Human attention guided explainable artificial intelligence for computer vision models.

Explainable artificial intelligence (XAI) has been increasingly investigated to enhance the transparency of black-box artificial intelligence models, promoting better user understanding and trust. Developing an XAI that is faithful to models and plausible to users is both a necessity and a challenge. This work examines whether embedding human attention knowledge into saliency-based XAI methods for computer vision models could enhance their plausibility and faithfulness. Two novel XAI methods for object detection models, namely FullGrad-CAM and FullGrad-CAM++, were first developed to generate object-specific explanations by extending the current gradient-based XAI methods for image classification models. Using human attention as the objective plausibility measure, these methods achieve higher explanation plausibility. Interestingly, all current XAI methods when applied to object detection models generally produce saliency maps that are less faithful to the model than human attention maps from the same object detection task. Accordingly, human attention-guided XAI (HAG-XAI) was proposed to learn from human attention how to best combine explanatory information from the models to enhance explanation plausibility by using trainable activation functions and smoothing kernels to maximize the similarity between XAI saliency map and human attention map. The proposed XAI methods were evaluated on widely used BDD-100K, MS-COCO, and ImageNet datasets and compared with typical gradient-based and perturbation-based XAI methods. Results suggest that HAG-XAI enhanced explanation plausibility and user trust at the expense of faithfulness for image classification models, and it enhanced plausibility, faithfulness, and user trust simultaneously and outperformed existing state-of-the-art XAI methods for object detection models.

Incidence, clinical features, and survival outcomes of primary malignant lacrimal gland tumors: A population-based analysis.

BACKGROUND: Studies on the epidemiological information and prognosis of primary malignant lacrimal gland tumors (MLGTs) are rare for its low occurrence. The goal of our research was to investigate the epidemiological characteristics and survival outcomes of patients with MLGTs. METHODS: Incidence and demographic information of patients with MLGTs were collected from the Surveillance, Epidemiology, and End Results (SEER) database. To identify independent prognostic factors for disease-specific survival (DSS) and overall survival (OS), univariate and multivariate Cox regression analysis were performed. RESULTS: The overall incidence of primary MLGTs from 1975 to 2020 was 0.413/1,000,000 (according to the 2000 American standard population), with a steadily increasing incidence over years. A total of 964 patients with primary MLGTs were diagnosed, with an average age of 59.3 years. Of these, 53.2% were aged ≥60 years, 57.4% were female, and 77.1% were whites. Multivariate Cox regression analysis demonstrated that year of diagnosis, age, sex, histological type, SEER stage, surgery, and chemotherapy were independent prognostic factors of DSS or OS. CONCLUSIONS: Although primary MLGT is rare, its incidence has steadily increased in the past 46 years, and surgery was related to a better prognosis.

KISS-1 knockdown inhibits cell growth, migration, and invasion in HTR-8/SVneo cells by regulating the GRP54-mediated PI3K/AKT signaling pathway.

Recurrent spontaneous abortions (RSA) affect reproductive health and increase the risk of subsequent abortions. To investigate the role of KISS-1/GPR-54 signaling in RSA progression. Villus tissue was collected from RSA patients, and human trophoblastic HTR-8/SVneo cells were used. KISS-1 and GRP54 levels were detected using RT-qPCR and immunohistochemistry. Western blotting was performed to analyze ZO-1 and ZEB1 levels. Cell proliferation was determined via CCK-8 and cell clone formation assays. Transwell assays were performed to assess cell migration and invasion abilities. KISS-1 was down-regulated in the villus tissues of RSA patients. KISS-1 overexpression dramatically inhibited trophoblast proliferation, migration, and invasion. Mechanistically, ZEB1 expression was down-regulated, whereas ZO-1 expression was up-regulated, after KISS-1 overexpression. GPR54 silencing neutralized the effect of KISS-1 in HTR-8/SVneo cells. Additionally, KISS-1 overexpression inactivated the PI3K/AKT signaling pathway through GRP54. The KISS-1/GPR-54 signaling axis regulates RSA progression by regulating the PI3K/AKT signaling pathway.

