High dimensionality reduction by matrix factorization for systems pharmacology

Identification of key genes and pathways involved in T-DM1-resistance in OE-19 esophageal cancer cells through bioinformatics analysis

Introduction

Esophageal Cancer (EC) ranks among the most common malignancies worldwide. Most EC patients acquire drug resistance to chemotherapy either intrinsically or acquired after T-DM1 treatment, which shows that increasing or decreasing the expression of particular genes might influence chemotherapeutic sensitivity or resistance. Therefore, gaining a deeper understanding of the altered expression of genes involved in EC drug resistance and developing new therapeutic methods are essential targets for continued advancement in EC therapy.

Methods

The present study aimed to find critical regulatory genes/pathways in the progression of T-DM1 resistance in OE-19 EC cells. Expression datasets were extracted from GEO omnibus. Gene interactions were analyzed, and the protein-protein interaction network was drawn. Then, enrichment analysis of the hub genes and network cluster analysis of the hub genes was performed. Finally, the genes were screened in the DrugBank database as therapeutic targets and molecular docking analysis was done on the selected targets.

Results

In the current study, nine hub genes were identified in TDM-1-resistant EC cells (CTGF, CDH17, THBS1, CXCL8, NRP1, ITGB5, EDN1, FAT1, and PTGS2). The KEGG analysis highlighted the IL-17 signaling pathway and ECM-receptor interaction pathway as the most critical pathways; cluster analysis also showed the significance of these pathways. Therefore, the genes involved in these two pathways, including CXCL8, FSCN1, PTGS2, SERPINE2, LEF1, THBS1, CCN2, TAGLN, CDH11, and ITGA6, were searched in DrugBank as therapeutic targets. The DrugBank analysis suggests a potential role for Nonsteroidal Anti-Inflammatory Drugs (NSAIDs) in reducing T-DM1 drug resistance in EC. The docking results revealed that NSAIDs, including Diclofenac, Mefenamic acid, Celecoxib, Naproxen, and Etoricoxib, significantly suppress resistant cancer cells.

Conclusion

This comprehensive bioinformatics analysis deeply explains the molecular mechanisms governing TDM-1 resistance in EC. The identified hub genes and their associated pathways offer potential targets for therapeutic interventions. Moreover, the possible role of NSAIDs in mitigating T-DM1 resistance presents an intriguing avenue for further investigation. This research contributes significantly to the field and establishes a basis for further research to enhance treatment efficacy for EC patients.

CTGF, FN1, IL-6, THBS1, and WISP1 genes and PI3K-Akt signaling pathway as prognostic and therapeutic targets in gastric cancer identified by gene network modeling

OBJECTIVE

Gastric cancer (GC) is one of the most common malignancies worldwide and it is considered the fourth most common cause of cancer death. This study aimed to find critical genes/pathways in GC pathogenesis to be used as biomarkers or therapeutic targets.

METHODS

Differentially expressed genes were explored between human gastric cancerous and noncancerous tissues, and Gene Ontology and Kyoto Encyclopedia of Genes and Genomes signaling pathway enrichment analyses were done. Hub genes were identified based on the protein-protein interaction network constructed in the STRING database with Cytoscape software. The hub genes were selected for further investigation using GEPIA2 and DrugBank databases.

RESULTS

Ten overexpressed hub genes in GC were identified in the current study, including FN1, TP53, IL-6, CXCL5, ELN, ADAMTS2, WISP1, MMP2, CTGF, and THBS1. The study demonstrated the PI3K-Akt pathway's central involvement in GC, with pronounced alterations in essential components. Survival analysis revealed significant correlations between CTGF, FN1, IL-6, THBS1, and WISP1 overexpression and reduced overall survival times in GC patients.

CONCLUSION

A mutual interplay emerged, where PI3K-Akt signaling could upregulate certain genes, forming feedback loops and intensifying cancer phenotypes. The interconnected overexpression of genes and the PI3K-Akt pathway fosters gastric tumorigenesis, suggesting therapeutic potential. DrugBank analysis identified limited FDA-approved drugs, advocating for further exploration while targeting these hub genes could reshape GC treatment. The identified genes could be novel diagnostic/prognostic biomarkers or potential therapeutic targets for GC, but further clinical validation is required.

