A robust and efficient AI assistant for breast tumor segmentation from DCE-MRI via a spatial-temporal framework.

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) allows screening, follow up, and diagnosis for breast tumor with high sensitivity. Accurate tumor segmentation from DCE-MRI can provide crucial information of tumor location and shape, which significantly influences the downstream clinical decisions. In this paper, we aim to develop an artificial intelligence (AI) assistant to automatically segment breast tumors by capturing dynamic changes in multi-phase DCE-MRI with a spatial-temporal framework. The main advantages of our AI assistant include (1) robustness, i.e., our model can handle MR data with different phase numbers and imaging intervals, as demonstrated on a large-scale dataset from seven medical centers, and (2) efficiency, i.e., our AI assistant significantly reduces the time required for manual annotation by a factor of 20, while maintaining accuracy comparable to that of physicians. More importantly, as the fundamental step to build an AI-assisted breast cancer diagnosis system, our AI assistant will promote the application of AI in more clinical diagnostic practices regarding breast cancer.

Development and validation of a radiomics-based nomogram for predicting a major pathological response to neoadjuvant immunochemotherapy for patients with potentially resectable non-small cell lung cancer.

Introduction: The treatment response to neoadjuvant immunochemotherapy varies among patients with potentially resectable non-small cell lung cancers (NSCLC) and may have severe immune-related adverse effects. We are currently unable to accurately predict therapeutic response. We aimed to develop a radiomics-based nomogram to predict a major pathological response (MPR) of potentially resectable NSCLC to neoadjuvant immunochemotherapy using pretreatment computed tomography (CT) images and clinical characteristics. Methods: A total of 89 eligible participants were included and randomly divided into training (N=64) and validation (N=25) sets. Radiomic features were extracted from tumor volumes of interest in pretreatment CT images. Following data dimension reduction, feature selection, and radiomic signature building, a radiomics-clinical combined nomogram was developed using logistic regression analysis. Results: The radiomics-clinical combined model achieved excellent discriminative performance, with AUCs of 0.84 (95% CI, 0.74-0.93) and 0.81(95% CI, 0.63-0.98) and accuracies of 80% and 80% in the training and validation sets, respectively. Decision curves analysis (DCA) indicated that the radiomics-clinical combined nomogram was clinically valuable. Discussion: The constructed nomogram was able to predict MPR to neoadjuvant immunochemotherapy with a high degree of accuracy and robustness, suggesting that it is a convenient tool for assisting with the individualized management of patients with potentially resectable NSCLC.

ViSTA: A Novel Network Improving Lung Adenocarcinoma Invasiveness Prediction from Follow-Up CT Series.

To investigate the value of the deep learning method in predicting the invasiveness of early lung adenocarcinoma based on irregularly sampled follow-up computed tomography (CT) scans. In total, 351 nodules were enrolled in the study. A new deep learning network based on temporal attention, named Visual Simple Temporal Attention (ViSTA), was proposed to process irregularly sampled follow-up CT scans. We conducted substantial experiments to investigate the supplemental value in predicting the invasiveness using serial CTs. A test set composed of 69 lung nodules was reviewed by three radiologists. The performance of the model and radiologists were compared and analyzed. We also performed a visual investigation to explore the inherent growth pattern of the early adenocarcinomas. Among counterpart models, ViSTA showed the best performance (AUC: 86.4% vs. 60.6%, 75.9%, 66.9%, 73.9%, 76.5%, 78.3%). ViSTA also outperformed the model based on Volume Doubling Time (AUC: 60.6%). ViSTA scored higher than two junior radiologists (accuracy of 81.2% vs. 75.4% and 71.0%) and came close to the senior radiologist (85.5%). Our proposed model using irregularly sampled follow-up CT scans achieved promising accuracy in evaluating the invasiveness of the early stage lung adenocarcinoma. Its performance is comparable with senior experts and better than junior experts and traditional deep learning models. With further validation, it can potentially be applied in clinical practice.

