Prospects

Background

Machine learning (ML) and deep learning (DL) methods have recently garnered a great deal of attention in the field of cancer research by making a noticeable contribution to the growth of predictive medicine and modern oncological practices. Considerable focus has been particularly directed toward hematologic malignancies because of the complexity in detecting early symptoms. Many patients with blood cancer do not get properly diagnosed until their cancer has reached an advanced stage with limited treatment prospects. Hence, the state-of-the-art revolves around the latest artificial intelligence (AI) applications in hematology management.

Objective

This comprehensive review provides an in-depth analysis of the current AI practices in the field of hematology. Our objective is to explore the ML and DL applications in blood cancer research, with a special focus on the type of hematologic malignancies and the patient’s cancer stage to determine future research directions in blood cancer.

Methods

We searched a set of recognized databases (Scopus, Springer, and Web of Science) using a selected number of keywords. We included studies written in English and published between 2015 and 2021. For each study, we identified the ML and DL techniques used and highlighted the performance of each model.

Results

Using the aforementioned inclusion criteria, the search resulted in 567 papers, of which 144 were selected for review.

Conclusions

The current literature suggests that the application of AI in the field of hematology has generated impressive results in the screening, diagnosis, and treatment stages. Nevertheless, optimizing the patient’s pathway to treatment requires a prior prediction of the malignancy based on the patient’s symptoms or blood records, which is an area that has still not been properly investigated.

Automated detection of leukemia by pretrained deep neural networks and transfer learning: A comparison

Leukemia is usually diagnosed by viewing the smears of blood and bone marrow using microscopes and complex Cytochemical tests can be used to authorize and classify leukemia. But these methods are costly, slow and affected by the proficiency and expertise of the specialists concerned. Leukemia can be detected with the help of image processing-based methods by analyzing microscopic smear images to detect the presence of leukemic cells and such techniques are simple, fast, cheap and not biased by the specialists. The proposed study presents a computer aided diagnosis system that uses pretrained deep Convolutional Neural Networks (CNNs) for detection of leukemia images against normal images. The use of pretrained networks is comparatively an easy method of applying deep learning for image analysis and the comparison results of the present study can be used to choose appropriate networks for diagnostic tasks. The microscopic images used in the proposed work were downloaded from a public dataset ALL-IDB. In the proposed work, image classification is done without using any image segregation and feature extraction practices and the study used pretrained series network AlexNet, VGG-16, VGG-19, Directed Acyclic Graph (DAG) networks GoogLeNet, Inceptionv3, MobileNet-v2, Xception, DenseNet-201, Inception-ResNet-v2 and residual networks ResNet-18, ResNet-50 and ResNet-101 for performing the classification and comparison. A classification accuracy of 100% is obtained with all the pretrained networks used in the study for ALL_IDB1 dataset and for ALL_IDB2 dataset, 100% accuracy is obtained with all networks except the AlexNet and VGG-16. The efficacy of three optimization algorithms Stochastic Gradient Descent with Momentum (SGDM), Root Mean Square propagation (RMSprop) and Adaptive Moment estimation (ADAM) is also compared in all the classifications performed. The study considered the detection of leukemia in general only, and classification of leukemia into different types can be attempted as a future work.

Deep learning detects acute myeloid leukemia and predicts NPM1 mutation status from bone marrow smears

The evaluation of bone marrow morphology by experienced hematopathologists is essential in the diagnosis of acute myeloid leukemia (AML); however, it suffers from a lack of standardization and inter-observer variability. Deep learning (DL) can process medical image data and provides data-driven class predictions. Here, we apply a multi-step DL approach to automatically segment cells from bone marrow images, distinguish between AML samples and healthy controls with an area under the receiver operating characteristic (AUROC) of 0.9699, and predict the mutation status of Nucleophosmin 1 (NPM1)—one of the most common mutations in AML—with an AUROC of 0.92 using only image data from bone marrow smears. Utilizing occlusion sensitivity maps, we observed so far unreported morphologic cell features such as a pattern of condensed chromatin and perinuclear lightening zones in myeloblasts of NPM1-mutated AML and prominent nucleoli in wild-type NPM1 AML enabling the DL model to provide accurate class predictions.