A novel disulfidptosis-related lncRNA signature for predicting prognosis and potential targeted therapy in hepatocellular carcinoma

Disulfidptosis is a recently discovered mode of cell death with a significant role in cancer. Long non-coding RNAs (lncRNAs) have been implicated in numerous biological processes including oncogenesis, invasion, and metastasis. In this work, we developed an lncRNA signature associated with disulfidptosis for prediction of survival of hepatocellular carcinoma (HCC) patients. Detailed HCC expression profiles and clinical information were obtained from The Cancer Genome Atlas, and 599 differentially expressed disulfidptosis-related lncRNAs were identified through Pearson correlation analysis. Finally, by the least absolute shrinkage and selection operator method, we constructed an HCC prognostic model containing 7 disulfidptosis-related lncRNAs. We split patients into high- and low-risk groups based on the risk values generated by this model and showed that patients in the high-risk group had shorter overall survival times. In the training dataset, receiver operating characteristic curves for 1-, 3-, and 5-year survival were drawn according to the standard (0.788, 0.801, 0.803) and internal validation set (0.684, 0.595, 0.704) to assess the efficacy of the signature. Risk value was confirmed as an independent predictor and used to construct a nomogram in combination with several clinical factors. We further assessed the signature with respect to tumor immune landscape, gene set enrichment analysis, principal component analysis, tumor mutation burden, tumor immune dysfunction and exclusion, and drug sensitivity. High-risk patients had higher immune function scores, except for type II IFN response, whereas low-risk patients had significantly lower tumor immune dysfunction and rejection scores, indicating that they were more sensitive to immune checkpoint inhibitors. Drug sensitivity analysis showed that low-risk patients could benefit more from certain anti-tumor drugs, including sulafenib. In summary, we have constructed a novel signature that shows good performance in predicting survival of patients with HCC and may provide new insights for targeted tumor therapy.

DSKCA-UNet: Dynamic selective kernel channel attention for medical image segmentation

U-Net has attained immense popularity owing to its performance in medical image segmentation. However, it cannot be modeled explicitly over remote dependencies. By contrast, the transformer can effectively capture remote dependencies by leveraging the self-attention (SA) of the encoder. Although SA, an important characteristic of the transformer, can find correlations between them based on the original data, secondary computational complexity might retard the processing rate of high-dimensional data (such as medical images). Furthermore, SA is limited because the correlation between samples is overlooked; thus, there is considerable scope for improvement. To this end, based on Swin-UNet, we introduce a dynamic selective attention mechanism for the convolution kernels. The weight of each convolution kernel is calculated to fuse the results dynamically. This attention mechanism permits each neuron to adaptively modify its receptive field size in response to multiscale input information. A local cross-channel interaction strategy without dimensionality reduction was introduced, which effectively eliminated the influence of downscaling on learning channel attention. Through suitable cross-channel interactions, model complexity can be significantly reduced while maintaining its performance. Subsequently, the global interaction between the encoder features is used to extract more fine-grained features. Simultaneously, the mixed loss function of the weighted cross-entropy loss and Dice loss is used to alleviate category imbalances and achieve better results when the sample number is unbalanced. We evaluated our proposed method on abdominal multiorgan segmentation and cardiac segmentation datasets, achieving Dice similarity coefficient and 95% Hausdorff distance metrics of 80.30 and 14.55%, respectively, on the Synapse dataset and Dice similarity coefficient metrics of 90.80 on the ACDC dataset. The experimental results show that our proposed method has good generalization ability and robustness, and it is a powerful tool for medical image segmentation.

The potential use and experimental validation of genomic instability-related lncRNA in pancreatic carcinoma

This study explored the potential role of long noncoding RNA (lncRNAs) associated with genomic instability in the diagnosis and treatment of pancreatic adenocarcinoma (PAAD). Transcriptome and single-nucleotide variation data of PAAD samples were downloaded from the cancer genome atlas database to explore genomic instability-associated lncRNAs. We constructed a genomic instability-associated lncRNA prognostic signature. Then gene ontology and Kyoto encyclopedia of genes and genomes enrichment analyses were used to explore the physiological role of lncRNAs involved in genomic instability. Tumor microenvironments, immunotherapy response, immune cell infiltration, immune checkpoint, and drug sensitivity were compared between high-risk and low-risk groups. In vitro experiments were performed for external validation. Six lncRNAs associated with genomic instability were identified, capable of predicting the prognosis of PAAD. Patients were assigned to low-risk or high-risk groups using these biomarkers, with better or worse prognosis, respectively. The tumor immune score, immune cell infiltration, and efficacy of immunotherapy were worse in the high-risk group. A drug sensitivity analysis revealed the high- and low-risk groups had different half-maximal inhibitory concentrations. The expression of cancer susceptibility candidate 8 was significantly higher in tumor tissues than in normal tissues, while the expression of LYPLAL1-AS1 exhibited an opposite pattern. They may be potential diagnostic or prognostic biomarkers for patients with pancreatic cancer. Genomic instability-associated lncRNAs were explored in this study and predicted the prognosis of PAAD and stratified patients risk in PAAD. These lncRNAs also predicted the efficacy of immunotherapy and potential therapeutic targets in PAAD.