Facile Synthesis of Ce-MOF for the Removal of Phosphate, Fluoride, and Arsenic.

Ce-MOF was synthesized by a solvothermal synthesis method and was used to simultaneously remove phosphate, fluoride and arsenic (V) from water by adsorption. Ce-MOF was characterized by a nitrogen adsorption-desorption isotherm, scanning electron microscopy, and infrared spectroscopy. The effects of initial concentration, adsorption time, adsorption temperature, pH value and adsorbent on the adsorption properties were investigated. A Langmuir isotherm model was used to fit the adsorption data, and the adsorption capacity of phosphate, fluoride, and arsenic (V) was calculated to be 41.2 mg·g-1, 101.8 mg·g-1 and 33.3 mg·g-1, respectively. Compared with the existing commercially available CeO2 and other MOFs, Ce-MOF has a much higher adsorption capacity. Furthermore, after two reuses, the performance of the adsorbent was almost unchanged, indicating it is a stable adsorbent and has good application potential in the field of wastewater treatment.

Single-cell sequencing in primary intraocular tumors: understanding heterogeneity, the microenvironment, and drug resistance.

Retinoblastoma (RB) and uveal melanoma (UM) are the most common primary intraocular tumors in children and adults, respectively. Despite continued increases in the likelihood of salvaging the eyeball due to advancements in local tumor control, prognosis remains poor once metastasis has occurred. Traditional sequencing technology obtains averaged information from pooled clusters of diverse cells. In contrast, single-cell sequencing (SCS) allows for investigations of tumor biology at the resolution of the individual cell, providing insights into tumor heterogeneity, microenvironmental properties, and cellular genomic mutations. SCS is a powerful tool that can help identify new biomarkers for diagnosis and targeted therapy, which may in turn greatly improve tumor management. In this review, we focus on the application of SCS for evaluating heterogeneity, microenvironmental characteristics, and drug resistance in patients with RB and UM.

Epidemiology and survival outcomes of patients with primary intraocular lymphoma: a population-based analysis.

BACKGROUND: Primary intraocular lymphoma (PIOL) is a rare malignancy with a poor prognosis, but its optimal therapy remains unclear. Herein, we aimed to analyze the epidemiology and survival outcomes of PIOL patients based on a population-based cancer registry in the United States. METHODS: Patients diagnosed with PIOL between 1992 and 2018 were identified from the Surveillance Epidemiology and End Results program. The patients were divided into two groups: those aged < 60 years and ≥ 60 years. We used the chi-squared test to analyze the differences between the two groups. Descriptive analyses were performed to analyze epidemiological characteristics and treatment. The likely prognostic factors were analyzed by Kaplan-Meier curves and Cox proportional hazards models. RESULTS: The overall incidence of PIOL was 0.23/1,000,000, which was steadily increasing from 1992 to 2018, with an annual percentage change of 2.35. In total, 326 patients (mean age, 66.1 years) with PIOL were included in this study, 72.1% were aged ≥ 60 years, 84.4% were White, and 60.4% were female. The most common pathological type was diffuse large B-cell lymphoma (DLBCL), but in patients aged < 60 years, extranodal marginal zone lymphoma of mucosa-associated lymphoid tissue was the most common. The disease-specific survival rates were 74.2% and 61.5% 5 and 10 years after diagnosis, respectively. Survival analysis found that surgery, radiation, and chemotherapy did not lead to better prognosis. CONCLUSIONS: PIOL is a rare disease with poor prognosis, and its incidence has been increasing for nearly 30 years. It usually affects people aged ≥ 60 years, and DLBCL is the most common pathological type of PIOL. Patients aged < 60 years and with non-DLBCL type have improved survival. Survival of PIOL has improved in recent years.

Aberrant brain functional networks in type 2 diabetes mellitus: A graph theoretical and support-vector machine approach.

