Interpretable Machine Learning for COVID-19: An Empirical Study on Severity Prediction Task

The black-box nature of machine learning models hinders the deployment of some high-accuracy medical diagnosis algorithms. It is risky to put one's life in the hands of models that medical researchers do not fully understand or trust. However, through model interpretation, black-box models can promptly reveal significant biomarkers that medical practitioners may have overlooked due to the surge of infected patients in the COVID-19 pandemic. This research leverages a database of 92 patients with confirmed SARS-CoV-2 laboratory tests between 18th January 2020 and 5th March 2020, in Zhuhai, China, to identify biomarkers indicative of infection severity prediction. Through the interpretation of four machine learning models, decision tree, random forests, gradient boosted trees, and neural networks using permutation feature importance, partial dependence plot, individual conditional expectation, accumulated local effects, local interpretable model-agnostic explanations, and Shapley additive explanation, we identify an increase in N-terminal pro-brain natriuretic peptide, C-reaction protein, and lactic dehydrogenase, a decrease in lymphocyte is associated with severe infection and an increased risk of death, which is consistent with recent medical research on COVID-19 and other research using dedicated models. We further validate our methods on a large open dataset with 5644 confirmed patients from the Hospital Israelita Albert Einstein, at São Paulo, Brazil from Kaggle, and unveil leukocytes, eosinophils, and platelets as three indicative biomarkers for COVID-19.

[Development and Application of Medical Imaging Analysis Platform Based on Radiomics and Machine Learning Technologies]

OBJECTIVE

In order to solve the technical problems, clinical researchers face the process of medical imaging analysis such as data labeling, feature extraction and algorithm selection, a medical imaging oriented multi-disease research platform based on radiomics and machine learning technology was designed and constructed.

METHODS

Five aspects including data acquisition, data management, data analysis, modeling and data management were considered. This platform provides comprehensive functions such as data retrieve and data annotation, image feature extraction and dimension reduction, machine learning model running, results validation, visual analysis and automatic generation of analysis reports, thus an integrated solution for the whole process of radiomics analysis has been generated.

RESULTS

Clinical researchers can use this platform for the whole process of radiomics and machine learning analysis for medical images, and quickly produce research results.

CONCLUSIONS

This platform greatly shortens the time for medical image analysis research, decreasing the work difficulty of clinical researchers, as well as significantly promoting their working efficiency.

[Association between the awareness of Nutrition Facts Panel and prepackaged food purchase behavior among residents]

Objective: To explore the association between the cognition of Nutrition Facts Panel and prepackaged food purchase behavior among residents in six provinces in China. Methods: Using a multi-stage sampling method, 3 002 adults aged 18-70 were selected from the western region (Sichuan), eastern region (Guangdong, Jiangsu, Beijing), central region (Henan), and northeastern region (Heilongjiang) of China from July 2020 to March 2021. Socio-demographic characteristics of participants and their cognition of Nutrition Facts Panel and prepackaged food purchase behavior were collected through questionnaire. A multivariate binary logistic regression model was used to analyze the association between cognition of Nutrition Facts Panel and prepackaged food purchase behavior. Results: The age of 3 002 subjects was (42.3±13.4) years, among which 63.8% (1 914) were female, 66.7% knew the Nutrition Facts Panel, 49.8% would read it when purchasing, 30.7% could understand it, and 56.6% (1 699) bought prepackaged food more than once a week. The results of multivariate analysis showed that after adjusting for relevant confounding factors, compared with the participants knowing but not reading the Nutrition Facts Panel, the group knowing and reading was more likely to buy 11 types of prepackaged food at least once a week (all P<0.05). Compared with the participants reading but not understanding the Nutrition Facts Panel, the group reading and understanding was less likely to buy 11 types of prepackaged food at least once a week (all P<0.05). Conclusion: There was a correlation between cognition of Nutrition Facts Panel and prepackaged food purchase behavior among residents.

Fine-Grained Assessment of COVID-19 Severity Based on Clinico-Radiological Data Using Machine Learning

BACKGROUND

The severe and critical cases of COVID-19 had high mortality rates. Clinical features, laboratory data, and radiological features provided important references for the assessment of COVID-19 severity. The machine learning analysis of clinico-radiological features, especially the quantitative computed tomography (CT) image analysis results, may achieve early, accurate, and fine-grained assessment of COVID-19 severity, which is an urgent clinical need.

OBJECTIVE

To evaluate if machine learning algorithms using CT-based clinico-radiological features could achieve the accurate fine-grained assessment of COVID-19 severity.

METHODS

The clinico-radiological features were collected from 78 COVID-19 patients with different severities. A neural network was developed to automatically measure the lesion volume from CT images. The severity was clinically diagnosed using two-type (severe and non-severe) and fine-grained four-type (mild, regular, severe, critical) classifications, respectively. To investigate the key features of COVID-19 severity, statistical analyses were performed between patients' clinico-radiological features and severity. Four machine learning algorithms (decision tree, random forest, SVM, and XGBoost) were trained and applied in the assessment of COVID-19 severity using clinico-radiological features.

RESULTS

The CT imaging features (CTscore and lesion volume) were significantly related with COVID-19 severity (p < 0.05 in statistical analysis for both in two-type and fine-grained four-type classifications). The CT imaging features significantly improved the accuracy of machine learning algorithms in assessing COVID-19 severity in the fine-grained four-type classification. With CT analysis results added, the four-type classification achieved comparable performance to the two-type one.

CONCLUSIONS

CT-based clinico-radiological features can provide an important reference for the accurate fine-grained assessment of illness severity using machine learning to achieve the early triage of COVID-19 patients.