Unsupervised feature selection based on variance-covariance subspace distance

Subspace distance is an invaluable tool exploited in a wide range of feature selection methods. The power of subspace distance is that it can identify a representative subspace, including a group of features that can efficiently approximate the space of original features. On the other hand, employing intrinsic statistical information of data can play a significant role in a feature selection process. Nevertheless, most of the existing feature selection methods founded on the subspace distance are limited in properly fulfilling this objective. To pursue this void, we propose a framework that takes a subspace distance into account which is called "Variance-Covariance subspace distance". The approach gains advantages from the correlation of information included in the features of data, thus determines all the feature subsets whose corresponding Variance-Covariance matrix has the minimum norm property. Consequently, a novel, yet efficient unsupervised feature selection framework is introduced based on the Variance-Covariance distance to handle both the dimensionality reduction and subspace learning tasks. The proposed framework has the ability to exclude those features that have the least variance from the original feature set. Moreover, an efficient update algorithm is provided along with its associated convergence analysis to solve the optimization side of the proposed approach. An extensive number of experiments on nine benchmark datasets are also conducted to assess the performance of our method from which the results demonstrate its superiority over a variety of state-of-the-art unsupervised feature selection methods. The source code is available at https://github.com/SaeedKarami/VCSDFS.

A systems biology approach to pathogenesis of gastric cancer: gene network modeling and pathway analysis

BACKGROUND

Gastric cancer (GC) ranks among the most common malignancies worldwide. This study aimed to find critical genes/pathways in GC pathogenesis.

METHODS

Gene interactions were analyzed, and the protein-protein interaction network was drawn. Then enrichment analysis of the hub genes was performed and network cluster analysis and promoter analysis of the hub genes were done. Age/sex analysis was done on the identified genes.

RESULTS

Eleven hub genes in GC were identified in the current study (ATP5A1, ATP5B, ATP5D, MT-ATP8, COX7A2, COX6C, ND4, ND6, NDUFS3, RPL8, and RPS16), mostly involved in mitochondrial functions. There was no report on the ATP5D, ND6, NDUFS3, RPL8, and RPS16 in GC. Our results showed that the most affected processes in GC are the metabolic processes, and the oxidative phosphorylation pathway was considerably enriched which showed the significance of mitochondria in GC pathogenesis. Most of the affected pathways in GC were also involved in neurodegenerative diseases. Promoter analysis showed that negative regulation of signal transduction might play an important role in GC pathogenesis. In the analysis of the basal expression pattern of the selected genes whose basal expression presented a change during the age, we found that a change in age may be an indicator of changes in disease insurgence and/or progression at different ages.

CONCLUSIONS

These results might open up new insights into GC pathogenesis. The identified genes might be novel diagnostic/prognostic biomarkers or potential therapeutic targets for GC. This work, being based on bioinformatics analysis act as a hypothesis generator that requires further clinical validation.

Multi-objective optimization determines when, which and how to fuse deep networks: An application to predict COVID-19 outcomes

The COVID-19 pandemic has caused millions of cases and deaths and the AI-related scientific community, after being involved with detecting COVID-19 signs in medical images, has been now directing the efforts towards the development of methods that can predict the progression of the disease. This task is multimodal by its very nature and, recently, baseline results achieved on the publicly available AIforCOVID dataset have shown that chest X-ray scans and clinical information are useful to identify patients at risk of severe outcomes. While deep learning has shown superior performance in several medical fields, in most of the cases it considers unimodal data only. In this respect, when, which and how to fuse the different modalities is an open challenge in multimodal deep learning. To cope with these three questions here we present a novel approach optimizing the setup of a multimodal end-to-end model. It exploits Pareto multi-objective optimization working with a performance metric and the diversity score of multiple candidate unimodal neural networks to be fused. We test our method on the AIforCOVID dataset, attaining state-of-the-art results, not only outperforming the baseline performance but also being robust to external validation. Moreover, exploiting XAI algorithms we figure out a hierarchy among the modalities and we extract the features' intra-modality importance, enriching the trust on the predictions made by the model.

