Multimodal Disentangled Variational Autoencoder with Game Theoretic Interpretability for Glioma grading

Bimodal PET/MRI generative reconstruction based on VAE architectures

Objective.In this study, we explore positron emission tomography (PET)/magnetic resonance imaging (MRI) joint reconstruction within a deep learning framework, introducing a novel synergistic method.Approach.We propose a new approach based on a variational autoencoder (VAE) constraint combined with the alternating direction method of multipliers (ADMM) optimization technique. We explore three VAE architectures, joint VAE, product of experts-VAE and multimodal JS divergence (MMJSD), to determine the optimal latent representation for the two modalities. We then trained and evaluated the architectures on a brain PET/MRI dataset.Main results.We showed that our approach takes advantage of each modality sharing information to each other, which results in improved peak signal-to-noise ratio and structural similarity as compared with traditional reconstruction, particularly for short acquisition times. We find that the one particular architecture, MMJSD, is the most effective for our methodology.Significance.The proposed method outperforms conventional approaches especially in noisy and undersampled conditions by making use of the two modalities together to compensate for the missing information.

Disentangled global and local features of multi-source data variational autoencoder: An interpretable model for diagnosing IgAN via multi-source Raman spectral fusion techniques

A single Raman spectrum reflects limited molecular information. Effective fusion of the Raman spectra of serum and urine source domains helps to obtain richer feature information. However, most of the current studies on immunoglobulin A nephropathy (IgAN) based on Raman spectroscopy are based on small sample data and low signal-to-noise ratio. If a multi-source data fusion strategy is directly adopted, it may even reduce the accuracy of disease diagnosis. To this end, this paper proposes a data enhancement and spectral optimization method based on variational autoencoders to obtain reconstructed Raman spectra with doubled sample size and improved signal-to-noise ratio. In the diagnosis of IgAN in multi-source domain Raman spectra, this paper builds a global and local feature decoupled variational autoencoder (DMSGL-VAE) model based on multi-source data. First, the statistical features after spectral segmentation are extracted, and the latent variables obtained by the variational encoder are decoupled through the decoupling module. The global representation and local representation obtained represent the global shared information and local unique information of the serum and urine source domains, respectively. Then, the cross-source reconstruction loss and decoupling loss are used to constrain the decoupling, and the effectiveness of the decoupling is proved quantitatively and qualitatively. Finally, the features of different source domains were integrated to diagnose IgAN, and the results were analyzed for important features using the SHapley Additive exPlanations algorithm. The experimental results showed that the AUC value of the DMSGL-VAE model for diagnosing IgAN on the test set was as high as 0.9958. The SHAP algorithm was used to further prove that proteins, hydroxybutyrate, and guanine are likely to be common biological fingerprint substances for the diagnosis of IgAN by serum and urine Raman spectroscopy. In summary, the DMSGL-VAE model designed based on Raman spectroscopy in this paper can achieve rapid, non-invasive, and accurate screening of IgAN in terms of classification performance. And interpretable analysis may help doctors further understand IgAN and make more efficient diagnostic measures in the future.

NETosis Genes and Pathomic Signature: A Novel Prognostic Marker for Ovarian Serous Cystadenocarcinoma

To evaluate the prognostic significance and molecular mechanism of NETosis markers in ovarian serous cystadenocarcinoma (OSC), we constructed a machine learning-based pathomic model utilizing hematoxylin and eosin (H&E) slides. We analyzed 333 patients with OSC from The Cancer Genome Atlas for prognostic-related neutrophil extracellular trap formation (NETosis) genes through bioinformatics analysis. Pathomic features were extracted from 54 cases with complete pathological images, genetic matrices, and clinical information. Two pathomic prognostic models were constructed using support vector machine (SVM) and logistic regression (LR) algorithms. Additionally, we established a predictive scoring system that integrated pathomic scores based on the NETcluster subtypes and clinical signature. We identified four NETosis genes significantly correlated with OSC prognosis, which were functionally associated with immune response, somatic mutations, tumor invasion, and metastasis. Five robust pathomic features were selected for overall survival prediction. The LR and SVM pathomic models demonstrated strong predictive performance for the NETcluster subtype classification through five-fold cross-validation. Time-dependent ROC analysis revealed excellent prognostic capability of the LR pathomic model's score for the overall survival (AUC values of 0.658, 0.761, and 0.735 at 36, 48, and 60 months, respectively), further validated by Kaplan-Meier analysis. The expression levels of NETosis genes greatly affected OSC patients' prognoses. The pathomic analysis of H&E slide pathological images provides an effective approach for predicting both NETcluster subtype and overall survival in OSC patients.

