SARS-CoV-2's brain impact: revealing cortical and cerebellar differences via cluster analysis in COVID-19 recovered patients.

BACKGROUND: COVID-19 is a disease known for its neurological involvement. SARS-CoV-2 infection triggers neuroinflammation, which could significantly contribute to the development of long-term neurological symptoms and structural alterations in the gray matter. However, the existence of a consistent pattern of cerebral atrophy remains uncertain. OBJECTIVE: Our study aimed to identify patterns of brain involvement in recovered COVID-19 patients and explore potential relationships with clinical variables during hospitalization. METHODOLOGY: In this study, we included 39 recovered patients and 39 controls from a pre-pandemic database to ensure their non-exposure to the virus. We obtained clinical data of the patients during hospitalization, and 3 months later; in addition we obtained T1-weighted magnetic resonance images and performed standard screening cognitive tests. RESULTS: We identified two groups of recovered patients based on a cluster analysis of the significant cortical thickness differences between patients and controls. Group 1 displayed significant cortical thickness differences in specific cerebral regions, while Group 2 exhibited significant differences in the cerebellum, though neither group showed cognitive deterioration at the group level. Notably, Group 1 showed a tendency of higher D-dimer values during hospitalization compared to Group 2, prior to p-value correction. CONCLUSION: This data-driven division into two groups based on the brain structural differences, and the possible link to D-dimer values may provide insights into the underlying mechanisms of SARS-COV-2 neurological disruption and its impact on the brain during and after recovery from the disease.

Performance of the new MC-80 automated digital cell morphology analyser in detection of normal and abnormal blood cells: Comparison with the CellaVision DM9600.

INTRODUCTION: Mindray MC-80 is an automated system for digital imaging of white blood cells (WBCs) and their pre-classification. The objective of this work is to analyse its performance comparing it with the CellaVision® DM9600. METHODS: A total of 445 samples were used, 194 normal and 251 abnormal: acute leukaemia (100), myelodysplastic syndromes/myeloproliferative neoplasms (33), lymphoid neoplasms (50), plasma cell neoplasms (14), infections (49) and thrombocytopenia (5). WBC pre-classification values with the MC-80 and DM9600 were compared with (1) the microscope, (2) Mindray BC-6800Plus differentials in only normal samples, and (3) confirmed or reclassified images (post-classification). Pearson's correlation, Lin's concordance, Passing-Bablok regression, and Bland-Altman plots were used. Sensitivity, specificity, positive (PPV) and negative (NPV) predictive values for abnormal cells using the MC-80 were calculated. RESULTS: The PPV and NPV were above 98% and 99%, for normal samples. For immature granulocytes (IG), NPV and PPV were 100% and 74.2%. When comparing the WBC differentials using the MC-80, the microscope and the BC-6800Plus, no differences were found except for basophils and IG. Our results showed good agreement between the pre- and post-classification of normal WBC, including IG, quantified by high correlation and concordance values (0.91-1). Sensitivity and specificity for blasts were 0.984 and 0.640. The MC-80 detected abnormal lymphocytes in 30% of the smears from patients with lymphoid neoplasm. Plasma cell identification was better using the DM9600. The sensitivity and specificity for erythroblast detection were 1 and 0.890. CONCLUSION: We found that the MC-80 shows high performance for WBC differentials for both normal samples and patients with haematological diseases.

Kimura disease as an uncommon cause of persistent hypereosinophilia: a diagnostic challenge.

Kimura disease (KD) is an unusual inflammatory disease of unknown etiology. Despite being described many years ago, KD might cause diagnostic difficulty or be confused with other conditions. Here, we present a 33-year-old Filipino woman who was referred to our hospital for evaluation of persistent eosinophilia and intense pruritus. Blood analysis and peripheral blood smear review showed high eosinophil counts (3.8 x109/L, 40%) that did not show morphological abnormalities. Besides, high serum IgE concentration was detected (33,528 kU/L). Serological tests were positive for Toxocara canis and treatment with albendazol was initiated. Nevertheless, increased eosinophil counts were still present after several months, alongside with high serum IgE concentrations and intense pruritus. During her follow-up, an inguinal adenopathy was detected. The biopsy revealed lymphoid hyperplasia with reactive germinal centers and massive eosinophil infiltration. Proteinaceous deposits of eosinophilic material were also observed. All these findings, together with peripheral blood eosinophilia and high IgE concentrations, confirmed the diagnosis of KD. The diagnosis of KD should be considered in the differential diagnosis of long-standing unexplained eosinophilia in association with high IgE concentrations, pruritus and lymphadenopathies.