Sodium MRI at 7T for Early Response Evaluation of Intracranial Tumors following Stereotactic Radiotherapy Using the CyberKnife

BACKGROUND AND PURPOSE

Conventionally, early treatment response to stereotactic radiotherapy in intracranial tumors is often determined by structural MR imaging. Tissue sodium concentration is altered by cellular integrity and energy status in cells. In this study, we aimed to investigate the feasibility of sodium MR imaging at 7T for the preliminary evaluation of radiotherapeutic efficacy for intracranial tumors.

MATERIALS AND METHODS

Data were collected from 16 patients (12 men and 4 women, 24-75 years of age) with 22 intracranial tumors who were treated with stereotactic radiation therapy using CyberKnife at our institution between December 1, 2016, and August 15, 2019. Sodium MR imaging was performed at 7T before and 48 hours, 1 week, and 1 month after CyberKnife radiation therapy. Tissue sodium concentration (TSC) was calculated and analyzed based on manually labeled regions of tumors.

RESULTS

Ultra-high-field sodium MR imaging clearly showed the intratumoral signal, which is significantly higher than that of normal tissue (t = 5.250, P <.001)., but the edema zone has some influence. The average TSC ratios of tumor to CSF in the 22 tumors, contralateral normal tissues, edema zones, frontal cortex, and frontal white matter were 0.66 (range, 0.23-1.5), 0.30 (range, 0.15-0.43), 0.58 (range, 0.25-1.21), 0.25 (range, 0.17-0.42), and 0.30 (range, 0.19-0.49), respectively. A total of 12 tumors in 8 patients were scanned at 48 hours, 1 week, and 1 month after treatment. The average TSC at 48 hours after treatment was 0.06 higher than that before treatment and began to decrease at 1 week. The TSC ratios of 10 continued to decline and 2 tumors increased at 1 month, respectively. Tumor volume decreased by 2.4%-99% after 3 months.

CONCLUSIONS

Changes in the TSC can be quantified by sodium MR imaging at 7T and used to detect radiobiologic alterations in intracranial tumors at early time points after CyberKnife radiation therapy.

Differentiating nontuberculous mycobacterium pulmonary disease from pulmonary tuberculosis through the analysis of the cavity features in CT images using radiomics

OBJECTIVE

To differentiate nontuberculous mycobacteria (NTM) pulmonary diseases from pulmonary tuberculosis (PTB) by analyzing the CT radiomics features of their cavity.

METHODS

73 patients of NTM pulmonary diseases and 69 patients of PTB with the cavity in Shandong Province Chest Hospital and Qilu Hospital of Shandong University were retrospectively analyzed. 20 patients of NTM pulmonary diseases and 20 patients of PTB with the cavity in Jinan Infectious Disease Hospitall were collected for external validation of the model. 379 cavities as the region of interesting (ROI) from chest CT images were performed by 2 experienced radiologists. 80% of cavities were allocated to the training set and 20% to the validation set using a random number generated by a computer. 1409 radiomics features extracted from the Huiying Radcloud platform were used to analyze the two kinds of diseases' CT cavity characteristics. Feature selection was performed using analysis of variance (ANOVA) and least absolute shrinkage and selection operator (LASSO) methods, and six supervised learning classifiers (KNN, SVM, XGBoost, RF, LR, and DT models) were used to analyze the features.

RESULTS

29 optimal features were selected by the variance threshold method, K best method, and Lasso algorithm.and the ROC curve values are obtained. In the training set, the AUC values of the six models were all greater than 0.97, 95% CI were 0.95-1.00, the sensitivity was greater than 0.92, and the specificity was greater than 0.92. In the validation set, the AUC values of the six models were all greater than 0.84, 95% CI were 0.76-1.00, the sensitivity was greater than 0.79, and the specificity was greater than 0.79. In the external validation set, The AUC values of the six models were all greater than 0.84, LR classifier has the highest precision, recall and F1-score, which were 0.92, 0.94, 0.93.

CONCLUSION

The radiomics features extracted from cavity on CT images can provide effective proof in distinguishing the NTM pulmonary disease from PTB, and the radiomics analysis shows a more accurate diagnosis than the radiologists. Among the six classifiers, LR classifier has the best performance in identifying two diseases.

Generative adversarial networks in medical image segmentation: A review

PURPOSE

Since Generative Adversarial Network (GAN) was introduced into the field of deep learning in 2014, it has received extensive attention from academia and industry, and a lot of high-quality papers have been published. GAN effectively improves the accuracy of medical image segmentation because of its good generating ability and capability to capture data distribution. This paper introduces the origin, working principle, and extended variant of GAN, and it reviews the latest development of GAN-based medical image segmentation methods.

METHOD

To find the papers, we searched on Google Scholar and PubMed with the keywords like "segmentation", "medical image", and "GAN (or generative adversarial network)". Also, additional searches were performed on Semantic Scholar, Springer, arXiv, and the top conferences in computer science with the above keywords related to GAN.

RESULTS

We reviewed more than 120 GAN-based architectures for medical image segmentation that were published before September 2021. We categorized and summarized these papers according to the segmentation regions, imaging modality, and classification methods. Besides, we discussed the advantages, challenges, and future research directions of GAN in medical image segmentation.

CONCLUSIONS

We discussed in detail the recent papers on medical image segmentation using GAN. The application of GAN and its extended variants has effectively improved the accuracy of medical image segmentation. Obtaining the recognition of clinicians and patients and overcoming the instability, low repeatability, and uninterpretability of GAN will be an important research direction in the future.