Multi-Loss Disentangled Generative-Discriminative Learning for Multimodal Representation in Schizophrenia

Schizophrenia (SCZ) is a multifactorial mental illness, thus it will be beneficial for exploring this disease using multimodal data, including functional magnetic resonance imaging (fMRI), genes, and the gut microbiome. Previous studies reported combining multimodal data can offer complementary information for better depicting the abnormalities of SCZ. However, the existing multimodal-based methods have multiple limitations. First, most approaches cannot fully use the relationships among different modalities for the downstream tasks. Second, representing multimodal data by the modality-common and modality-specific components can improve the performance of multimodal analysis but often be ignored. Third, most methods conduct the model for classification or regression, thus a unified model is needed for finishing these tasks simultaneously. To this end, a multi-loss disentangled generative-discriminative learning (MDGDL) model was developed to tackle these issues. Specifically, using disentangled learning method, the genes and gut microbial biomarkers were represented and separated into two modality-specific vectors and one modality-common vector. Then, a generative-discriminative framework was introduced to uncover the relationships between fMRI features and these three latent vectors, further producing the attentive vectors, which can help fMRI features for the downstream tasks. To validate the performance of MDGDL, an SCZ classification task and a cognitive score regression task were conducted. Results showed the MDGDL achieved superior performance and identified the most important multimodal biomarkers for the SCZ. Our proposed model could be a supplementary approach for multimodal data analysis. Based on this method, we could analyze the SCZ by combining multimodal data, and further obtain some interesting findings.

Integrated diagnosis of glioma based on magnetic resonance images with incomplete ground truth labels

BACKGROUND

Since the 2016 WHO guidelines, glioma diagnosis has entered an era of integrated diagnosis, combining tissue pathology and molecular pathology. The WHO has focused on promoting the application of molecular diagnosis in the classification of central nervous system tumors. Genetic information such as IDH1 and 1p/19q are important molecular markers, and pathological grading is also a key clinical indicator. However, obtaining genetic pathology labels is more costly than conventional MRI images, resulting in a large number of missing labels in realistic modeling.

METHOD

We propose a training strategy based on label encoding and a corresponding loss function to enable the model to effectively utilize data with missing labels. Additionally, we integrate a graph model with genes and pathology-related clinical prior knowledge into the ResNet backbone to further improve the efficacy of diagnosis. Ten-fold cross-validation experiments were conducted on a large dataset of 1072 patients.

RESULTS

The classification area under the curve (AUC) values are 0.93, 0.91, and 0.90 for IDH1, 1p/19q status, and grade (LGG/HGG), respectively. When the label miss rate reached 59.3 %, the method improved the AUC by 0.09, 0.10, and 0.04 for IDH1, 1p/19q, and pathological grade, respectively, compared to the same backbone without the missing label strategy.

CONCLUSIONS

Our method effectively utilizes data with missing labels and integrates clinical prior knowledge, resulting in improved diagnostic performance for glioma genetic and pathological markers, even with high rates of missing labels.