Nucleus incertus projections to rat medial septum and entorhinal cortex: rare collateralization and septal-gating of temporal lobe theta rhythm activity.

Nucleus incertus (NI) neurons in the pontine tegmentum give rise to ascending forebrain projections and express the neuropeptide relaxin-3 (RLN3) which acts via the relaxin-family peptide 3 receptor (RXFP3). Activity in the hippocampus and entorhinal cortex can be driven from the medial septum (MS), and the NI projects to all these centers, where a prominent pattern of activity is theta rhythm, which is related to spatial memory processing. Therefore, we examined the degree of collateralization of NI projections to the MS and the medial temporal lobe (MTL), comprising medial and lateral entorhinal cortex (MEnt, LEnt) and dentate gyrus (DG), and the ability of the MS to drive entorhinal theta in the adult rat. We injected fluorogold and cholera toxin-B into the MS septum and either MEnt, LEnt or DG, to determine the percentage of retrogradely labeled neurons in the NI projecting to both or single targets, and the relative proportion of these neurons that were RLN3-positive ( +). The projection to the MS was threefold stronger than that to the MTL. Moreover, a majority of NI neurons projected independently to either MS or the MTL. However, RLN3 + neurons collateralize significantly more than RLN3-negative (-) neurons. In in vivo studies, electrical stimulation of the NI induced theta activity in the MS and the entorhinal cortex, which was impaired by intraseptal infusion of an RXFP3 antagonist, R3(BΔ23-27)R/I5, particularly at ~ 20 min post-injection. These findings suggest that the MS plays an important relay function in the NI-induced generation of theta within the entorhinal cortex.

Continuous Prediction of Web User Visual Attention on Short Span Windows Based on Gaze Data Analytics.

Understanding users' visual attention on websites is paramount to enhance the browsing experience, such as providing emergent information or dynamically adapting Web interfaces. Existing approaches to accomplish these challenges are generally based on the computation of salience maps of static Web interfaces, while websites increasingly become more dynamic and interactive. This paper proposes a method and provides a proof-of-concept to predict user's visual attention on specific regions of a website with dynamic components. This method predicts the regions of a user's visual attention without requiring a constant recording of the current layout of the website, but rather by knowing the structure it presented in a past period. To address this challenge, the concept of visit intention is introduced in this paper, defined as the probability that a user, while browsing, will fixate their gaze on a specific region of the website in the next period. Our approach uses the gaze patterns of a population that browsed a specific website, captured via an eye-tracker device, to aid personalized prediction models built with individual visual kinetics features. We show experimentally that it is possible to conduct such a prediction through multilabel classification models using a small number of users, obtaining an average area under curve of 84.3%, and an average accuracy of 79%. Furthermore, the user's visual kinetics features are consistently selected in every set of a cross-validation evaluation.

Liberté, Égalité, Crédibilité: An experimental study of citizens' perceptions of government responses to COVID-19 in eight countries.

During a global pandemic, individual views of government can be linked to citizens' trust and cooperation with government and their propensity to resist state policies or to take action that influences the course of a pandemic. This article explores citizens' assessments of government responses to COVID-19 as a function of policy substance (restrictions on civil liberties), information about performance, and socioeconomic inequity in outcomes. We conducted a survey experiment and analyzed data on over 7000 respondents from eight democratic countries. We find that across countries, citizens are less favorable toward COVID-19 policies that are more restrictive of civil liberties. Additionally, citizens' views of government performance are significantly influenced by objective performance information from reputable sources and information on the disproportionate impacts of COVID-19 on low-income groups. This study reinforces the importance of policy design and outcomes and the consideration of multiple public values in the implementation of public policies.

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.

Frontostriatal Circuits Alterations Associated with Cognitive Flexibility Deterioration in Huntington's Disease.