Objective: Type 2 diabetes mellitus (T2DM) is a high risk of cognitive decline and dementia, but the underlying mechanisms are not yet clearly understood. This study aimed to explore the functional connectivity (FC) and topological properties among whole brain networks and correlations with impaired cognition and distinguish T2DM from healthy controls (HC) to identify potential biomarkers for cognition abnormalities. Methods: A total of 80 T2DM and 55 well-matched HC were recruited in this study. Subjects' clinical data, neuropsychological tests and resting-state functional magnetic resonance imaging data were acquired. Whole-brain network FC were mapped, the topological characteristics were analyzed using a graph-theoretic approach, the FC and topological characteristics of the network were compared between T2DM and HC using a general linear model, and correlations between networks and clinical and cognitive characteristics were identified. The support vector machine (SVM) model was used to identify differences between T2DM and HC. Results: In patients with T2DM, FC was higher in two core regions [precuneus/posterior cingulated cortex (PCC)_1 and later prefrontal cortex_1] in the default mode network and lower in bilateral superior parietal lobes (within dorsal attention network), and decreased between the right medial frontal cortex and left auditory cortex. The FC of the right frontal medial-left auditory cortex was positively correlated with the Montreal Cognitive Assessment scales and negatively correlated with the blood glucose levels. Long-range connectivity between bilateral auditory cortex was missing in the T2DM. The nodal degree centrality and efficiency of PCC were higher in T2DM than in HC (P < 0.005). The nodal degree centrality in the PCC in the SVM model was 97.56% accurate in distinguishing T2DM patients from HC, demonstrating the reliability of the prediction model. Conclusion: Functional abnormalities in the auditory cortex in T2DM may be related to cognitive impairment, such as memory and attention, and nodal degree centrality in the PCC might serve as a potential neuroimaging biomarker to predict and identify T2DM.

Preoperative Prediction of Metastasis for Ovarian Cancer Based on Computed Tomography Radiomics Features and Clinical Factors.

BACKGROUND: There is urgent need for an accurate preoperative prediction of metastatic status to optimize treatment for patients with ovarian cancer (OC). The feasibility of predicting the metastatic status based on radiomics features from preoperative computed tomography (CT) images alone or combined with clinical factors were investigated. METHODS: A total of 101 OC patients who underwent primary debulking surgery were enrolled. Radiomics features were extracted from the tumor volumes contoured on CT images with LIFEx package. Mann-Whitney U tests, least absolute shrinkage selection operator (LASSO), and Ridge Regression were applied to select features and to build prediction models. Univariate and regression analysis were applied to select clinical factors for metastatic prediction. The performance of models generated with radiomics features alone, clinical factors, and combined factors were evaluated and compared. RESULTS: Nine radiomics features were screened out from 184 extracted features to classify patients with and without metastasis. Age and cancer antigen 125 (CA125) were the two clinical factors that were associated with metastasis. The area under curves (AUCs) for the radiomics signature, clinical factors model, and combined model were 0.82 (95% CI, 0.66-0.98; sensitivity = 0.90, specificity = 0.70), 0.83 (95% CI, 0.67-0.95; sensitivity = 0.71, specificity = 0.8), and 0.86 (95% CI, 0.72-0.99, sensitivity = 0.81, specificity = 0.8), respectively. CONCLUSIONS: Radiomics features alone or radiomics features combined with clinical factors are feasible and accurate enough to predict the metastatic status for OC patients.

Preoperative Nomogram for Differentiation of Histological Subtypes in Ovarian Cancer Based on Computer Tomography Radiomics.