Joint learning method with teacher-student knowledge distillation for on-device breast cancer image classification

The deep learning models such as AlexNet, VGG, and ResNet achieved a good performance in classifying the breast cancer histopathological images in BreakHis dataset. However, these models are not practically appropriate due to their computational complexity and too many parameters; as a result, they are rarely utilized on devices with limited computational resources. This paper develops a lightweight learning model based on knowledge distillation to classify the histopathological images of breast cancer in BreakHis. This method employs two teacher models based on VGG and ResNext to train two student models, which are similar to the teacher models in development but have fewer deep layers. In the proposed method, the adaptive joint learning approach is adopted to transfer the knowledge in the final-layer output of a teacher model along with the feature maps of its middle layers as the dark knowledge to a student model. According to the experimental results, the student model designed by ResNeXt architecture obtained the recognition rate 97.09% for all histopathological images. In addition, this model has ∼69.40 million fewer parameters, ∼0.93 G less GPU memory use, and 268.17 times greater compression rate than its teacher model. While in the student model the recognition rate merely dropped down to 1.75%. The comparisons indicated that the student model had a rather acceptable outputs compared with state-of-the-art methods in classifying the images of breast cancer in BreakHis.

Deep learning based classification of multi-label chest X-ray images via dual-weighted metric loss

-Thoracic disease, like many other diseases, can lead to complications. Existing multi-label medical image learning problems typically include rich pathological information, such as images, attributes, and labels, which are crucial for supplementary clinical diagnosis. However, the majority of contemporary efforts exclusively focus on regression from input to binary labels, ignoring the relationship between visual features and semantic vectors of labels. In addition, there is an imbalance in data amount between diseases, which frequently causes intelligent diagnostic systems to make erroneous disease predictions. Therefore, we aim to improve the accuracy of the multi-label classification of chest X-ray images. Chest X-ray14 pictures were utilized as the multi-label dataset for the experiments in this study. By fine-tuning the ConvNeXt network, we got visual vectors, which we combined with semantic vectors encoded by BioBert to map the two different forms of features into a common metric space and made semantic vectors the prototype of each class in metric space. The metric relationship between images and labels is then considered from the image level and disease category level, respectively, and a new dual-weighted metric loss function is proposed. Finally, the average AUC score achieved in the experiment reached 0.826, and our model outperformed the comparison models.

BUPNN: Manifold Learning Regularizer-Based Blood Usage Prediction Neural Network for Blood Centers

Blood centers are an essential component of the healthcare system, as timely blood collection, processing, and efficient blood dispatch are critical to the treatment of patients and the performance of the entire healthcare system. At the same time, an efficient blood dispatching system through the high-precision predictive capability of artificial intelligence is crucial for the efficiency improvement of the blood centers. However, the current artificial intelligence (AI) models for predicting blood usage do not meet the needs of blood centers. The challenges of AI models mainly include lower generalization ability in different hospitals, limited stability under missing values, and low interpretability. An artificial neural network-based model named the blood usage prediction neural network (BUPNN) has been developed to address these challenges. BUPNN includes a novel similarity-based manifold regularizer that aims to enhance network mapping consistency and, thus, overcome the domain bise of different hospitals. Moreover, BUPNN diminishes the performance degradation caused by missing values through data enhancement. Experimental results on a large amount of accurate data demonstrate that BUPNN outperforms the baseline method in classification and regression tasks and excels in generalization and consistency. Moreover, BUPNN has solid potential to be interpreted. Therefore, the decision-making process of BUPNN is explored to the extent that it acts as an aid to the experts in the blood center.