BACKGROUND: Recent resting-state functional magnetic resonance imaging studies have reported abnormal functional connectivity (FC) in the prefrontal cortex (PFC)-striatum circuit in patients with premanifest Huntington's disease (HD). However, there is a lack of evidence showing persistence of abnormal frontostriatal FC and its relation to cognitive flexibility performance in patients with clinically manifest HD. OBJECTIVE: The aim of this study was to evaluate the resting-state FC integrity of the frontostriatal circuit and its relation to cognitive flexibility in HD patients and healthy controls (HCs). METHOD: Eighteen patients with early clinical HD manifestation and 18 HCs matched for age, sex, and education participated in this study. Both groups performed the Cambridge Neuropsychological Test Automated Battery (CANTAB) Intra-Extra Dimensional (IED) set-shift task, which measures cognitive flexibility. Resting-state functional magnetic resonance images were also acquired to examine the FC in specific frontostriatal circuits. Eight regions of interest were preselected based on regions previously associated with extradimensional (ED) shifting in patients with premanifest HD. RESULTS: Significant negative correlations between the number of attentional set-shifting errors and the ventral striatum-ventrolateral PFC FC were found in the HD group. This group also showed negative FC correlations between the total errors and the FC between right ventral striatum-right ventrolateral PFC, left ventral striatum-left ventrolateral PFC, and right ventral striatum-left ventrolateral PFC. Negative correlations between the ED errors and left ventral striatum-left ventrolateral PFC and right ventral striatum-right ventrolateral PFC FC were also found. Finally, a positive correlation between the number of stages completed and left ventral striatum-left ventrolateral PFC FC was found. CONCLUSIONS: Manifest HD patients show significant cognitive flexibility deficits in attentional set-shifting that are associated with FC alterations in the frontostriatal circuit. These results show that FC abnormalities found in the prodromal stage of the disease can also be associated with cognitive flexibility deficits at a later clinical stage, making them good candidates to be explored in longitudinal studies.

ATTCT and ATTCC repeat expansions in the ATXN10 gene affect disease penetrance of spinocerebellar ataxia type 10.

Spinocerebellar ataxia type 10 (SCA10) is an autosomal-dominant disorder caused by an expanded pentanucleotide repeat in the ATXN10 gene. This repeat expansion, when fully penetrant, has a size of 850-4,500 repeats. It has been shown that the repeat composition can be a modifier of disease, e.g., seizures. Here, we describe a Mexican kindred in which we identified both pure (ATTCT)n and mixed (ATTCT)n-(ATTCC)n expansions in the same family. We used amplification-free targeted sequencing and optical genome mapping to decipher the composition of these repeat expansions. We found a considerable degree of mosaicism of the repeat expansion. This mosaicism was confirmed in skin fibroblasts from individuals with ATXN10 expansions with RNAScope in situ hybridization. All affected family members with the mixed ATXN10 repeat expansion showed typical clinical signs of spinocerebellar ataxia and epilepsy. In contrast, individuals with the pure ATXN10 expansion present with Parkinson's disease or are unaffected, even in individuals more than 20 years older than the average age at onset for SCA10. Our findings suggest that the pure (ATTCT)n expansion is non-pathogenic, while repeat interruptions, e.g., (ATTCC)n, are necessary to cause SCA10. This mechanism has been recently described for several other repeat expansions including SCA31 (BEAN1), SCA37 (DAB1), and three loci for benign adult familial myoclonic epilepsy BAFME (SAMD12, TNRC6A, RAPGEF2). Therefore, long-read sequencing and optical genome mapping of the entire genomic structure of repeat expansions are critical for clinical practice and genetic counseling, as variations in the repeat can affect disease penetrance, symptoms, and disease trajectory.

Serous body fluid evaluation using the new automated haematology analyser Mindray BC-6800Plus.