Hybrid Inception v3 XGBoost Model for Acute Lymphoblastic Leukemia Classification

Acute lymphoblastic leukemia (ALL) is the most common type of pediatric malignancy which accounts for 25% of all pediatric cancers. It is a life-threatening disease which if left untreated can cause death within a few weeks. Many computerized methods have been proposed for the detection of ALL from microscopic cell images. In this paper, we propose a hybrid Inception v3 XGBoost model for the classification of acute lymphoblastic leukemia (ALL) from microscopic white blood cell images. In the proposed model, Inception v3 acts as the image feature extractor and the XGBoost model acts as the classification head. Experiments indicate that the proposed model performs better than the other methods identified in literature. The proposed hybrid model achieves a weighted F1 score of 0.986. Through experiments, we demonstrate that using an XGBoost classification head instead of a softmax classification head improves classification performance for this dataset for several different CNN backbones (feature extractors). We also visualize the attention map of the features extracted by Inception v3 to interpret the features learnt by the proposed model.

Accent labeling algorithm based on morphological rules and machine learning in English conversion system

The dependency of a speech recognition system on the accent of a user leads to the variation in its performance, as the people from different backgrounds have different accents. Accent labeling and conversion have been reported as a prospective solution for the challenges faced in language learning and various other voice-based advents. In the English TTS system, the accent labeling of unregistered words is another very important link besides the phonetic conversion. Since the importance of the primary stress is much greater than that of the secondary stress, and the primary stress is easier to call than the secondary stress, the labeling of the primary stress is separated from the secondary stress. In this work, the labeling of primary accents uses a labeling algorithm that combines morphological rules and machine learning; the labeling of secondary accents is done entirely through machine learning algorithms. After 10 rounds of cross-validation, the average tagging accuracy rate of primary stress was 94%, the average tagging accuracy rate of secondary stress was 94%, and the total tagging accuracy rate was 83.6%. This perceptual study separates the labeling of primary and secondary accents providing the promising outcomes.

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 new convolutional neural network predictive model for the automatic recognition of hypogranulated neutrophils in myelodysplastic syndromes

BACKGROUND

Dysplastic neutrophils commonly show at least 2/3 reduction of the content of cytoplasmic granules by morphologic examination. Recognition of less granulated dysplastic neutrophils by human eyes is difficult and prone to inter-observer variability. To tackle this problem, we proposed a new deep learning model (DysplasiaNet) able to automatically recognize the presence of hypogranulated dysplastic neutrophils in peripheral blood.

METHODS

Eight models were generated by varying convolutional blocks, number of layer nodes and fully connected layers. Each model was trained for 20 epochs. The five most accurate models were selected for a second stage, being trained again from scratch for 100 epochs. After training, cut-off values were calculated for a granularity score that discerns between normal and dysplastic neutrophils. Furthermore, a threshold value was obtained to quantify the minimum proportion of dysplastic neutrophils in the smear to consider that the patient might have a myelodysplastic syndrome (MDS). The final selected model was the one with the highest accuracy (95.5%).

RESULTS

We performed a final proof of concept with new patients not involved in previous steps. We reported 95.5% sensitivity, 94.3% specificity, 94% precision, and a global accuracy of 94.85%.

CONCLUSIONS

The primary contribution of this work is a predictive model for the automatic recognition in an objective way of hypogranulated neutrophils in peripheral blood smears. We envision the utility of the model implemented as an evaluation tool for MDS diagnosis integrated in the clinical laboratory workflow.

Object-oriented classification approach for bone metastasis mapping from whole-body bone scintigraphy

PURPOSE

Whole-body bone scintigraphy is the most widely used method for detecting bone metastases in advanced cancer. However, its interpretation depends on the experience of the radiologist. Some automatic interpretation systems have been developed in order to improve diagnostic accuracy. These systems are pixel-based and do not use spatial or textural information of groups of pixels, which could be very important for classifying images with better accuracy. This paper presents a fast method of object-oriented classification that facilitates easier interpretation of bone scintigraphy images.