Future forecasting prediction of Covid-19 using hybrid deep learning algorithm

Due the quick spread of coronavirus disease 2019 (COVID-19), identification of that disease, prediction of mortality rate and recovery rate are considered as one of the critical challenges in the whole world. The occurrence of COVID-19 dissemination beyond the world is analyzed in this research and an artificial-intelligence (AI) based deep learning algorithm is suggested to detect positive cases of COVID19 patients, mortality rate and recovery rate using real-world datasets. Initially, the unwanted data like prepositions, links, hashtags etc., are removed using some pre-processing techniques. After that, term frequency inverse-term frequency (TF-IDF) andBag of Words (BoW) techniques are utilized to extract the features from pre-processed dataset. Then, Mayfly Optimization (MO) algorithm is performed to pick the relevant features from the set of features. Finally, two deep learning procedures, ResNet model and GoogleNet model, are hybridized to achieve the prediction process. Our system examines two different kinds of publicly available text datasets to identify COVID-19 disease as well as to predict mortality rate and recovery rate using those datasets. There are four different datasets are taken to analyse the performance, in which the proposed method achieves 97.56% accuracy which is 1.40% greater than Linear Regression (LR) and Multinomial Naive Bayesian (MNB), 3.39% higher than Random Forest (RF) and Stochastic gradient boosting (SGB) as well as 5.32% higher than Decision tree (DT) and Bagging techniques if first dataset. When compared to existing machine learning models, the simulation result indicates that a proposed hybrid deep learning method is valuable in corona virus identification and future mortality forecast study.

Comparative Study on International Research Hotspots and National-Level Policy Keywords of Dynamic Disaster Monitoring and Early Warning in China (2000-2021)

For more than 20 years, disaster dynamic monitoring and early warning have achieved orderly and sustainable development in China, forming a systematic academic research system and top-down policy design, which are inseparable from the research of China's scientific community and the promotion of government departments. In the past, most of the research on dynamic disaster monitoring and early warning focused on specific research in a certain field, scene, and discipline, while a few studies focused on research review or policy analysis, and few studies combined macro and meso research reviews in academia with national policy analysis for comparative analysis. It is necessary and urgent to explore the interaction between scholars' research and policy deployment, which can bring theoretical contributions and policy references to the top-down design, implementation promotion, and academic research of China's dynamic disaster monitoring and early warning. Based on 608 international research articles on dynamic disaster monitoring and early warning published by Chinese scholars from 2000-2021 and 187 national policy documents published during this period, this paper conducts a comparative analysis between the knowledge maps of international research hotspots and the co-occurrence maps of policy keywords on dynamic disaster monitoring and early warning. The research shows that in the stage of initial development (2000-2007), international research articles are few and focused, and research hotspots are somewhat alienated from policy keywords. In the stage of rising development (2008-2015), after the Wenchuan earthquake, research hotspots are closely related to policy keywords, mainly in the fields of geology, engineering disasters, meteorological disasters, natural disasters, etc. Meanwhile, research hotspots also focus on cutting-edge technologies and theories, while national-level policy keywords focus more on overall governance and macro promotion, but the two are gradually closely integrated. In the stage of rapid development (2016-2021), with the continuous attention and policy promotion of the national government, the establishment of the Ministry of Emergency Management, and the gradual establishment and improvement of the disaster early warning and monitoring system, research hotspots and policy keywords are integrated and overlapped with each other, realizing the organic linkage and mutual promotion between academic research and political deployment. The motivation, innovation, integration, and transformation of dynamic disaster monitoring and early warning are promoted by both policy and academic research. The institutions that issue policies at the national level include the State Council and relevant departments, the Ministry of Emergency Management, the Ministry of Water Resources, and other national ministries and commissions. The leading affiliated institutions of scholars' international research include China University of Mining and Technology, Chinese Academy of Sciences, Wuhan University, Shandong University of Science and Technology, and other institutions. The disciplines involved are mainly multidisciplinary geosciences, environmental sciences, electrical and electronic engineering, remote sensing, etc. It is worth noting that in the past two to three years, research and policies focusing on COVID-19, public health, epidemic prevention, environmental governance, and emergency management have gradually increased.

Face mask detection and social distance monitoring system for COVID-19 pandemic

Coronavirus triggers several respirational infections such as sneezing, coughing, and pneumonia, which transmit humans to humans through airborne droplets. According to the guidelines of the World Health Organization, the spread of COVID-19 can be mitigated by avoiding public interactions in proximity and following standard operating procedures (SOPs) including wearing a face mask and maintaining social distancing in schools, shopping malls, and crowded areas. However, enforcing the adaptation of these SOPs on a larger scale is still a challenging task. With the emergence of deep learning-based visual object detection networks, numerous methods have been proposed to perform face mask detection on public spots. However, these methods require a huge amount of data to ensure robustness in real-time applications. Also, to the best of our knowledge, there is no standard outdoor surveillance-based dataset available to ensure the efficacy of face mask detection and social distancing methods in public spots. To this end, we present a large-scale dataset comprising of 10,000 outdoor images categorized into a binary class labeling i.e., face mask, and non-face masked people to accelerate the development of automated face mask detection and social distance measurement on public spots. Alongside, we also present an end-to-end pipeline to perform real-time face mask detection and social distance measurement in an outdoor environment. Initially, existing state-of-the-art single and multi-stage object detection networks are fine-tuned on the proposed dataset to evaluate their performance in terms of accuracy and inference time. Based on better performance, YOLO-v3 architecture is further optimized by tuning its feature extraction and region proposal generation layers to improve the performance in real-time applications. Our results indicate that the presented pipeline performed better than the baseline version, showing an improvement of 5.3% in terms of accuracy.