GMILT: A Novel Transformer Network That Can Noninvasively Predict EGFR Mutation Status.

Noninvasively and accurately predicting the epidermal growth factor receptor (EGFR) mutation status is a clinically vital problem. Moreover, further identifying the most suspicious area related to the EGFR mutation status can guide the biopsy to avoid false negatives. Deep learning methods based on computed tomography (CT) images may improve the noninvasive prediction of EGFR mutation status and potentially help clinicians guide biopsies by visual methods. Inspired by the potential inherent links between EGFR mutation status and invasiveness information, we hypothesized that the predictive performance of a deep learning network can be improved through extra utilization of the invasiveness information. Here, we created a novel explainable transformer network for EGFR classification named gated multiple instance learning transformer (GMILT) by integrating multi-instance learning and discriminative weakly supervised feature learning. Pathological invasiveness information was first introduced into the multitask model as embeddings. GMILT was trained and validated on a total of 512 patients with adenocarcinoma and tested on three datasets (the internal test dataset, the external test dataset, and The Cancer Imaging Archive (TCIA) public dataset). The performance (area under the curve (AUC) = 0.772 on the internal test dataset) of GMILT exceeded that of previously published methods and radiomics-based methods (i.e., random forest and support vector machine) and attained a preferable generalization ability (AUC = 0.856 in the TCIA test dataset and AUC = 0.756 in the external dataset). A diameter-based subgroup analysis further verified the efficiency of our model (most of the AUCs exceeded 0.772) to noninvasively predict EGFR mutation status from computed tomography (CT) images. In addition, because our method also identified the "core area" of the most suspicious area related to the EGFR mutation status, it has the potential ability to guide biopsies.

Survivors of COVID-19 exhibit altered amplitudes of low frequency fluctuation in the brain: a resting-state functional magnetic resonance imaging study at 1-year follow-up.

Although some short-term follow-up studies have found that individuals recovering from coronavirus disease 2019 (COVID-19) exhibit anxiety, depression, and altered brain microstructure, their long-term physical problems, neuropsychiatric sequelae, and changes in brain function remain unknown. This observational cohort study collected 1-year follow-up data from 22 patients who had been hospitalized with COVID-19 (8 males and 11 females, aged 54.2 ± 8.7 years). Fatigue and myalgia were persistent symptoms at the 1-year follow-up. The resting state functional magnetic resonance imaging revealed that compared with 29 healthy controls (7 males and 18 females, aged 50.5 ± 11.6 years), COVID-19 survivors had greatly increased amplitude of low-frequency fluctuation (ALFF) values in the left precentral gyrus, middle frontal gyrus, inferior frontal gyrus of operculum, inferior frontal gyrus of triangle, insula, hippocampus, parahippocampal gyrus, fusiform gyrus, postcentral gyrus, inferior parietal angular gyrus, supramarginal gyrus, angular gyrus, thalamus, middle temporal gyrus, inferior temporal gyrus, caudate, and putamen. ALFF values in the left caudate of the COVID-19 survivors were positively correlated with their Athens Insomnia Scale scores, and those in the left precentral gyrus were positively correlated with neutrophil count during hospitalization. The long-term follow-up results suggest that the ALFF in brain regions related to mood and sleep regulation were altered in COVID-19 survivors. This can help us understand the neurobiological mechanisms of COVID-19-related neuropsychiatric sequelae. This study was approved by the Ethics Committee of the Second Xiangya Hospital of Central South University (approval No. 2020S004) on March 19, 2020.

Persistent white matter changes in recovered COVID-19 patients at the 1-year follow-up.