METHODS

Nine whole-body images from patients suspected with bone metastases were analyzed in this preliminary study. First, an edge-based segmentation algorithm together with the full lambda-schedule algorithm were used to identify the object in the bone scintigraphy and the textural and spatial attributes of these objects were calculated. Then, a set of objects (224 objects, ~ 46% of the total objects) were selected as training data based on visual examination of the image, and were assigned to various levels of radionuclide accumulation before performing the data classification using both k-nearest-neighbor and support vector machine classifiers. The performance of the proposed method was evaluated using as metric the statistical parameters calculated from error matrix.

RESULTS

The results revealed that the proposed object-oriented classification approach using either k-nearest-neighbor or support vector machine as classification methods performed well in detecting bone metastasis in terms of overall accuracy (86.62 ± 2.163% and 86.81 ± 2.137% respectively) and kappa coefficient (0.6395 ± 0.0143 and 0.6481 ± 0.0218 respectively).

CONCLUSIONS

In conclusion, the described method provided encouraging results in mapping bone metastases in whole-body bone scintigraphy.

Image Analysis Using Machine Learning for Automated Detection of Hemoglobin H Inclusions in Blood Smears - A Method for Morphologic Detection of Rare Cells

Background

Morphologic rare cell detection is a laborious, operator-dependent process which has the potential to be improved by the use of image analysis using artificial intelligence. Detection of rare hemoglobin H (HbH) inclusions in red cells in the peripheral blood is a common screening method for alpha-thalassemia. This study aims to develop a convolutional neural network-based algorithm for the detection of HbH inclusions.

Methods

Digital images of HbH-positive and HbH-negative blood smears were used to train and test the software. The software performance was tested on images obtained at various magnifications and on different scanning platforms. Another model was developed for total red cell counting and was used to confirm HbH cell frequency in alpha-thalassemia trait. The threshold minimum red cells to image for analysis was determined by Poisson modeling and validated on image sets.

Results

The sensitivity and specificity of the software for HbH+ cells on images obtained at ×100, ×60, and ×40 objectives were close to 91% and 99%, respectively. When an AI-aided diagnostic model was tested on a pilot of 40 whole slide images (WSIs), good inter-rater reliability and high sensitivity and specificity of slide-level classification were obtained. Using the lowest frequency of HbH+ cells (1 in 100,000) observed in our study, we estimated that a minimum of 2.4 × 10⁶ red cells would need to be analyzed to reduce misclassification at the slide level. The minimum required smear size was validated on 78 image sets which confirmed its validity.

Conclusions

WSI image analysis can be utilized effectively for morphologic rare cell detection. The software can be further developed on WISs and evaluated in future clinical validation studies comparing AI-aided diagnosis with the routine diagnostic method.

A deep learning model (ALNet) for the diagnosis of acute leukaemia lineage using peripheral blood cell images

BACKGROUND AND OBJECTIVES

Morphological differentiation among blasts circulating in blood in acute leukaemia is challenging. Artificial intelligence decision support systems hold substantial promise as part of clinical practise in detecting haematological malignancy. This study aims to develop a deep learning-based system to predict the diagnosis of acute leukaemia using blood cell images.

METHODS

A set of 731 blood smears containing 16,450 single-cell images was analysed from 100 healthy controls, 191 patients with viral infections and 148 with acute leukaemia. Training and testing sets were arranged with 85% and 15% of these smears, respectively. To find the best architecture for acute leukaemia classification VGG16, ResNet101, DenseNet121 and SENet154 were evaluated. Fine-tuning was implemented to these pre-trained CNNs to adapt their layers to our data. Once the best architecture was chosen, a system with two modules working sequentially was configured (ALNet). The first module recognised abnormal promyelocytes among other mononuclear blood cell images, such as lymphocytes, monocytes, reactive lymphocytes and blasts. The second distinguished if blasts were myeloid or lymphoid lineage. The final strategy was to predict patients' initial diagnosis of acute leukaemia lineage using the blood smear review. ALNet was assessed with smears of the testing set.

RESULTS

ALNet provided the correct diagnostic prediction of all patients with promyelocytic and myeloid leukaemia. Sensitivity, specificity and precision values of 100%, 92.3% and 93.7%, respectively, were obtained for myeloid leukaemia. Regarding lymphoid leukaemia, a sensitivity of 89% and specificity and precision values of 100% were obtained.