OBJECTIVES: Non-invasive method to predict the histological subtypes preoperatively is essential for the overall management of ovarian cancer (OC). The feasibility of radiomics in the differentiating of epithelial ovarian cancer (EOC) and non-epithelial ovarian cancer (NEOC) based on computed tomography (CT) images was investigated. METHODS: Radiomics features were extracted from preoperative CT for 101 patients with pathologically proven OC. Radiomics signature was built using the least absolute shrinkage and selection operator (LASSO) logistic regression. A nomogram was developed with the combination of radiomics features and clinical factors to differentiate EOC and NEOC. RESULTS: Eight radiomics features were selected to build a radiomics signature with an area under curve (AUC) of 0.781 (95% confidence interval (CI), 0.666 -0.897) in the discrimination between EOC and NEOC. The AUC of the combined model integrating clinical factors and radiomics features was 0.869 (95% CI, 0.783 -0.955). The nomogram demonstrated that the combined model provides a better net benefit to predict histological subtypes compared with radiomics signature and clinical factors alone when the threshold probability is within a range from 0.43 to 0.97. CONCLUSIONS: Nomogram developed with CT radiomics signature and clinical factors is feasible to predict the histological subtypes preoperative for patients with OC.

SCAU-Net: Spatial-Channel Attention U-Net for Gland Segmentation.

With the development of medical technology, image semantic segmentation is of great significance for morphological analysis, quantification, and diagnosis of human tissues. However, manual detection and segmentation is a time-consuming task. Especially for biomedical image, only experts are able to identify tissues and mark their contours. In recent years, the development of deep learning has greatly improved the accuracy of computer automatic segmentation. This paper proposes a deep learning image semantic segmentation network named Spatial-Channel Attention U-Net (SCAU-Net) based on current research status of medical image. SCAU-Net has an encoder-decoder-style symmetrical structure integrated with spatial and channel attention as plug-and-play modules. The main idea is to enhance local related features and restrain irrelevant features at the spatial and channel levels. Experiments on the gland dataset GlaS and CRAG show that the proposed SCAU-Net model is superior to the classic U-Net model in image segmentation task, with 1% improvement on Dice score and 1.5% improvement on Jaccard score.

Development, Validation and Comparison of Artificial Neural Network Models and Logistic Regression Models Predicting Survival of Unresectable Pancreatic Cancer.

Background: Prediction models for the overall survival of pancreatic cancer remain unsatisfactory. We aimed to explore artificial neural networks (ANNs) modeling to predict the survival of unresectable pancreatic cancer patients. Methods: Thirty-two clinical parameters were collected from 221 unresectable pancreatic cancer patients, and their prognostic ability was evaluated using univariate and multivariate logistic regression. ANN and logistic regression (LR) models were developed on a training group (168 patients), and the area under the ROC curve (AUC) was used for comparison of the ANN and LR models. The models were further tested on the testing group (53 patients), and k-statistics were used for accuracy comparison. Results: We built three ANN models, based on 3, 7, and 32 basic features, to predict 8 month survival. All 3 ANN models showed better performance, with AUCs significantly higher than those from the respective LR models (0.811 vs. 0.680, 0.844 vs. 0.722, 0.921 vs. 0.849, all p < 0.05). The ability of the ANN models to discriminate 8 month survival with higher accuracy than the respective LR models was further confirmed in 53 consecutive patients. Conclusion: We developed ANN models predicting the 8 month survival of unresectable pancreatic cancer patients. These models may help to optimize personalized patient management.

Multiple U-Net-Based Automatic Segmentations and Radiomics Feature Stability on Ultrasound Images for Patients With Ovarian Cancer.