A deep learning-based pipeline for developing multi-rib shape generative model with populational percentiles or anthropometrics as predictors

Rib cross-sectional shapes (characterized by the outer contour and cortical bone thickness) affect the rib mechanical response under impact loading, thereby influence the rib injury pattern and risk. A statistical description of the rib shapes or their correlations to anthropometrics is a prerequisite to the development of numerical human body models representing target demographics. Variational autoencoders (VAE) as anatomical shape generators remain to be explored in terms of utilizing the latent vectors to control or interpret the representativeness of the generated results. In this paper, we propose a pipeline for developing a multi-rib cross-sectional shape generative model from CT images, which consists of the achievement of rib cross-sectional shape data from CT images using an anatomical indexing system and regular grids, and a unified framework to fit shape distributions and associate shapes to anthropometrics for different rib categories. Specifically, we collected CT images including 3193 ribs, surface regular grid is generated for each rib based on anatomical coordinates, the rib cross-sectional shapes are characterized by nodal coordinates and cortical bone thickness. The tensor structure of shape data based on regular grids enable the implementation of CNNs in the conditional variational autoencoder (CVAE). The CVAE is trained against an auxiliary classifier to decouple the low-dimensional representations of the inter- and intra- variations and fit each intra-variation by a Gaussian distribution simultaneously. Random tree regressors are further leveraged to associate each continuous intra-class space with the corresponding anthropometrics of the subjects, i.e., age, height and weight. As a result, with the rib class labels and the latent vectors sampled from Gaussian distributions or predicted from anthropometrics as the inputs, the decoder can generate valid rib cross-sectional shapes of given class labels (male/female, 2nd to 11th ribs) for arbitrary populational percentiles or specific age, height and weight, which paves the road for future biomedical and biomechanical studies considering the diversity of rib shapes across the population.

Physical frailty identification using machine learning to explore the 5-item FRAIL scale, Cardiovascular Health Study index, and Study of Osteoporotic Fractures index

Background

Physical frailty is an important issue in aging societies. Three models of physical frailty assessment, the 5-Item fatigue, resistance, ambulation, illness and loss of weight (FRAIL); Cardiovascular Health Study (CHS); and Study of Osteoporotic Fractures (SOF) indices, have been regularly used in clinical and research studies. However, no previous studies have investigated the predictive ability of machine learning (ML) for physical frailty assessment. The aim was to use two ML algorithms, random forest (RF) and extreme gradient boosting (XGBoost), to predict these three physical frailty assessment models.

Materials and methods

Questionnaires regarding demographic characteristics, lifestyle habits, living environment, and physical frailty assessment were answered by 445 participants aged 60 years and above. The RF and XGBoost algorithms were used to assess their scores for the three physical frailty indices. Furthermore, feature importance and Shapley additive explanations (SHAP) were used to determine the important physical frailty factors.

Results

The XGBoost algorithm obtained higher accuracy for predicting the three physical frailty indices; the areas under the curve obtained by the XGBoost algorithm for the 5-Item FRAIL, CHS, and SOF indices were 0.84. 0.79, and 0.69, respectively. The feature importance and SHAP of the XGBoost algorithm revealed that systolic blood pressure, diastolic blood pressure, age, and body mass index play important roles in all three physical frailty models.

Conclusion

The XGBoost algorithm has a more accurate predictive rate than RF across all three physical frailty assessments. Thus, ML can be a useful tool for the early detection of physical frailty.

Dual-space disentangled-multimodal network (DDM-net) for glioma diagnosis and prognosis with incomplete pathology and genomic data

Objective. Effective fusion of histology slides and molecular profiles from genomic data has shown great potential in the diagnosis and prognosis of gliomas. However, it remains challenging to explicitly utilize the consistent-complementary information among different modalities and create comprehensive representations of patients. Additionally, existing researches mainly focus on complete multi-modality data and usually fail to construct robust models for incomplete samples.Approach. In this paper, we propose adual-space disentangled-multimodal network (DDM-net)for glioma diagnosis and prognosis. DDM-net disentangles the latent features generated by two separate variational autoencoders (VAEs) into common and specific components through a dual-space disentangled approach, facilitating the construction of comprehensive representations of patients. More importantly, DDM-net imputes the unavailable modality in the latent feature space, making it robust to incomplete samples.Main results. We evaluated our approach on the TCGA-GBMLGG dataset for glioma grading and survival analysis tasks. Experimental results demonstrate that the proposed method achieves superior performance compared to state-of-the-art methods, with a competitive AUC of 0.952 and a C-index of 0.768.Significance. The proposed model may help the clinical understanding of gliomas and can serve as an effective fusion model with multimodal data. Additionally, it is capable of handling incomplete samples, making it less constrained by clinical limitations.