CONCLUSIONS

ALNet is a predictive model designed with two serially connected convolutional networks. It is proposed to assist clinical pathologists in the diagnosis of acute leukaemia during the blood smear review. It has been proved to distinguish neoplastic (leukaemia) and non-neoplastic (infections) diseases, as well as recognise the leukaemia lineage.

Deep Learning-Based Nuclear Lobe Count Method for Differential Count of Neutrophils

Differentiating neutrophils based on the count of nuclear lobulation is useful for diagnosing various hematological disorders, including megaloblastic anemia, myelodysplastic syndrome, and sepsis. It has been reported that one-fifth of sepsis-infected patients worldwide died between 1990 and 2017. Notably, fewer nuclear-lobed and stab-formed neutrophils develop in the peripheral blood during sepsis. This abnormality can serve as an early diagnostic criterion. However, testing this feature is a complex and time-consuming task that is rife with human error. For this reason, we apply deep learning to automatically differentiate neutrophil and nuclear lobulation counts and report the world's first small-scale pilot. Blood films are prepared using venous peripheral blood taken from four healthy volunteers and are stained with May-Grünwald Giemsa stain. Six-hundred 360 × 363-pixel images of neutrophils having five different nuclear lobulations are automatically captured by Cellavision DM-96, an automatic digital microscope camera. Images are input to an original architecture with five convolutional layers built on a deep learning neural-network platform by Sony, Neural Network Console. The deep learning system distinguishes the four groups (i.e., band-formed, two-, three-, and four- and five- segmented) of neutrophils with up to 99% accuracy, suggesting that neutrophils can be automatically differentiated based on their count of segmented nuclei using deep learning.

Application of machine learning in the management of acute myeloid leukemia: current practice and future prospects

Machine learning (ML) is rapidly emerging in several fields of cancer research. ML algorithms can deal with vast amounts of medical data and provide a better understanding of malignant disease. Its ability to process information from different diagnostic modalities and functions to predict prognosis and suggest therapeutic strategies indicates that ML is a promising tool for the future management of hematologic malignancies; acute myeloid leukemia (AML) is a model disease of various recent studies. An integration of these ML techniques into various applications in AML management can assure fast and accurate diagnosis as well as precise risk stratification and optimal therapy. Nevertheless, these techniques come with various pitfalls and need a strict regulatory framework to ensure safe use of ML. This comprehensive review highlights and discusses recent advances in ML techniques in the management of AML as a model disease of hematologic neoplasms, enabling researchers and clinicians alike to critically evaluate this upcoming, potentially practice-changing technology.

Probing the characteristics and biofunctional effects of disease-affected cells and drug response via machine learning applications

Drug-induced transformations in disease characteristics at the cellular and molecular level offers the opportunity to predict and evaluate the efficacy of pharmaceutical ingredients whilst enabling the optimal design of new and improved drugs with enhanced pharmacokinetics and pharmacodynamics. Machine learning is a promising in-silico tool used to simulate cells with specific disease properties and to determine their response toward drug uptake. Differences in the properties of normal and infected cells, including biophysical, biochemical and physiological characteristics, plays a key role in developing fundamental cellular probing platforms for machine learning applications. Cellular features can be extracted periodically from both the drug treated, infected, and normal cells via image segmentations in order to probe dynamic differences in cell behavior. Cellular segmentation can be evaluated to reflect the levels of drug effect on a distinct cell or group of cells via probability scoring. This article provides an account for the use of machine learning methods to probe differences in the biophysical, biochemical and physiological characteristics of infected cells in response to pharmacokinetics uptake of drug ingredients for application in cancer, diabetes and neurodegenerative disease therapies.