Few studies have reported the reproducibility and stability of ultrasound (US) images based radiomics features obtained from automatic segmentation in oncology. The purpose of this study is to study the accuracy of automatic segmentation algorithms based on multiple U-net models and their effects on radiomics features from US images for patients with ovarian cancer. A total of 469 US images from 127 patients were collected and randomly divided into three groups: training sets (353 images), validation sets (23 images), and test sets (93 images) for automatic segmentation models building. Manual segmentation of target volumes was delineated as ground truth. Automatic segmentations were conducted with U-net, U-net++, U-net with Resnet as the backbone (U-net with Resnet), and CE-Net. A python 3.7.0 and package Pyradiomics 2.2.0 were used to extract radiomic features from the segmented target volumes. The accuracy of automatic segmentations was evaluated by Jaccard similarity coefficient (JSC), dice similarity coefficient (DSC), and average surface distance (ASD). The reliability of radiomics features were evaluated by Pearson correlation and intraclass correlation coefficients (ICC). CE-Net and U-net with Resnet outperformed U-net and U-net++ in accuracy performance by achieving a DSC, JSC, and ASD of 0.87, 0.79, 8.54, and 0.86, 0.78, 10.00, respectively. A total of 97 features were extracted from the delineated target volumes. The average Pearson correlation was 0.86 (95% CI, 0.83-0.89), 0.87 (95% CI, 0.84-0.90), 0.88 (95% CI, 0.86-0.91), and 0.90 (95% CI, 0.88-0.92) for U-net++, U-net, U-net with Resnet, and CE-Net, respectively. The average ICC was 0.84 (95% CI, 0.81-0.87), 0.85 (95% CI, 0.82-0.88), 0.88 (95% CI, 0.85-0.90), and 0.89 (95% CI, 0.86-0.91) for U-net++, U-net, U-net with Resnet, and CE-Net, respectively. CE-Net based segmentation achieved the best radiomics reliability. In conclusion, U-net based automatic segmentation was accurate enough to delineate the target volumes on US images for patients with ovarian cancer. Radiomics features extracted from automatic segmented targets showed good reproducibility and for reliability further radiomics investigations.

Efficacy And Safety Of Apatinib Treatment In Platinum-Resistant Recurrent Epithelial Ovarian Cancer: A Real World Study.

OBJECTIVE: To evaluate real-world use and outcomes of apatinib treatment in platinum-resistant recurrent epithelial ovarian cancer. METHODS: This is an observational study. Patients with platinum-resistant recurrent epithelial ovarian cancer initiating apatinib treatment from January 2016 to December 2018 were included. The primary end point was progression-free survival. Other end points included overall survival, objective response rate, disease control rate, and toxicity. RESULTS: A total of 28 platinum-resistant epithelial ovarian cancer patients were enrolled in this study. Thirteen cases received apatinib as maintenance therapy following chemotherapy with a median progression-free survival of 6.0 months and a medium overall survival of 11.0 months. Four patients received apatinib as palliative following chemotherapy with 2 cases in progressive disease and 2 cases in stable disease. Eleven cases received apatinib alone as salvage therapy with a disease control rate of 81.8% and a median progression-free survival of 3.0 months. The most common adverse effects were hand-foot syndrome (53.57%), secondary hypertension (46.43%) and fatigue (14.29%). Five patients discontinued treatment due to grade 3 toxicities and 4 patients required dose reduction because of adverse effects. CONCLUSION: Apatinib produced moderate improvements in progression-free survival in patients with platinum-resistant epithelial ovarian cancer both as maintenance therapy following chemotherapy and as single-agent salvage therapy. Our study suggests that apatinib may be effective for women with platinum-resistant recurrent epithelial ovarian cancer.

The Efficacy And Safety Of Apatinib Treatment For Patients With Metastatic Or Recurrent Cervical Cancer: A Retrospective Study.

BACKGROUND: This study was performed to assess the efficacy and safety of apatinib in patients with metastatic or recurrent cervical cancer. METHODS: Twenty-six patients with metastatic or recurrent cervical cancer and treated with apatinib until progressive disease or unacceptable toxicity were included in this multicenter, retrospective, observational study from January 2016 to April 2018. The primary end point was progression free survival (PFS). Secondary end points included overall survival (OS), objective response rate (ORR), disease control rate (DCR), and toxicity. Toxicities were assessed according to Common Terminology Criteria for Adverse Events. RESULTS: A total of 26 metastatic or recurrent cervical cancer patients were enrolled in this study. No complete response (CR) occurred, 4 patients (15.4%) showed partial response (PR), 11 patients (42.3%) had stable disease (SD), and 11 patients (42.3%) had progressive disease (PD), with DCR of 57.7% and ORR of 15.4%. Median progression-free survival (PFS) was 3.0 months (95% confidence interval [CI]: 0-6.3 months) and overall survival (OS) was 7.0 months (95% CI: 5.1-8.9 months) respectively. The most common adverse effects were hand-foot syndrome (50.0%), secondary hypertension (26.9%) and fatigue (26.9%). Three patients discontinued treatment due to grade 3 toxicities (one case for hand-foot syndrome, two cases for diarrhea) and 6 patients required dose reduction because of adverse effects. CONCLUSION: Apatinib seems active in heavily-pretreated metastatic or recurrent cervical cancer. The adverse effects were moderate but manageable.