A Phase II Study Evaluating the Safety and Efficacy of Sunitinib Malate in Combination With Weekly Paclitaxel Followed by Doxorubicin and Daily Oral Cyclophosphamide Plus G-CSF as Neoadjuvant Chemotherapy for Locally Advanced or Inflammatory Breast Cancer

INTRODUCTION

Neoadjuvant chemotherapy is standard treatment for locally advanced breast cancer (LABC) or inflammatory breast cancer (IBC). We hypothesized that adding sunitinib, a tyrosine kinase inhibitor with antitumor and antiangiogenic activity, to an anthracycline and taxane regimen would improve pathologic complete response (pCR) rates to a prespecified endpoint of 45% in patients with HER2-negative LABC or IBC.

METHODS

We conducted a multicenter, phase II trial of neoadjuvant sunitinib with paclitaxel (S+T) followed by doxorubicin and cyclophosphamide plus G-CSF for patients with HER2-negative LABC or IBC. Patients received sunitinib 25 mg PO daily with paclitaxel 80 mg/m² IV weekly ×12 followed by doxorubicin 24 mg/m² IV weekly + cyclophosphamide 60 mg/m² PO daily with G-CSF support. Response was evaluated using pCR in the breast and the CPS + EG score (clinical-pathologic scoring + estrogen receptor [ER] and grade).

RESULTS

Seventy patients enrolled, and 66 were evaluable for efficacy. Eighteen patients (27%) had pCR in the breast (10 had ER⁺ disease and 8 had triple-negative disease). When defining response as pCR and/or CPS + EG score ≤2, 31 (47%) were responders. In pateints with ER positive disease, 23 (64%) were responders. The most common toxicities were cytopenias and fatigue.

CONCLUSIONS

Neoadjuvant S+T followed by AC+G-CSF was safe and tolerable in LABC and IBC. The study did not meet the prespecified endpoint for pCR; however, 47% were responders using pCR and/or CPS + EG score ≤2. ER positive patients had the highest response rate (64%). The addition of sunitinib to neoadjuvant chemotherapy may provide promising incremental benefit for patients with ER positive LABC.

Investigation of MRI-based radiomics model in differentiation between sinonasal primary lymphomas and squamous cell carcinomas

PURPOSE

To develop and validate an MRI-based radiomics model in differentiation between sinonasal primary lymphomas and squamous cell carcinomas (SCCs).

MATERIALS AND METHODS

One-hundred-and-fifty-four patients were enrolled (74 individuals with SCCs and 80 with lymphomas). After feature analysis and feature selection with variance threshold and least absolute shrinkage and selection operator (LASSO) methods, an MRI-based radiomics model with the support vector machine (SVM) classifier was constructed in differentiation between lymphomas and SCCs. Areas under the receiver operating characteristic curves (AUCs) of the MRI-based radiomics model were compared with those of radiologists using Delong test.

RESULTS

Five features (T1 original shape Compactness2, T1 wavelet-HHH first-order Total Energy, T2 wavelet-HLH GLCM Informational Measure of Correlation1, T1 wavelet-LHL GLCM Inverse Variance and T1 square GLRLM Long Run Low Gray Level Emphasis) were finally selected in the radiomics model. The AUC values in differentiation between lymphomas and SCCs were 0.94 for the training dataset and 0.85 for the validation dataset, respectively. For all the patient datasets, the AUC values of radiomics model, readers 1, 2 and 3 were 0.92, 0.76, 0.77 and 0.80, respectively. For the validation datasets, no significant difference was found between the AUCs of the radiomics model and those of the three radiologist (P = 0.459, 0.469, 0.738 for radiologist 1, 2 and 3, respectively).

CONCLUSION

An MRI-based radiomics model can help to differentiate sinonasal lymphomas from SCCs with high accuracy.

Iron deficiency plays essential roles in the trigger, treatment, and prognosis of autosomal dominant hypophosphatemic rickets

By analyzing iron status of 14 ADHR patients, we found that iron deficiency was an important trigger of ADHR. Correcting iron deficiency significantly improved patients' symptoms. Meanwhile, patients' serum phosphate showed positive correlations with iron metabolism parameters and hemoglobin-related parameters, suggesting the necessity of monitoring and correcting the iron status in ADHR.

INTRODUCTION

Autosomal dominant hypophosphatemic rickets (ADHR) is unique for its incomplete penetrance, variety of disease onsets, and waxing and waning phenotypes. Iron deficiency is a trigger of ADHR. This study aimed to clarify the role of iron deficiency in ADHR.

METHODS

Data of clinical manifestations and laboratory examinations were collected from patients among eight kindreds with ADHR. Multiple regression and Pearson's correlation tests were performed to test the relationships of serum phosphate levels and other laboratory variables during the patients' follow-ups.

RESULTS

Among 23 ADHR patients with fibroblast growth factor 23 (FGF23) mutations, 14 patients presented with obvious symptoms. Ten patients had iron deficiency at the onset of ADHR, coinciding with menarche, menorrhagia, pregnancy, and chronic gastrointestinal bleeding. Two patients who did not have their iron status tested presented with symptoms after abortion and pregnancy in one patient each, which suggested that they also had iron deficiency at onset. Patients were treated with ferrous succinate tablets, vitamin C, and neutral phosphate and calcitriol. With correction of the iron status, the patients' symptoms showed notable improvement, with increased serum phosphate levels. Two patients' FGF23 levels also declined to the normal range. There were strong correlations between serum phosphate and serum iron levels (r = 0.7689, p < 0.0001), serum ferritin levels (r = 0.5312, p = 0.002), iron saturation (r = 0.7907, p < 0.0001), and transferrin saturation (r = 0.7875, p < 0.001). We also examined the relationships between serum phosphate levels and hemoglobin-related indices, which were significant (hemoglobin: r = 0.71, p < 0.0001; MCV: r = 0.7589, p < 0.0001; MCH: r = 0.8218, p < 0.0001; and MCHC: r = 0.7751, p < 0.0001). Longitudinal data of six patients' follow-up also showed synchronous changes in serum phosphate with serum iron levels.