Machine learning in haematological malignancies

Machine learning is a branch of computer science and statistics that generates predictive or descriptive models by learning from training data rather than by being rigidly programmed. It has attracted substantial attention for its many applications in medicine, both as a catalyst for research and as a means of improving clinical care across the cycle of diagnosis, prognosis, and treatment of disease. These applications include the management of haematological malignancy, in which machine learning has created inroads in pathology, radiology, genomics, and the analysis of electronic health record data. As computational power becomes cheaper and the tools for implementing machine learning become increasingly democratised, it is likely to become increasingly integrated into the research and practice landscape of haematology. As such, machine learning merits understanding and attention from researchers and clinicians alike. This narrative Review describes important concepts in machine learning for unfamiliar readers, details machine learning's current applications in haematological malignancy, and summarises important concepts for clinicians to be aware of when appraising research that uses machine learning.

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

Malaria is an endemic life-threating disease caused by the unicellular protozoan parasites of the genus Plasmodium. Confirming the presence of parasites early in all malaria cases ensures species-specific antimalarial treatment, reducing the mortality rate, and points to other illnesses in negative cases. However, the gold standard remains the light microscopy of May-Grünwald–Giemsa (MGG)-stained thin and thick peripheral blood (PB) films. This is a time-consuming procedure, dependent on a pathologist’s skills, meaning that healthcare providers may encounter difficulty in diagnosing malaria in places where it is not endemic. This work presents a novel three-stage pipeline to (1) segment erythrocytes, (2) crop and mask them, and (3) classify them into malaria infected or not. The first and third steps involved the design, training, validation and testing of a Segmentation Neural Network and a Convolutional Neural Network from scratch using a Graphic Processing Unit. Segmentation achieved a global accuracy of 93.72% over the test set and the specificity for malaria detection in red blood cells (RBCs) was 87.04%. This work shows the potential that deep learning has in the digital pathology field and opens the way for future improvements, as well as for broadening the use of the created networks.

A dataset of microscopic peripheral blood cell images for development of automatic recognition systems

This article makes available a dataset that was used for the development of an automatic recognition system of peripheral blood cell images using convolutional neural networks [1]. The dataset contains a total of 17,092 images of individual normal cells, which were acquired using the analyzer CellaVision DM96 in the Core Laboratory at the Hospital Clinic of Barcelona. The dataset is organized in the following eight groups: neutrophils, eosinophils, basophils, lymphocytes, monocytes, immature granulocytes (promyelocytes, myelocytes, and metamyelocytes), erythroblasts and platelets or thrombocytes. The size of the images is 360 × 363 pixels, in format jpg, and they were annotated by expert clinical pathologists. The images were captured from individuals without infection, hematologic or oncologic disease and free of any pharmacologic treatment at the moment of blood collection.

This high-quality labelled dataset may be used to train and test machine learning and deep learning models to recognize different types of normal peripheral blood cells. To our knowledge, this is the first publicly available set with large numbers of normal peripheral blood cells, so that it is expected to be a canonical dataset for model benchmarking.

Diffusion Mapping of Eosinophil-Activation State

Eosinophils are granular leukocytes that play a role in mediating inflammatory responses linked to infection and allergic disease. Their activation during an immune response triggers spatial reorganization and eventual cargo release from intracellular granules. Understanding this process is important in diagnosing eosinophilic disorders and in assessing treatment efficacy; however, current protocols are limited to simply quantifying the number of eosinophils within a blood sample. Given that high optical absorption and scattering by the granular structure of these cells lead to marked image features, the physical changes that occur during activation should be trackable using image analysis. Here, we present a study in which imaging flow cytometry is used to quantify eosinophil activation state, based on the extraction of 85 distinct spatial features from dark-field images formed by light scattered orthogonally to the illuminating beam. We apply diffusion mapping, a time inference method that orders cells on a trajectory based on similar image features. Analysis of exogenous cell activation using eotaxin and endogenous activation in donor samples with elevated eosinophil counts shows that cell position along the diffusion-path line correlates with activation level (99% confidence level). Thus, the diffusion mapping provides an activation metric for each cell. Assessment of activated and control populations using both this spatial image-based, activation score and the integrated side-scatter intensity shows an improved Fisher discriminant ratio rd = 0.7 for the multivariate technique compared with an rd = 0.47 for the traditional whole-cell scatter metric. © 2019 The Authors. Cytometry Part A published by Wiley Periodicals, Inc. on behalf of International Society for Advancement of Cytometry.