Patient-specific game-based transfer method for Parkinson's disease severity prediction

Dysphonia is one of the early symptoms of Parkinson's disease (PD). Most existing methods use feature selection methods to find the optimal subset of voice features for all PD patients. Few have considered the heterogeneity between patients, which implies the need to provide specific prediction models for different patients. However, building the specific model faces the challenge of small sample size, which makes it lack generalization ability. Instance transfer is an effective way to solve this problem. Therefore, this paper proposes a patient-specific game-based transfer (PSGT) method for PD severity prediction. First, a selection mechanism is used to select PD patients with similar disease trends to the target patient from the source domain, which reduces the risk of negative transfer. Then, the contribution of the transferred subjects and their instances to the disease estimation of the target subject is fairly evaluated by the Shapley value, which improves the interpretability of the method. Next, the proportion of valid instances in the transferred subjects is determined, and the instances with higher contribution are transferred to further reduce the difference between the transferred instance subset and the target subject. Finally, the selected subset of instances is added to the training set of the target subject, and the extended data is fed into the random forest to improve the performance of the method. Parkinson's telemonitoring dataset is used to evaluate the feasibility and effectiveness. The mean values of mean absolute error, root mean square error, and volatility obtained by predicting motor-UPDRS and total-UPDRS for target patients are 1.59, 1.95, 1.56 and 1.98, 2.54, 1.94, respectively. Experiment results show that the PSGT has better performance in both prediction error and stability over compared methods.

Immune, metabolic landscapes of prognostic signatures for lung adenocarcinoma based on a novel deep learning framework

Lung adenocarcinoma (LUAD) is a malignant tumor with high lethality, and the aim of this study was to identify promising biomarkers for LUAD. Using the TCGA-LUAD dataset as a discovery cohort, a novel joint framework VAEjMLP based on variational autoencoder (VAE) and multilayer perceptron (MLP) was proposed. And the Shapley Additive Explanations (SHAP) method was introduced to evaluate the contribution of feature genes to the classification decision, which helped us to develop a biologically meaningful biomarker potential scoring algorithm. Nineteen potential biomarkers for LUAD were identified, which were involved in the regulation of immune and metabolic functions in LUAD. A prognostic risk model for LUAD was constructed by the biomarkers HLA-DRB1, SCGB1A1, and HLA-DRB5 screened by Cox regression analysis, dividing the patients into high-risk and low-risk groups. The prognostic risk model was validated with external datasets. The low-risk group was characterized by enrichment of immune pathways and higher immune infiltration compared to the high-risk group. While, the high-risk group was accompanied by an increase in metabolic pathway activity. There were significant differences between the high- and low-risk groups in metabolic reprogramming of aerobic glycolysis, amino acids, and lipids, as well as in angiogenic activity, epithelial-mesenchymal transition, tumorigenic cytokines, and inflammatory response. Furthermore, high-risk patients were more sensitive to Afatinib, Gefitinib, and Gemcitabine as predicted by the pRRophetic algorithm. This study provides prognostic signatures capable of revealing the immune and metabolic landscapes for LUAD, and may shed light on the identification of other cancer biomarkers.

Brain Structure-Function Fusing Representation Learning Using Adversarial Decomposed-VAE for Analyzing MCI

Integrating the brain structural and functional connectivity features is of great significance in both exploring brain science and analyzing cognitive impairment clinically. However, it remains a challenge to effectively fuse structural and functional features in exploring the complex brain network. In this paper, a novel brain structure-function fusing-representation learning (BSFL) model is proposed to effectively learn fused representation from diffusion tensor imaging (DTI) and resting-state functional magnetic resonance imaging (fMRI) for mild cognitive impairment (MCI) analysis. Specifically, the decomposition-fusion framework is developed to first decompose the feature space into the union of the uniform and unique spaces for each modality, and then adaptively fuse the decomposed features to learn MCI-related representation. Moreover, a knowledge-aware transformer module is designed to automatically capture local and global connectivity features throughout the brain. Also, a uniform-unique contrastive loss is further devised to make the decomposition more effective and enhance the complementarity of structural and functional features. The extensive experiments demonstrate that the proposed model achieves better performance than other competitive methods in predicting and analyzing MCI. More importantly, the proposed model could be a potential tool for reconstructing unified brain networks and predicting abnormal connections during the degenerative processes in MCI.