CONCLUSIONS

Iron deficiency plays an important role in triggering ADHR. Monitoring and correcting the iron status are helpful for diagnosing and treating ADHR. Iron metabolism parameters and hemoglobin-related parameters are positively correlated with serum phosphate levels in patients with ADHR and iron deficiency, and these might serve as good indicators of prognosis of ADHR.

Involvement of soluble proteins in growth and metabolic adjustments of drought-stressed Calligonum mongolicum seedlings under nitrogen addition

The planting of seedlings is the most effective measure for vegetation restoration. However, this practice is challenging in desert ecosystems where water and nutrients are scarce. Calligonum mongolicum is a sand-fixing pioneer shrub species, and its adaptive strategy for nitrogen (N) deposition and drought is poorly understood. Thus, in a pot experiment, we studied the impacts of four N levels (0, 3, 6, 9 gN·m^-2 ·year^-1 ) under drought or a well-watered regime on multiple eco-physiological responses of 1-year-old C. mongolicum seedlings. Compared to well-watered conditions, drought considerably influenced seedling growth by impairing photosynthesis, osmolyte accumulation and activity of superoxide dismutase and enzymes related to N metabolism. Nitrogen addition improved the productivity of drought-stressed seedlings, as revealed by increased water use efficiency, enhanced superoxide dismutase and nitrite reductase activity and elevated N and phosphorus (P) levels in seedlings. Nevertheless, the addition of moderate to high levels of N (6-9 gN·m^-2 ·year^-1 ) impaired net photosynthesis, osmolyte accumulation and nitrate reductase activity. N addition and water regimes did not markedly change the N:P ratios of aboveground parts; while more biomass and nutrients were allocated to fine roots to assimilate the insufficient resources. Soluble protein in assimilating shoots might play a vital role in adaptation to the desert environment. The response of C. mongolicum seedlings to N addtion and drought involved an interdependency between soluble protein and morphological, physiological and biochemical processes. These findings provide an important reference for vegetation restoration in arid lands under global change.

Spatial variation of the soil bacterial community in major apple producing regions of China

AIMS

In China, apple production areas are largely from the coastal to inland areas and across varied climate zones. However, the relationship among soil micro-organisms, environmental factors and fruit quality has not been clearly confirmed in orchards. Here we attempted to identify the variation of soil bacteria in the main apple producing regions and reveal the relationship among climatic factor, soil properties, soil bacterial community and fruit quality.

METHODS AND RESULTS

Sixty soil samples were collected from six main apple producing areas in China. We examined the soil bacteria using bacterial 16S rRNA gene amplicon profiling. The results show that the soil bacterial diversity of apple orchards varied from the Bohai Bay Region to the Loess Plateau Region. Proteobacteria, Acidobacteria and Actinobacteria were the predominant taxa at the phylum level for all six areas. In the Bohai Bay and the Loess Plateau region, which are the two largest apple producing areas, Proteobacteria and Actinobacteria had the highest relative abundance, respectively. Furthermore, soil bacterial diversity showed positive correlation with the mean annual temperature (MAT), soil organic matter (SOM) and pH. Excluding a direct effect on the apple fruit quality, MAT exerted an indirect influence through soil SOM and pH to alter the relative abundance of dominant taxa and shift the bacterial diversity, which affects the apple fruit titratable acids and soluble solids.

CONCLUSIONS

Geographic variables underlie apple orchard soil bacterial communities vary according to spatial scale. Environmental factors exert an indirect effect on apple fruit quality via shaping soil bacterial community.

SIGNIFICANCE AND IMPACT OF THE STUDY

This study provides a list of bacteria associated with environmental factors and the ecological attributes of their interactions in apple orchards, which will improve our ability to promote soil bacterial functional capabilities in order to reduce the fertilizer input and enhance the fruit quality.

Magnetic Solitons in a Spin-1 Bose-Einstein Condensate

Vector solitons are a type of solitary or nonspreading wave packet occurring in a nonlinear medium composed of multiple components. As such, a variety of synthetic systems can be constructed to explore their properties, from nonlinear optics to ultracold atoms, and even in metamaterials. Bose-Einstein condensates have a rich panoply of internal hyperfine levels, or spin components, which make them a unique platform for exploring these solitary waves. However, existing experimental work has focused largely on binary systems confined to the Manakov limit of the nonlinear equations governing the soliton behavior, where quantum magnetism plays no role. Here we observe, using a "magnetic shadowing" technique, a new type of soliton in a spinor Bose-Einstein condensate, one that exists only when the underlying interactions are antiferromagnetic and which is deeply embedded within a full spin-1 quantum system. Our approach opens up a vista for future studies of "solitonic matter" whereby multiple solitons interact with one another at deterministic locations.

Hypereosinophilia (HE) in acute myeloid leukemia (AML) with normal karyotype: A report of two cases

We present two rare cases of hypereosinophilia (HE) in acute myeloid leukemia with normal karyotype (NK-AML) at diagnosis. The first case is a 29-year-old female who presented with HE. On evaluation, she was found to have NK-AML. She failed to achieve complete remission (CR) after the first induction therapy with standard idarubicin and cytarabine (IA). She achieved CR after two cycles of reinduction chemotherapy with cytarabine, aclarubicin, and granulocyte colony-stimulating factor (G-CSF) (CAG) but had early relapsed. Reinduction chemotherapy with fludarabine, Ara-C, and G-CSF (FLAG) led to her second remission, followed by unrelated umbilical cord hematopoietic stem cell transplantation (HSCT). Unfortunately, she died of thrombotic thrombocytopenic purpura. The second case is a 23-year-old male who was diagnosed as NK-AML with HE. IA regimen was successively used in two cycles treatment achieving CR. He underwent haploidentical HSCT but had a relapse after 17 months of sustained remission and died 4 months later. The presence of HE may be a poor prognostic feature in NK-AML.