Image processing techniques represent innovative tools for comparative analysis of proteins

Different bioinformatic and data-mining approaches have been used for the analysis of proteins. Here, we describe a novel, robust, and reliable approach for comparative analysis of a large number of proteins by combining Image Processing Techniques and Convolutional Deep Neural Network (IPT-CNN). As proof of principle, we used IPT-CNN to predict different subtypes of Influenza A virus (IAV). Over 8000 sequences of surface proteins haemagglutinin (HA) and neuraminidase (NA) from different IAV subtypes were used to create polynomial or binary vector datasets. The datasets were then converted into binary images. Analysis of these images enabled the classification of IAV subtypes with 100% accuracy and, compared to non-image-based approaches, within a shorter time frame. The proteome-based IPT-CNN approach described here may be used for analysis and proteome-based classification of other proteins.

Recognition of peripheral blood cell images using convolutional neural networks

BACKGROUND AND OBJECTIVES

Morphological analysis is the starting point for the diagnostic approach of more than 80% of hematological diseases. However, the morphological differentiation among different types of normal and abnormal peripheral blood cells is a difficult task that requires experience and skills. Therefore, the paper proposes a system for the automatic classification of eight groups of peripheral blood cells with high accuracy by means of a transfer learning approach using convolutional neural networks. With this new approach, it is not necessary to implement image segmentation, the feature extraction becomes automatic and existing models can be fine-tuned to obtain specific classifiers.

METHODS

A dataset of 17,092 images of eight classes of normal peripheral blood cells was acquired using the CellaVision DM96 analyzer. All images were identified by pathologists as the ground truth to train a model to classify different cell types: neutrophils, eosinophils, basophils, lymphocytes, monocytes, immature granulocytes (myelocytes, metamyelocytes and promyelocytes), erythroblasts and platelets. Two designs were performed based on two architectures of convolutional neural networks, Vgg-16 and Inceptionv3. In the first case, the networks were used as feature extractors and these features were used to train a support vector machine classifier. In the second case, the same networks were fine-tuned with our dataset to obtain two end-to-end models for classification of the eight classes of blood cells.

RESULTS

In the first case, the experimental test accuracies obtained were 86% and 90% when extracting features with Vgg-16 and Inceptionv3, respectively. On the other hand, in the fine-tuning experiment, global accuracy values of 96% and 95% were obtained using Vgg-16 and Inceptionv3, respectively. All the models were trained and tested using Keras and Tensorflow with a Nvidia Titan XP Graphics Processing Unit.

CONCLUSIONS

The main contribution of this paper is a classification scheme involving a convolutional neural network trained to discriminate among eight classes of cells circulating in peripheral blood. Starting from a state-of-the-art general architecture, we have established a fine-tuning procedure to develop an end-to-end classifier trained using a dataset with over 17,000 cell images obtained from clinical practice. The performance obtained when testing the system has been truly satisfactory, the values of precision, sensitivity, and specificity being excellent. To summarize, the best overall classification accuracy has been 96.2%.

Machine learning applications in the diagnosis of leukemia: Current trends and future directions

Machine learning (ML) offers opportunities to advance pathological diagnosis, especially with increasing trends in digitalizing microscopic images. Diagnosing leukemia is time-consuming and challenging in many areas globally and there is a growing trend in utilizing ML techniques for its diagnosis. In this review, we aimed to describe the literature of ML utilization in the diagnosis of the four common types of leukemia: acute lymphocytic leukemia (ALL), chronic lymphocytic leukemia (CLL), acute myeloid leukemia (AML), and chronic myelogenous leukemia (CML). Using a strict selection criterion, utilizing MeSH terminology and Boolean logic, an electronic search of MEDLINE and IEEE Xplore Digital Library was performed. The electronic search was complemented by handsearching of references of related studies and the top results of Google Scholar. The full texts of 58 articles were reviewed, out of which, 22 studies were included. The number of studies discussing ALL, AML, CLL, and CML was 12, 8, 3, and 1, respectively. No studies were prospectively applying algorithms in real-world scenarios. Majority of studies had small and homogenous samples and used supervised learning for classification tasks. 91% of the studies were performed after 2010, and 74% of the included studies applied ML algorithms to microscopic diagnosis of leukemia. The included studies illustrated the need to develop the field of ML research, including the transformation from solely designing algorithms to practically applying them clinically.