Leveraging Reddit for Suicidal Ideation Detection: A Review of Machine Learning and Natural Language Processing Techniques

Suicide is a major public-health problem that exists in virtually every part of the world. Hundreds of thousands of people commit suicide every year. The early detection of suicidal ideation is critical for suicide prevention. However, there are challenges associated with conventional suicide-risk screening methods. At the same time, individuals contemplating suicide are increasingly turning to social media and online forums, such as Reddit, to express their feelings and share their struggles with suicidal thoughts. This prompted research that applies machine learning and natural language processing techniques to detect suicidality among social media and forum users. The objective of this paper is to investigate methods employed to detect suicidal ideations on the Reddit forum. To achieve this objective, we conducted a literature review of the recent articles detailing machine learning and natural language processing techniques applied to Reddit data to detect the presence of suicidal ideations. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, we selected 26 recent studies, published between 2018 and 2022. The findings of the review outline the prevalent methods of data collection, data annotation, data preprocessing, feature engineering, model development, and evaluation. Furthermore, we present several Reddit-based datasets utilized to construct suicidal ideation detection models. Finally, we conclude by discussing the current limitations and future directions in the research of suicidal ideation detection.

Decoding clinical biomarker space of COVID-19: Exploring matrix factorization-based feature selection methods

One of the most critical challenges in managing complex diseases like COVID-19 is to establish an intelligent triage system that can optimize the clinical decision-making at the time of a global pandemic. The clinical presentation and patients' characteristics are usually utilized to identify those patients who need more critical care. However, the clinical evidence shows an unmet need to determine more accurate and optimal clinical biomarkers to triage patients under a condition like the COVID-19 crisis. Here we have presented a machine learning approach to find a group of clinical indicators from the blood tests of a set of COVID-19 patients that are predictive of poor prognosis and morbidity. Our approach consists of two interconnected schemes: Feature Selection and Prognosis Classification. The former is based on different Matrix Factorization (MF)-based methods, and the latter is performed using Random Forest algorithm. Our model reveals that Arterial Blood Gas (ABG) O₂ Saturation and C-Reactive Protein (CRP) are the most important clinical biomarkers determining the poor prognosis in these patients. Our approach paves the path of building quantitative and optimized clinical management systems for COVID-19 and similar diseases.

Classification of acute myeloid leukemia M1 and M2 subtypes using machine learning

BACKGROUND

Classification of acute myeloid leukemia (AML) relies on manual analysis of bone marrow or peripheral blood smear images. We aimed to construct a machine learning model for automatic classification of AML-M1 and M2 subtypes in bone marrow smear images.

METHODS

Bone marrow smear images of AML patients were extracted from the Cancer Imaging Archive (TCIA) open database. Classification criteria of AML subtypes were based on the French-American-British (FAB) classification system. Random forest method and broad learning system (BLS) were used to develop the classification model. Morphological features, radiomics features, and clinical features were extracted. The performance of the classification model was evaluated by calculating accuracy, precision, recall, F1-score, and area under the curve (AUC). A total of 50 bone marrow smear images (AML-M1, 31 cases; AML-M2, 19 cases) with 500 slices were included in this study.

RESULTS

A total of 43 morphological features, 276 radiomics features, and 1 clinical feature were extracted. Finally, 9 variables including 2 morphological features, 6 radiomics features, and 1 clinical feature were selected into the classification model. The best classification performance was observed in the random forest model with 9 variables, with the average accuracy, AUC, F1-score, recall, and precision of the model being 0.998 ± 0.003, 0.998 ± 0.004, 0.998 ± 0.004, 0.996 ± 0.009, and 1 ± 0, respectively.

CONCLUSION

The random forest model performed well for the classification of AML-M1 and M2, which may provide a tool for clinicians to classify AML-M1 and M2.