Apple rootstocks with different phosphorus efficiency exhibit alterations in rhizosphere bacterial structure

AIMS

The purpose of this study was to select phosphorus-efficient apple rootstocks under phosphorus deficiency and to reveal the effects of different apple rootstocks on the rhizosphere bacterial community.

METHODS AND RESULTS

We used 83 hybrid lines of Malus robusta Rehd. × Malling 9 (M.9) to investigate their physiological traits and the phosphorus deficiency phenotypes of leaves in response to phosphorus deficiency (0·1 mmol l^-1 PO₄ ^3- ). All the plants were cultivated in pots in the greenhouse and watered using drip irrigation. In accordance with the results of investigation, we selected the phosphorus-efficient hybrid lines (PE) and the phosphorus-inefficient hybrid lines (PI) to research their root morphology and root hairs (RH). In addition, we used Illumina MiSeq sequencing to determine the bacterial community of the rhizosphere from different rootstocks. The results showed that the PE plants had better growth characteristics and stronger root plasticity than that of the PI plants, and phosphorus deficiency can stimulate the RH growth of PE plants. There was no significant difference in the rhizosphere bacterial diversity, but we found that the bacterial community structure was significantly different at the genus levels; in addition, 89 genera were found to have significant differences between PE and PI plants, especially Bacillus. The PE rhizosphere had more abundant Bacillus compared to the PI. High positive Pearson correlations with the phosphorus concentration in the plantlets of apple rootstocks were detected for the bacterial genera Bacillus (r: 0·776).

CONCLUSIONS

The phosphorus-efficient apple rootstocks adapted to phosphorus deficiency by shaping the root morphology. Notably, different apple rootstocks showed alteration of the microbes in rhizosphere.

SIGNIFICANCE AND IMPACT OF THE STUDY

This study can provide the materials for exploring the mechanism of apple rootstock phosphorus absorption. In accordance with the different bacterial community compositions, we can develop the inoculants to promote nutrient uptake.

Fully automatic segmentation of type B aortic dissection from CTA images enabled by deep learning

PURPOSE

This study sought to establish a robust and fully automated Type B aortic dissection (TBAD) segmentation method by leveraging the emerging deep learning techniques.

METHODS

Preoperative CTA images of 276 patients with TBAD were retrospectively collected from January 2011 to December 2018. Using a reproducible manual segmentation protocol of three labels (whole aorta, true lumen (TL), and false lumen (FL)), a ground truth database (n = 276) was established and randomly divided into training and testing sets in a rough 8:1 ratio. Three convolutional neural network (CNN) models were developed on the training set (n = 246): single one-task (CNN1), single multi-task (CNN2), and serial multi-task (CNN3) models. Performance was evaluated using the Dice coefficient score (DCS) and lumen volume accuracy on the testing set (n = 30). Pearson correlation, Intra-class correlation coefficients and Bland-Altman plots were used to evaluate the inter-observer measurement agreement.

RESULTS

CNN3 performed the best, with mean DCSs of 0.93 ± 0.01, 0.93 ± 0.01 and 0.91 ± 0.02 for the whole aorta, TL, and FL, respectively (p < 0.05). Each label volume from CNN3 showed excellent agreement with the ground truth, with mean volume differences of -31.05 (-82.76 to 20.65) ml, 4.79 (-11.04 to 20.63) ml, and 8.67(-11.40 to 28.74) ml for the whole aorta, TL, and FL, respectively. The segmentation speed of CNN3 was 0.038 ± 0.006 s/image.

CONCLUSION

Deep learning-based model provides a promising approach for accurate and efficient segmentation of TBAD and makes it possible for automated measurements of TBAD anatomical features.

let-7i inhibits proliferation and migration of bladder cancer cells by targeting HMGA1

Background

Let-7 is one of the earliest discovered microRNAs(miRNAs) and has been reported to be down-regulated in multiple malignant tumors. The effects and molecular mechanisms of let-7i in bladder cancer are still unclear. This study was to investigate the effects and potential mechanisms of let-7i on bladder cancer cells.

Methods

Total RNA was extracted from bladder cancer cell lines. The expression levels of let-7i and HMGA1 were examined by quantitative real-time PCR. Cell viability was detected using the CCK-8 and colony formation assays, while transwell and wound healing assays were used to evaluate migration ability. Luciferase reporter assay and western blot were used to confirm the target gene of let-7i.

Results

Compared with the SV-40 immortalized human uroepithelial cell line (SV-HUC-1), bladder cancer cell lines T24 and 5637 had low levels of let-7i expression, but high levels of high mobility group protein A1 (HMGA1) expression. Transfection of cell lines T24 and 5637 with let-7i mimic suppressed cell proliferation and migration. Luciferase reporter assay confirmed HMGA1 may be one of the target genes of let-7i-5p. Protein and mRNA expression of HMGA1 was significantly downregulated in let-7i mimic transfected cell lines T24 and 5637.

Conclusions

Up-regulation of let-7i suppressed proliferation and migration of the human bladder cancer cell lines T24 and 5637 by targeting HMGA1. These findings suggest that let-7i might be considered as a novel therapeutic target for bladder cancer.