OBJECTIVES: Cellular analysis of body fluids (BF) has clinical relevance in several medical conditions. The objective of this study is twofold: (1) evaluate the analytical performance of the BF mode of Mindray BC-6800 Plus compared to manual counts under microscopy and (2) analyse if the high-fluorescent cell counts provided by the analyser (HF-BF) are useful to detect malignancy. METHODS: A total of 285 BF was analysed: 250 corresponding to patients without neoplasia and 35 to patients with malignant diseases. Manual differential counts were performed in BF with ≥250 cells/µL. Percentages and absolute counts were obtained on the BC-6800Plus for total nucleated cells (TC-BF), mononuclear, polymorphonuclear and HF-BF. Statistical analysis was performed using Mann-Whitney U-test, Spearman's correlation, Passing-Bablok regression, Bland-Altman graph and ROC curve. RESULTS: To compare manual and automatic total cell counts, samples were divided in three groups: <250, 250-1,000 and >1,000 cells/µL. Correlation was good in all cases (r=0.72, 0.73 and 0.92, respectively) without significant differences between both methods (p=0.65, 0.39 and 0.30, respectively). The concordance between methods showed values of 90%. Considering malignant samples, median HF-BF values showed significant higher values (102 cells/µL) with respect to non-malignant (4 cells/µL) (p<0.001). The cut-off value of 8.5 HF-BF/µL was able to discriminate samples containing malignant cells showing sensitivity and specificity values of 89 and 71%, respectively. Considering both, HF-BF and TC-BF values, sensitivity and specificity values were 100 and 53%, respectively. CONCLUSIONS: This study reveals that the Mindray BC-6800Plus offers an accurate and acceptable performance, showing results consistent with the manual method. It is recommended to consider both HF-BF and TC-BF values for the screening of the microscopic evaluation to ensure the detection of all malignant samples.

A Deep Learning Approach for the Morphological Recognition of Reactive Lymphocytes in Patients with COVID-19 Infection.

Laboratory medicine plays a fundamental role in the detection, diagnosis and management of COVID-19 infection. Recent observations of the morphology of cells circulating in blood found the presence of particular reactive lymphocytes (COVID-19 RL) in some of the infected patients and demonstrated that it was an indicator of a better prognosis of the disease. Visual morphological analysis is time consuming, requires smear review by expert clinical pathologists, and is prone to subjectivity. This paper presents a convolutional neural network system designed for automatic recognition of COVID-19 RL. It is based on the Xception71 structure and is trained using images of blood cells from real infected patients. An experimental study is carried out with a group of 92 individuals. The input for the system is a set of images selected by the clinical pathologist from the blood smear of a patient. The output is the prediction whether the patient belongs to the group associated with better prognosis of the disease. A threshold is obtained for the classification system to predict that the smear belongs to this group. With this threshold, the experimental test shows excellent performance metrics: 98.3% sensitivity and precision, 97.1% specificity, and 97.8% accuracy. The system does not require costly calculations and can potentially be integrated into clinical practice to assist clinical pathologists in a more objective smear review for early prognosis.

Atypical lymphoid cells circulating in blood in COVID-19 infection: morphology, immunophenotype and prognosis value.

AIMS: Atypical lymphocytes circulating in blood have been reported in COVID-19 patients. This study aims to (1) analyse if patients with reactive lymphocytes (COVID-19 RL) show clinical or biological characteristics related to outcome; (2) develop an automatic system to recognise them in an objective way and (3) study their immunophenotype. METHODS: Clinical and laboratory findings in 36 COVID-19 patients were compared between those showing COVID-19 RL in blood (18) and those without (18). Blood samples were analysed in Advia2120i and stained with May Grünwald-Giemsa. Digital images were acquired in CellaVisionDM96. Convolutional neural networks (CNNs) were used to accurately recognise COVID-19 RL. Immunophenotypic study was performed throughflow cytometry. RESULTS: Neutrophils, D-dimer, procalcitonin, glomerular filtration rate and total protein values were higher in patients without COVID-19 RL (p<0.05) and four of these patients died. Haemoglobin and lymphocyte counts were higher (p<0.02) and no patients died in the group showing COVID-19 RL. COVID-19 RL showed a distinct deep blue cytoplasm with nucleus mostly in eccentric position. Through two sequential CNNs, they were automatically distinguished from normal lymphocytes and classical RL with sensitivity, specificity and overall accuracy values of 90.5%, 99.4% and 98.7%, respectively. Immunophenotypic analysis revealed COVID-19 RL are mostly activated effector memory CD4 and CD8 T cells. CONCLUSION: We found that COVID-19 RL are related to a better evolution and prognosis. They can be detected by morphology in the smear review, being the computerised approach proposed useful to enhance a more objective recognition. Their presence suggests an abundant production of virus-specific T cells, thus explaining the better outcome of patients showing these cells circulating in blood.