Patient-based prediction algorithm of relapse after allo-HSCT for acute Leukemia and its usefulness in the decision-making process using a machine learning approach

Although allogeneic hematopoietic stem cell transplantation (allo‐HSCT) is a curative therapy for high‐risk acute leukemia (AL), some patients still relapse. Since patients simultaneously have many prognostic factors, difficulties are associated with the construction of a patient‐based prediction algorithm of relapse. The alternating decision tree (ADTree) is a successful classification method that combines decision trees with the predictive accuracy of boosting. It is a component of machine learning (ML) and has the capacity to simultaneously analyze multiple factors. Using ADTree, we attempted to construct a prediction model of leukemia relapse within 1 year of transplantation. With the model of training data (n = 148), prediction accuracy, the AUC of ROC, and the κ‐statistic value were 78.4%, 0.746, and 0.508, respectively. The false positive rate (FPR) of the relapse prediction was as low as 0.134. In an evaluation of the model with validation data (n = 69), prediction accuracy, AUC, and FPR of the relapse prediction were similar at 71.0%, 0.667, and 0.216, respectively. These results suggest that the model is generalized and highly accurate. Furthermore, the output of ADTree may visualize the branch point of treatment. For example, the selection of donor types resulted in different relapse predictions. Therefore, clinicians may change treatment options by referring to the model, thereby improving outcomes. The present results indicate that ML, such as ADTree, will contribute to the decision‐making process in the diversified allo‐HSCT field and be useful for preventing the relapse of leukemia.

Automatic recognition of different types of acute leukaemia in peripheral blood by image analysis

AIMS

Morphological differentiation among different blast cell lineages is a difficult task and there is a lack of automated analysers able to recognise these abnormal cells. This study aims to develop a machine learning approach to predict the diagnosis of acute leukaemia using peripheral blood (PB) images.

METHODS

A set of 442 smears was analysed from 206 patients. It was split into a training set with 75% of these smears and a testing set with the remaining 25%. Colour clustering and mathematical morphology were used to segment cell images, which allowed the extraction of 2,867 geometric, colour and texture features. Several classification techniques were studied to obtain the most accurate classification method. Afterwards, the classifier was assessed with the images of the testing set. The final strategy was to predict the patient's diagnosis using the PB smear, and the final assessment was done with the cell images of the smears of the testing set.

RESULTS

The highest classification accuracy was achieved with the selection of 700 features with linear discriminant analysis. The overall classification accuracy for the six groups of cell types was 85.8%, while the overall classification accuracy for individual smears was 94% as compared with the true confirmed diagnosis.

CONCLUSIONS

The proposed method achieves a high diagnostic precision in the recognition of different types of blast cells among other mononuclear cells circulating in blood. It is the first encouraging step towards the idea of being a diagnostic support tool in the future.

Quantitative Cytologic Descriptors to Differentiate CLL, Sézary, Granular, and Villous Lymphocytes Through Image Analysis

OBJECTIVES

We aimed to find descriptors to identify chronic lymphocytic leukemia (CLL), Sézary, granular, and villous lymphocytes among normal and abnormal lymphocytes in peripheral blood.

METHODS

Image analysis was applied to 768 images from 15 different types of lymphoid cells and monocytes to determine four discriminant descriptors. For each descriptor, numerical scales were obtained using 627 images from 79 patients. An assessment of the four descriptors was performed using smears from 209 new patients.

RESULTS

Cyan correlation of the nucleus identified clumped chromatin, and standard deviation of the granulometric curve of the cyan of the nucleus was specific for cerebriform chromatin. Skewness of the histogram of the u component of the cytoplasm identified cytoplasmic granulation. Hairiness showed specificity for cytoplasmic villi. In the assessment, 96% of the smears were correctly classified.

CONCLUSIONS

The quantitative descriptors obtained through image analysis may contribute to the morphologic identification of the abnormal lymphoid cells considered in this article.