Apple rootstocks of different nitrogen tolerance affect the rhizosphere bacterial community composition

AIMS

To select apple rootstocks that are tolerant to low nitrogen and reveal the relationship between the rhizosphere bacterial communities and the low nitrogen tolerance of the apple rootstock.

METHODS AND RESULTS

In total, 235 lines of hybrids of Malus robusta Rehd. × M.9 with low nitrogen stress were cultivated in pots in a greenhouse equipped with a drip irrigation system, and growth characteristics, photosynthesis traits and mineral elements were monitored. The bacterial community structure of the rhizosphere from different rootstocks was determined via Illumina MiSeq sequencing. This study selected three low nitrogen-tolerant (NT) lines that had higher nitrogen concentration, and higher photosynthesis rate than the three low nitrogen-sensitive (NS) lines. The bacterial community structure significantly differed (P ≤ 0·001) among the rootstocks. The bacterial phyla Proteobacteria and Actinobacteria were the dominant groups in the rhizosphere and presented higher abundance in the NT rhizosphere. The N concentration in the apple rootstocks exhibited highly positive Pearson correlations with the bacterial genera Sphingomonas, Pseudoxanthomonas, Bacillus and Acinetobacter, and negative correlations with the bacterial genera Pseudarthrobacter and Bradyrhizobium.

CONCLUSIONS

This study showed that investigated rootstocks achieved increased nitrogen concentration by enhancing their photosynthetic production capacity and shaping their rhizobacteria community structure.

SIGNIFICANCE AND IMPACT OF THE STUDY

The findings provide a basis for studying the mechanisms of resistance to low nitrogen stress in apple rootstocks. Based on these beneficial bacteria, microbial inoculants can be developed for use in sustainable agricultural and horticultural production.

Observation of crossover from intraband to interband nonlinear terahertz optics

In this Letter, we investigate the nonlinear effects of extremely intense few-cycle terahertz (THz) pulses (generated from the organic crystal 4-NN, NN-dimethylamino-4^'4^'-N^'N^'-methyl-stilbazolium 2, 4, 6 trimethylbenzenesulfonate, with peak electrical fields of a few MV/cm) on the carrier dynamics in n-doped semiconductor thin film In_0.53Ga_0.47As. By performing open-aperture Z-scan measurements and recording the transmitted THz energy through semiconductor sample, we observed a strong THz absorption bleaching effect at high fields, followed by an absorption enhancement at even higher fields. We attribute our observations to a crossover from pure intraband carrier dynamics to an interplay between intraband carrier heating and interband carrier generations.

Giant self-induced transparency of intense few-cycle terahertz pulses in n-doped silicon

The results of high-field terahertz transmission experiments on n-doped silicon (carrier concentration of 8.7×10¹⁶  cm^-3) are presented. We use terahertz pulses with electric field strengths up to 3.1  MV cm^-1 and a pulse duration of 700 fs. A huge transmittance enhancement of ∼90 times is observed with increasing of the terahertz electric field strengths within the range of 1.5-3.1  MV cm^-1.

HLA-A*24:388N: a novel HLA-A*24 allele identified by sequence-based typing

The novel allele, A*24:388N, was identified by sequence-based typing in a Chinese individual.

Automatic multiorgan segmentation in CT images of the male pelvis using region-specific hierarchical appearance cluster models

PURPOSE

Accurate segmentation of pelvic organs in CT images is of great importance in external beam radiotherapy for prostate cancer. The aim of this studying is to develop a novel method for automatic, multiorgan segmentation of the male pelvis.

METHODS

The authors' segmentation method consists of several stages. First, a pretreatment includes parameterization, principal component analysis (PCA), and an established process of region-specific hierarchical appearance cluster (RSHAC) model which was executed on the training dataset. After the preprocessing, online automatic segmentation of new CT images is achieved by combining the RSHAC model with the PCA-based point distribution model. Fifty pelvic CT from eight prostate cancer patients were used as the training dataset. From another 20 prostate cancer patients, 210 CT images were used for independent validation of the segmentation method.

RESULTS

In the training dataset, 15 PCA modes were needed to represent 95% of shape variations of pelvic organs. When tested on the validation dataset, the authors' segmentation method had an average Dice similarity coefficient and mean absolute distance of 0.751 and 0.371 cm, 0.783 and 0.303 cm, 0.573 and 0.604 cm for prostate, bladder, and rectum, respectively. The automated segmentation process took on average 5 min on a personal computer equipped with Core 2 Duo CPU of 2.8 GHz and 8 GB RAM.

CONCLUSIONS

The authors have developed an efficient and reliable method for automatic segmentation of multiple organs in the male pelvis. This method should be useful for treatment planning and adaptive replanning for prostate cancer radiotherapy. With this method, the physicist can improve the work efficiency and stability.

Abstract P5-01-02: Multimodality molecular imaging with dynamic 18F-fluorodeoxyglucose positron emission tomography (FDG PET) and MRI to evaluate response and resistance to neoadjuvant chemotherapy (NAC)

Background: Using quantitative FDG PET to measure glucose metabolism and perfusion, and dynamic contrast-enhanced (DCE) MRI to measure perfusion, we previously identified a metabolic signature for breast cancer resistant to NAC. This imaging signature is (1) persistent or increased tumor perfusion despite treatment, (2) an altered pattern of glucose kinetics in response to therapy, and (3) pre-therapy mismatch between tumor metabolism (MRFDG) and glucose delivery (K1) (high ratio of MRFDG/K1). These patterns predict poor response, early relapse and death independent of established prognostic factors, including pathologic response. Identification of factors associated with resistance or response to therapy is the translational goal of "Quantitative Dynamic PET and MRI in Breast Cancer Therapy," part of the Seattle Breast SPORE (1P50CA138293).