Immune-Mediated Cutaneous Paraneoplastic Syndromes Associated With Hematologic Malignancies: Skin as a Mirror of Hematologic Neoplasms.

Malignant neoplasms may present as paraneoplastic syndromes with mucocutaneous manifestations, which may or may not be chronologically associated. The pathophysiological mechanism is complex and not completely understood; therefore, definitive diagnosis may be achieved with a precise differential diagnosis based on the morphology of skin lesions, clinical picture, and histological pattern. The complexities, and low frequency, make the therapeutic approach quite challenging; consequently, the cornerstone of therapy is the eradication of the underlying neoplasms. Corticosteroids are the therapy of choice for most of these immune-mediated manifestations, but for the most part, the successful resolution requires the eradication of the underlying malignancy.

External quality assessment in the absence of proficiency testing: A split-sample testing program experience.

In the field of laboratory medicine, proficiency testing is a vehicle used to improve the reliability of reported results. When proficiency tests are unavailable for a given analyte, an alternative approach is required to ensure adherence to the International Organization for Standardization (ISO) 15189:2012 standard. In this study, we report the results of a split-sample testing program performed as an alternative to a formal PT. This testing method was based on recommendations provided in the Clinical and Laboratory Standards Institute (CLSI) QMS24 guideline. Two different laboratories measured, in duplicate, the heparan sulfate concentration in five samples using ultra-performance liquid chromatography and tandem mass spectrometry. The data analysis to determine the criterion used for the comparability assessment between the two laboratories was based on Appendix E of the QMS24 guideline. Mean interlaboratory differences fell within the maximum allowable differences calculated from the application of the QMS24 guideline, indicating that the results obtained by the two laboratories were comparable across the concentrations tested. Application of the QMS24 split-sample testing procedure allows laboratories to objectively assess test results, thus providing the evidence needed to face an accreditation audit with confidence. However, due to the limitations of statistical analyses in small samples (participants and/or materials), laboratory specialists should assess whether the maximum allowable differences obtained are suitable for the intended use, and make adjustments if necessary.

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.

[High frequency of endoluminal thrombus in patients with ischaemic stroke following COVID-19]. / Alta Frecuencia de Trombo Endoluminal en pacientes con ictus isquémico tras la infección por Coronavirus 2019.

BACKGROUND: Ischaemic stroke may be a major complication of SARS-CoV-2 infection.Studying and characterising the different aetiological subtypes, clinical characteristics, and functional outcomes may be valuable in guiding patient selection for optimal management and treatment. METHODS: Data were collected retrospectively on consecutive patients with COVID-19 who developed acute focal brain ischaemia (between 1 March and 19 April 2020) at a tertiary university hospital in Madrid (Spain). RESULTS: During the study period, 1594 patients were diagnosed with COVID-19. We found 22 patients with ischaemic stroke (1.38%), 6 of whom did not meet the inclusion criteria. The remaining 16 patients were included in the study (15 cases of ischaemic stroke and one case of transient ischaemic attack).Median baseline National Institutes of Health Stroke Scale score was 9 (interquartile range: 16), and mean (standard deviation) age was 73 years (12.8). Twelve patients (75%) were men. Mean time from COVID-19 symptom onset to stroke onset was 13 days. Large vessel occlusion was identified in 12 patients (75%).We detected elevated levels of D-dimer in 87.5% of patients and C-reactive protein in 81.2%. The main aetiology was atherothrombotic stroke (9 patients, 56.3%), with the predominant subtype being endoluminal thrombus (5 patients, 31.2%), involving the internal carotid artery in 4 cases and the aortic arch in one. The mortality rate in our series was 44% (7 of 16 patients). CONCLUSIONS: In patients with COVID-19, the most frequent stroke aetiology was atherothrombosis, with a high proportion of endoluminal thrombus (31.2% of patients). Our clinical and laboratory data support COVID-19-associated coagulopathy as a relevant pathophysiological mechanism for ischaemic stroke in these patients.

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.

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.