Reducing the actuation threshold by incorporating a nonliquid crystal chain into a liquid crystal elastomer.

Liquid crystal elastomers (LCEs) are important smart materials that can undergo reversible deformation in response to liquid crystal (LC) phase transitions. A low threshold temperature for LC phase transition is advantageous because the LCE material can be more conveniently actuated by the applied stimulus. In this work, we investigated the effect of a nonliquid crystal chain on the reduction of threshold temperature of the LC phase transition by linking a nonliquid crystal side chain, 4-methoxyphenyl-1-hexenyloxy (MOCH3), to the network backbone of a classical polysiloxane-based side-chain nematic LCE. The nematic-isotropic transition temperature (T ni) of the MOCH3 incorporated nematic LCE was lower than that of the normal nematic LCE without the incorporation of a nonliquid crystal chain by about 27 °C. Compared to the normal nematic LCE or its nanocomposite, the MOCH3 incorporated nematic LCE or its nanocomposite demonstrated more rapid thermo-actuated deformation or photo-actuated deformation, and can be actuated to attain full axial contraction at an obviously lowered temperature or by light with obviously lowered intensity, while the maximum contraction ratio basically did not vary. These research results indicate that some nonliquid crystal chains show potential for improving the characteristics and enhancing the application significance of LCE materials.

Altered expression of nNOS/NIDD in the retina of a glaucoma model of DBA/2J mice and the intervention by nNOS inhibition.

The NIDD gene, neuronal NOS (nNOS)-interacting DHHC domain-containing protein with dendritic mRNA, codes a protein that upregulates nNOS enzyme activity by the interaction with the postsynaptic density protein 95/discsslarge/zon occlusens-1 (PDZ) domain of nNOS. Glial cell activation, especially Müller cells, may be an important factor contributing to retinal ganglion cell (RGC) death in glaucoma. The present study was to measure nNOS and NIDD expression in DBA/2J mice, a mouse model of glaucoma, and their correlation with glaucomatous phenotypes. Slit-lamp biomicroscopy, fundus photography, intraocular pressure (IOP) measurement, histology, and optic nerve axon counts were used to examine the ocular phenotypes of DBA/2J mice. Quantitative real-time PCR(RT-PCR) and Western blot analysis were used to analyze mRNA and protein expression of nNOS and NIDD. Their spatial distribution was evaluated by immunohistochemistry. Immunofluorescence was performed to observe the colocalization of nNOS and NIDD and the association of NIDD with Müller cells. The results showed that the prevalence and severity of ocular abnormalities, IOP, optic nerve cupping, and optic nerve atrophy increased with age. The mRNA and protein expression of nNOS reached the peak at 9 months old. The protein of NIDD underwent a similar change, while the mRNA of NIDD significantly increased at 6 months old. The expression of NIDD physically coexisted with nNOS in Müller cells. Administration of NOS inhibitor N(G)-Nitro-L-arginine-methyl-ester (L-NAME) by intraperitoneal injection (i.p.) prevented RGCs from apoptosis as shown in the increase of Brn-3a (RGC marker) expression, which was accompanied by decreased expression of NIDD. The spatiotemporal changes of nNOS/NIDD expression and its interference suggest that NIDD-nNOS axis may play a role in the degenerative process of RGC in glaucoma.