CDNet: Contrastive Disentangled Network for Fine-Grained Image Categorization of Ocular B-Scan Ultrasound

Precise and rapid categorization of images in the B-scan ultrasound modality is vital for diagnosing ocular diseases. Nevertheless, distinguishing various diseases in ultrasound still challenges experienced ophthalmologists. Thus a novel contrastive disentangled network (CDNet) is developed in this work, aiming to tackle the fine-grained image categorization (FGIC) challenges of ocular abnormalities in ultrasound images, including intraocular tumor (IOT), retinal detachment (RD), posterior scleral staphyloma (PSS), and vitreous hemorrhage (VH). Three essential components of CDNet are the weakly-supervised lesion localization module (WSLL), contrastive multi-zoom (CMZ) strategy, and hyperspherical contrastive disentangled loss (HCD-Loss), respectively. These components facilitate feature disentanglement for fine-grained recognition in both the input and output aspects. The proposed CDNet is validated on our ZJU Ocular Ultrasound Dataset (ZJUOUSD), consisting of 5213 samples. Furthermore, the generalization ability of CDNet is validated on two public and widely-used chest X-ray FGIC benchmarks. Quantitative and qualitative results demonstrate the efficacy of our proposed CDNet, which achieves state-of-the-art performance in the FGIC task.

SASG-GCN: Self-Attention Similarity Guided Graph Convolutional Network for Multi-Type Lower-Grade Glioma Classification

Identifying the subtypes of low-grade glioma (LGG) can help prevent brain tumor progression and patient death. However, the complicated non-linear relationship and high dimensionality of 3D brain MRI limit the performance of machine learning methods. Therefore, it is important to develop a classification method that can overcome these limitations. This study proposes a self-attention similarity-guided graph convolutional network (SASG-GCN) that uses the constructed graphs to complete multi-classification (tumor-free (TF), WG, and TMG). In the pipeline of SASG-GCN, we use a convolutional deep belief network and a self-attention similarity-based method to construct the vertices and edges of the constructed graphs at 3D MRI level, respectively. The multi-classification experiment is performed in a two-layer GCN model. SASG-GCN is trained and evaluated on 402 3D MRI images which are produced from the TCGA-LGG dataset. Empirical tests demonstrate that SASG-GCN accurately classifies the subtypes of LGG. The accuracy of SASG-GCN achieves 93.62%, outperforming several other state-of-the-art classification methods. In-depth discussion and analysis reveal that the self-attention similarity-guided strategy improves the performance of SASG-GCN. The visualization revealed differences between different gliomas.

A correlation graph attention network for classifying chromosomal instabilities from histopathology whole-slide images

The chromosome instability (CIN) is one of the hallmarks of cancer and is closely related to tumor metastasis. However, the sheer size and resolution of histopathology whole-slide images (WSIs) already challenges the capabilities of computational pathology. In this study, we propose a correlation graph attention network (MLP-GAT) that can construct graphs for classifying multi-type CINs from the WSIs of breast cancer. We construct a WSIs dataset of breast cancer from the Cancer Genome Atlas Breast Invasive Carcinoma (TCGA-BRCA). Extensive experiments show that MLP-GAT far outperforms accepted state-of-the-art methods and demonstrate the advantages of the constructed graph networks for analyzing WSI data. The visualization shows the difference among the tiles in a WSI. Furthermore, the generalization performance of the proposed method was verified on the stomach cancer. This study provides guidance for studying the relationship between CIN and cancer from the perspective of image phenotype.