There is growing evidence that severe acute respiratory syndrome coronavirus 2 can affect the CNS. However, data on white matter and cognitive sequelae at the 1-year follow-up are lacking. Therefore, we explored these characteristics in this study. We investigated 22 recovered coronavirus disease 2019 (COVID-19) patients and 21 matched healthy controls. Diffusion tensor imaging, diffusion kurtosis imaging and neurite orientation dispersion and density imaging were performed to identify white matter changes, and the subscales of the Wechsler Intelligence scale were used to assess cognitive function. Correlations between diffusion metrics, cognitive function and other clinical characteristics were then examined. We also conducted subgroup analysis based on patient admission to the intensive care unit. The corona radiata, corpus callosum and superior longitudinal fasciculus had a lower volume fraction of intracellular water in the recovered COVID-19 group than in the healthy control group. Patients who had been admitted to the intensive care unit had lower fractional anisotropy in the body of the corpus callosum than those who had not. Compared with the healthy controls, the recovered COVID-19 patients demonstrated no significant decline in cognitive function. White matter tended to present with fewer abnormalities for shorter hospital stays and longer follow-up times. Lower axonal density was detected in clinically recovered COVID-19 patients after 1 year. Patients who had been admitted to the intensive care unit had slightly more white matter abnormalities. No significant decline in cognitive function was found in recovered COVID-19 patients. The duration of hospital stay may be a predictor for white matter changes at the 1-year follow-up.

[Trend analysis on mortality and the mortal causes among children under 5 in Sichuan province, from 2001 to 2013].

OBJECTIVE: To study the mortality of children under five and the causes of death together with related trend of dynamics, from 2001 to 2013 in Sichuan province. METHODS: Using the Children Death Monitoring Network under five in Sichuan province to obtain basic data. Descriptive statistics and chi-square were used to describe the mortalities in children and infants as well as the causes of death, in both rural and urban areas of Sichuan province. RESULTS: In Sichuan province, the mortality of children under five decreased from 35.30‰ in 2001 to 11.77‰ in 2013. In 2013, mortality in the rural areas was 2.37 times more than that in the urban area. The proportion of neonatal deaths among the mortality in children under five was 44.72%. Pneumonia, congenital heart diseases and premature or low birth weigh were the top three causes of death for children under five. Among them, the top three causes of death for urban area were congenital heart disease, drowning, and premature or low birth weight/birth asphyxia. Meanwhile, the top three causes of death in rural areas were pneumonia, premature birth/low birth weight and birth asphyxia. Overall, the mortality rates of birth asphyxia, pneumonia and low birth weight gradually decreasd but drowning, diarrhea and traffic accidents fluctuated. CONCLUSION: The mortality of children under five in Sichuan province was 13‰, which had already met the goal set for the year 2020. However, reducing the mortality in rural areas, narrowing the gap between urban and rural areas seemed the main part of the future endeavor while focus of prevention should be adjusted according to the causes of death.

[Infant mortality and its leading causes in Sichuan province, from 2001 to 2009].

OBJECTIVE: To study the trend of infant mortality and the leading cause of the deaths in Sichuan province from 2001 to 2009. METHODS: Data presented in this report was obtained from the child mortality surveillance network with target population as children under 5 years of age. Rates on infant mortality, neonatal mortality and indirect estimation of infant mortality were calculated. RESULTS: The neonatal mortality rate and infant mortality rate in Sichuan dropped from 18.6, 25.5 in 2001 to 7.6, 12.1 per 1000 live birth in 2009, with rates of decline as 59.1% and 35.0%, from 2001 to 2009. In urban areas of Sichuan, the neonatal and infant mortality rates dropped from 4.7, 7.5 in 2001 to 3.7 and 6.5 per 1000 live birth in 2009, with the rates of decline as 22.3% and 13.1%. In the rural areas of Sichuan, the neonatal and infant mortality rates dropped from 25.2 and 34.0 in 2001 to 9.6, 14.3 per 1000 live birth in 2009, with rates of decline as 62.0%, 57.9% from 2001 to 2009. CONCLUSION: In both urban and rural areas, the neonatal and infant mortality rates had decreased drastically from 2001 to 2009, due to the decrease of avoidable deaths as pneumonia and diarrhea in infants.