Methods: Patients (Pts) undergoing NAC for histologically confirmed breast cancer (stage II-III) were approached for this trial (CCIRB# 7587). FDG PET and DCE-MRI were obtained pre-therapy, 2-12 weeks after start of NAC (mid-therapy) and after completion of NAC. Breast biopsies were obtained pre-therapy and post-NAC. FDG PET included a dynamic scan with kinetic analysis. PET measures included SUVmax, MRFDG, K1, Ki, and Patlak. 3T DCE-MRI measurements included semi-quantitative vascular parameters of peak enhancement (PE), signal enhancement ratio (SER), washout fraction, functional tumor volume, and apparent diffusion coefficient (ADC) from diffusion-weighted MRI (DWI). Breast biopsies were assayed by immunohistochemistry and gene expression profiling. NAC was per physician's choice with most pts receiving weekly paclitaxel (with trastuzumab if HER2+) followed by doxorubicin/cyclophosphamide.

Results: 32 pts have completed the study. Pathologic complete response (pCR), defined as absence of invasive cancer in the breast, was observed in 9 (28%); near pCR defined as only microscopic residual invasive cancer in 3 (9%) more pts. Mid-therapy decline in SUVmax and K1 was associated with near pCR; (p-value 0.06, 0.04, respectively). Pre-therapy PET measures of MRFDG and K1 were not predictive of pCR. On MRI, pre-therapy PE (p=0.009), SER (p=0.01), washout fraction (p=0.02), ADC (p=0.08, trend) and mid-therapy change in volume (p=0.05) were each predictive of pCR. Gene profiling of pre-therapy biopsies showed correlation between high MRFDG/K1 ratio in basal and luminal B tumors.

Conclusions: Assessment of serial changes in tumor metabolism and perfusion by FDG PET and DCE-MRI is feasible in the clinic. Mid-therapy decline in metabolism and glucose delivery was predictive of pCR; consistent with prior retrospective series. Baseline DCE-MRI and DWI measures show promise to predict response, and associations of mid-therapy change in MR functional tumor volume with pCR agree with findings of another multisite clinical trial (ISPY). These imaging parameters may serve as useful biomarkers to inform future neoadjuvant trials. Integration of imaging data with gene expression profiling revealed that the pattern of metabolism in luminal B tumors was closer to that of the basal subtype compared to other ER-positive tumors.

Citation Format: Specht JM, Partridge S, Chai X, Novakova A, Peterson L, Shields A, Guenthoer J, Linden HM, Gralow JR, Gadi V, Korde L, Hills D, Hsu L, Hockenbery DM, Kinahan P, Mankoff DA, Porter PL. Multimodality molecular imaging with dynamic 18F-fluorodeoxyglucose positron emission tomography (FDG PET) and MRI to evaluate response and resistance to neoadjuvant chemotherapy (NAC). [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P5-01-02.

Oral chronic GVHD outcomes and resource utilization: a subanalysis from the chronic GVHD consortium

OBJECTIVES

This study evaluated the extent to which oral chronic graft-versus-host disease (cGVHD) consensus assessments are predictive of management across institutions with and without oral medicine (OM) centers, and whether ancillary care guidelines are followed within clinical practice.

METHODS

Longitudinal oral cGVHD data were abstracted from the cGVHD Consortium, and additional mouth-specific management data were analyzed across five transplant centers.

RESULTS

Seventy-nine patients with 656 visits were observed for a median of 7.1 months with one visit per follow-up month. Ancillary therapies for oral cGVHD were prescribed for 67% of patients for a median of 0.46 months (per follow-up month) at OM centers and 0.78 months at non-OM centers. Patients treated with ancillary therapy were more likely to have an National Institutes of Health (NIH) mouth score of ≥1 (P < 0.001, odds ratio: 5.1) and mouth pain (P = 0.01, odds ratio: 2.6). The odds ratios of receiving ancillary therapy from OM experts were higher than transplant physicians (53%; P = 0.03).

CONCLUSIONS

Oral cGVHD consensus assessments corresponding with ancillary therapy use were mouth pain and NIH mouth score, with higher odds ratios of receiving therapy from OM experts. Ancillary care guidelines for oral cGVHD are reflected in academic clinical practice with respect to utilization of recommended prescriptions.

Abstract P4-01-03: HDACi (vorinostat) in metastatic breast cancer to restore sensitivity to ER-directed (AI) therapy: A phase II clinical trial with FES imaging correlates

Background: Histone deacetylase inhibitors (HDACi) have shown pre-clinical promise in estrogen receptor(ER)-modulation and restoring sensitivity to endocrine manipulation, suggesting potential clinical benefit (Sabnis 2011) (Huang 2000) in ER+ breast cancer. Vorinostat is an FDA-approved HDACi for CTCL, and could have a beneficial role in restoring ER-signaling in endocrine-resistant tumors (Munster 2011) (Yardley 2011). [F-18]fluoroestradiol (FES) PET imaging may be used to monitor regional tumor ER expression in patients with breast cancer (Linden 2011).

Methods: Patients with metastatic breast cancer with prior clinical benefit from endocrine manipulation who progressed on an AI therapy are eligible for this ongoing trial. In part A, patients were given vorinostat for 2 weeks, then resumed AI for 6 W. In part B (reflecting results of prior HDACi trials) patients are given vorinostat 400mg po daily 5/7 days 3/4 weeks while AI is given continuously. Paired FES and FDG PET are performed at baseline, week 2 and 8; clinical/radiologic assessment of disease is also performed at week 8. Patients with clinical benefit (response or stable disease) may continue on treatment until progressive disease or study withdrawal. Lesion-level analysis of the association between baseline FES uptake (logged) and FES/FDG ratio used generalized estimating equations (GEE) with small-sample adjustments to standard errors.