Generative AI for brain image computing and brain network computing: a review

Recent years have witnessed a significant advancement in brain imaging techniques that offer a non-invasive approach to mapping the structure and function of the brain. Concurrently, generative artificial intelligence (AI) has experienced substantial growth, involving using existing data to create new content with a similar underlying pattern to real-world data. The integration of these two domains, generative AI in neuroimaging, presents a promising avenue for exploring various fields of brain imaging and brain network computing, particularly in the areas of extracting spatiotemporal brain features and reconstructing the topological connectivity of brain networks. Therefore, this study reviewed the advanced models, tasks, challenges, and prospects of brain imaging and brain network computing techniques and intends to provide a comprehensive picture of current generative AI techniques in brain imaging. This review is focused on novel methodological approaches and applications of related new methods. It discussed fundamental theories and algorithms of four classic generative models and provided a systematic survey and categorization of tasks, including co-registration, super-resolution, enhancement, classification, segmentation, cross-modality, brain network analysis, and brain decoding. This paper also highlighted the challenges and future directions of the latest work with the expectation that future research can be beneficial.

Deep multimodality-disentangled association analysis network for imaging genetics in neurodegenerative diseases

Imaging genetics is a crucial tool that is applied to explore potentially disease-related biomarkers, particularly for neurodegenerative diseases (NDs). With the development of imaging technology, the association analysis between multimodal imaging data and genetic data is gradually being concerned by a wide range of imaging genetics studies. However, multimodal data are fused first and then correlated with genetic data in traditional methods, which leads to an incomplete exploration of their common and complementary information. In addition, the inaccurate formulation in the complex relationships between imaging and genetic data and information loss caused by missing multimodal data are still open problems in imaging genetics studies. Therefore, in this study, a deep multimodality-disentangled association analysis network (DMAAN) is proposed to solve the aforementioned issues and detect the disease-related biomarkers of NDs simultaneously. First, the imaging data are nonlinearly projected into a latent space and imaging representations can be achieved. The imaging representations are further disentangled into common and specific parts by using a multimodal-disentangled module. Second, the genetic data are encoded to achieve genetic representations, and then, the achieved genetic representations are nonlinearly mapped to the common and specific imaging representations to build nonlinear associations between imaging and genetic data through an association analysis module. Moreover, modality mask vectors are synchronously synthesized to integrate the genetic and imaging data, which helps the following disease diagnosis. Finally, the proposed method achieves reasonable diagnosis performance via a disease diagnosis module and utilizes the label information to detect the disease-related modality-shared and modality-specific biomarkers. Furthermore, the genetic representation can be used to impute the missing multimodal data with our learning strategy. Two publicly available datasets with different NDs are used to demonstrate the effectiveness of the proposed DMAAN. The experimental results show that the proposed DMAAN can identify the disease-related biomarkers, which suggests the proposed DMAAN may provide new insights into the pathological mechanism and early diagnosis of NDs. The codes are publicly available at https://github.com/Meiyan88/DMAAN.

Variational autoencoder-based estimation of chronological age and changes in morphological features of teeth

This study led to the development of a variational autoencoder (VAE) for estimating the chronological age of subjects using feature values extracted from their teeth. Further, it determined how given teeth images affected the estimation accuracy. The developed VAE was trained with the first molar and canine tooth images, and a parallel VAE structure was further constructed to extract common features shared by the two types of teeth more effectively. The encoder of the VAE was combined with a regression model to estimate the age. To determine which parts of the tooth images were more or less important when estimating age, a method of visualizing the obtained regression coefficient using the decoder of the VAE was developed. The developed age estimation model was trained using data from 910 individuals aged 10-79. This model showed a median absolute error (MAE) of 6.99 years, demonstrating its ability to estimate age accurately. Furthermore, this method of visualizing the influence of particular parts of tooth images on the accuracy of age estimation using a decoder is expected to provide novel insights for future research on explainable artificial intelligence.

A deep learning framework integrating MRI image preprocessing methods for brain tumor segmentation and classification

Glioma grading is critical in treatment planning and prognosis. This study aims to address this issue through MRI-based classification to develop an accurate model for glioma diagnosis. Here, we employed a deep learning pipeline with three essential steps: (1) MRI images were segmented using preprocessing approaches and UNet architecture, (2) brain tumor regions were extracted using segmentation, then (3) high-grade gliomas and low-grade gliomas were classified using the VGG and GoogleNet implementations. Among the additional preprocessing techniques used in conjunction with the segmentation task, the combination of data augmentation and Window Setting Optimization was found to be the most effective tool, resulting in the Dice coefficient of 0.82, 0.91, and 0.72 for enhancing tumor, whole tumor, and tumor core, respectively. While most of the proposed models achieve comparable accuracies of about 93 % on the testing dataset, the pipeline of VGG combined with UNet segmentation obtains the highest accuracy of 97.44 %. In conclusion, the presented architecture illustrates a realistic model for detecting gliomas; moreover, it emphasizes the significance of data augmentation and segmentation in improving model performance.

Automated Diagnosis of COVID-19 Using Deep Supervised Autoencoder With Multi-View Features From CT Images

Accurate and rapid diagnosis of coronavirus disease 2019 (COVID-19) from chest CT scans is of great importance and urgency during the worldwide outbreak. However, radiologists have to distinguish COVID-19 pneumonia from other pneumonia in a large number of CT scans, which is tedious and inefficient. Thus, it is urgently and clinically needed to develop an efficient and accurate diagnostic tool to help radiologists to fulfill the difficult task. In this study, we proposed a deep supervised autoencoder (DSAE) framework to automatically identify COVID-19 using multi-view features extracted from CT images. To fully explore features characterizing CT images from different frequency domains, DSAE was proposed to learn the latent representation by multi-task learning. The proposal was designed to both encode valuable information from different frequency features and construct a compact class structure for separability. To achieve this, we designed a multi-task loss function, which consists of a supervised loss and a reconstruction loss. Our proposed method was evaluated on a newly collected dataset of 787 subjects including COVID-19 pneumonia patients, other pneumonia patients, and normal subjects without abnormal CT findings. Extensive experimental results demonstrated that our proposed method achieved encouraging diagnostic performance and may have potential clinical application for the diagnosis of COVID-19.

A triple-classification for the evaluation of lung nodules manifesting as pure ground-glass sign: a CT-based radiomic analysis

OBJECTIVES

To construct a noninvasive radiomics model for evaluating the pathological degree and an individualized treatment strategy for patients with the manifestation of ground glass nodules (GGNs) on CT images.

METHODS

The retrospective primary cohort investigation included patients with GGNs on CT images who underwent resection between June 2015 and June 2020. The intratumoral regions of interest were segmented semiautomatically, and radiomics features were extracted from the intratumoral and peritumoral regions. After feature selection by ANOVA, Max-Relevance and Min-Redundancy (mRMR) and Least Absolute Shrinkage and Selection Operator (Lasso) regression, a random forest (RF) model was generated. Receiver operating characteristic (ROC) analysis was calculated to evaluate each classification. Shapley additive explanations (SHAP) was applied to interpret the radiomics features.

RESULTS

In this study, 241 patients including atypical adenomatous hyperplasia (AAH) or adenocarcinoma in situ (AIS) (n = 72), minimally invasive adenocarcinoma (MIA) (n = 83) and invasive adenocarcinoma (IAC) (n = 86) were selected for radiomics analysis. Three intratumoral radiomics features and one peritumoral feature were finally identified by the triple RF classifier with an average area under the curve (AUC) of 0.960 (0.963 for AAH/AIS, 0.940 for MIA, 0.978 for IAC) in the training set and 0.944 (0.955 for AAH/AIS, 0.952 for MIA, 0.926 for IAC) in the testing set for evaluation of the GGNs.

CONCLUSION

The triple classification based on intra- and peritumoral radiomics features derived from the noncontrast CT images had satisfactory performance and may be used as a noninvasive tool for preoperative evaluation of the pure ground-glass nodules and developing of individualized treatment strategies.

Multi-task disentangled autoencoder for time-series data in glucose dynamics

The objective of this study is to propose MD-VAE: a multi-task disentangled variational autoencoders (VAE) for exploring characteristics of latent representations (LR) and exploiting LR for diverse tasks including glucose forecasting, event detection, and temporal clustering.