Results: 12/ 20 planned patients have accrued, and the treatment is well tolerated. Enrolled women were postmenopausal, the majority with primary infiltrating ductal tumors, bone/soft tissue dominant with longstanding metastatic disease, exposed to multiple endocrine and chemotherapy regimens. Five patients have had clinical benefit (2/4 on part B with greater HDACi exposure). One patient withdrew from the study due to toxicity. FES and FDG uptake was analyzed in 42 lesions in 11 patients. Average FES uptake was 2.0 (SULmean) for patients with clinical benefit, and 1.2 in patients with progressive disease by 8 weeks (p = 0.09). FES/FDG ratio at baseline was also associated with response (p = 0.04).

Conclusions: HDACi therapy is promising in relapsed ER+ breast cancer. Imaging of metabolic pathways in parallel with clinical trials may accelerate understanding of the underlying tumor biology and refine treatment selection.

Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P4-01-03.

Abstract P4-01-02: The role of FLT PET early assessment of response to endocrine therapy for early stage breast cancer

Background: In estrogen receptor positive (ER+) tumors, a low proliferative index (Ki-67) two weeks into endocrine therapy predicts response. FLT PET non-invasively measures tumor proliferation in vivo. The pre-operative window is an opportunity to assess impact of systemic therapies. We tested associations between FLT PET qualitative and quantitative measures and Ki-67 following two weeks of aromatase inhibitor (AI) therapy.

Methods: Women with clinical stage I-II ER+ HER2– breast cancer underwent “run-in” of AI monotherapy prior to definitive surgery. Premenopausal women were given GNRH agonist treatment 2 W prior to AI therapy. FLT PET was performed before AI therapy, and 1-7 days before surgery. Ki-67 was measured in baseline core biopsy and surgical specimens.

Results: Fourteen patients (8 postmenopausal, 6 premenopausal) have been enrolled. All have undergone baseline FLT PET imaging; 11 have completed imaging and surgery, including one premenopausal patient with no residual invasive carcinoma following 26 days of AI therapy. The majority harbored ductal carcinomas (n = 9, 5 with lobular histology) with the majority histologic grade ≥ 2 (n = 11). The median number of days exposed to AI was 19 (range, 9-42). Baseline SUVmax ranged from 1.2 to 3.9 (median 2.2), and post run-in SUV (6-64 days later) ranged from 1.2 to 2.8 (median 1.8). Baseline Ki-67 ranged from 6-26.2, median 11.6; surgical Ki-67 post AI therapy ranged from 0- 20.3 median 3.7, with seven below 5%. SUV and flux declined in most patients, as did Ki-67.

Quantitative FLT flux correlated with tumor response assessed by proliferative index (Ki-67) before the “run-in” period, with a stronger correlation at surgery, Pearson correlation coefficients = 0.41 and 0.82, respectively. FLT SUV and qualitative changes were not strongly associated with Ki-67.

Conclusions: Both pre and postmenopausal women with early stage breast cancer showed imaging and tissue response to endocrine therapy. Quantitative, but not qualitative FLT is a promising tool to assess tumor proliferation and response to therapy. Accrual is ongoing and updated results will be reported.

Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P4-01-02.

The clinical application of absorbable intramedullary nail and claw plate on treating multiple rib fractures

AIM

The absorption intramedullary nail and claw plate indications and efficacy were investigated in the treatment of a life-threatening multiple rib fractures.

METHODS

A retrospective analysis of 248 surgically treated rib fracture patients was performed who admitted to our hospital from March 2007 to December 2012. Intramedullary nailing was performed in 28 cases, a claw-type bone plate was fixed in 141 cases, and a combination of both was fixed in 79 cases.

RESULTS

All internal fixation patients were clinically cured except 1 patient died 14 days after a massive pulmonary embolism. The patients with flail chest and floating chest wall causing respiratory and circulatory disorders were promptly corrected. Routine follow-up was from 1 to 2 years, displaced fractures were in 2 cases, and there were 11 cases of internal fixation and extraction.

CONCLUSION

Internal fixation is a simple and reliable method for the treatment of multiple rib fractures. Both internal fixation materials have their pros and cons but the claw bone plate is more robust. The actual selection of appropriate treatment options helps to improve the treatment efficacy.

Measurement of oral chronic GVHD: results from the Chronic GVHD Consortium

Oral chronic GVHD (cGVHD) is a serious complication of alloSCT. Scales and instruments to measure oral cGVHD activity and severity have not been prospectively validated. The objective of this study was to describe the characteristics of oral cGVHD and determine the measures most sensitive to change. Patients enrolled in the cGVHD Consortium with oral involvement were included. Clinicians scored oral changes according to the National Institutes of Health (NIH) criteria, and patients completed symptom and quality-of-life measures at each visit. Both rated change on an eight-point scale. Of the 458 participants, 72% (n=331) had objective oral involvement at enrollment. Lichenoid change was the most common feature (n=293; 89%). At visits where oral change could be assessed, 50% of clinicians and 56% of patients reported improvement, with worsening reported in 4-5% for both the groups (weighted kappa=0.41). Multivariable regression modeling suggested that the measurement changes most predictive of perceived change by clinicians and patients were erythema and lichenoid, NIH severity and symptom scores. Oral cGVHD is common and associated with a range of signs and symptoms. Measurement of erythema and lichenoid changes and symptoms may adequately capture the activity of oral cGVHD in clinical trials but require prospective validation.