Multi-phase features interaction transformer network for liver tumor segmentation and microvascular invasion assessment in contrast-enhanced CT.

Precise segmentation of liver tumors from computed tomography (CT) scans is a prerequisite step in various clinical applications. Multi-phase CT imaging enhances tumor characterization, thereby assisting radiologists in accurate identification. However, existing automatic liver tumor segmentation models did not fully exploit multi-phase information and lacked the capability to capture global information. In this study, we developed a pioneering multi-phase feature interaction Transformer network (MI-TransSeg) for accurate liver tumor segmentation and a subsequent microvascular invasion (MVI) assessment in contrast-enhanced CT images. In the proposed network, an efficient multi-phase features interaction module was introduced to enable bi-directional feature interaction among multiple phases, thus maximally exploiting the available multi-phase information. To enhance the model's capability to extract global information, a hierarchical transformer-based encoder and decoder architecture was designed. Importantly, we devised a multi-resolution scales feature aggregation strategy (MSFA) to optimize the parameters and performance of the proposed model. Subsequent to segmentation, the liver tumor masks generated by MI-TransSeg were applied to extract radiomic features for the clinical applications of the MVI assessment. With Institutional Review Board (IRB) approval, a clinical multi-phase contrast-enhanced CT abdominal dataset was collected that included 164 patients with liver tumors. The experimental results demonstrated that the proposed MI-TransSeg was superior to various state-of-the-art methods. Additionally, we found that the tumor mask predicted by our method showed promising potential in the assessment of microvascular invasion. In conclusion, MI-TransSeg presents an innovative paradigm for the segmentation of complex liver tumors, thus underscoring the significance of multi-phase CT data exploitation. The proposed MI-TransSeg network has the potential to assist radiologists in diagnosing liver tumors and assessing microvascular invasion.

Semi-supervised low-dose SPECT restoration using sinogram inner-structure aware graph neural network.

Objective.To mitigate the potential radiation risk, low-dose single photon emission computed tomography (SPECT) is of increasing interest. Numerous deep learning-based methods have been developed to perform low-dose imaging while maintaining image quality. However, most existing methods seldom explore the unique inner-structure inherent within sinograms. In addition, traditional supervised learning methods require large-scale labeled data, where the normal-dose data serves as annotation and is intractable to acquire in low-dose imaging. In this study, we aim to develop a novel sinogram inner-structure-aware semi-supervised framework for the task of low-dose SPECT sinogram restoration.Approach.The proposed framework retains the strengths of UNet, meanwhile introducing a sinogram-structure-based non-local neighbors graph neural network (SSN-GNN) module and a window-based K-nearest neighbors GNN (W-KNN-GNN) module to effectively exploit the inherent inner-structure within SPECT sinograms. Moreover, the proposed framework employs the mean teacher semi-supervised learning approach to leverage the information available in abundant unlabeled low-dose sinograms.Main results.The datasets exploited in this study were acquired from the (Extended Cardiac-Torso) XCAT anthropomorphic digital phantoms, which provide realistic images for imaging research of various modalities. Quantitative as well as qualitative results demonstrate that the proposed framework achieves superior performance compared to several state-of-the-art reconstruction methods. To further validate the effectiveness of the proposed framework, ablation and robustness experiments were also performed. The experimental results show that each component of the proposed framework effectively improves the model performance, and the framework exhibits superior robustness with respect to various noise levels. Besides, the proposed semi-supervised paradigm showcases the efficacy of incorporating supplementary unlabeled low-dose sinograms.Significance.The proposed framework improves the quality of low-dose SPECT reconstructed images by utilizing sinogram inner-structure and incorporating supplementary unlabeled data, which provides an important tool for dose reduction without sacrificing the image quality.

Dynamic light scattering microscopy sensing mitochondria dynamics for label-free detection of triple-negative breast cancer enhanced by deep learning.

We established a deep learning-based dynamic light scattering (DLS) microscopy sensing mitochondria dynamic for label-free identification of triple-negative breast cancer (TNBC) cells. The capacity of DLS microscopy to detect the intracellular motility of subcellular scatters was verified with the analysis of the autocorrelation function. We also conducted an in-depth examination of the impact of mitochondrial dynamics on DLS within TNBC cells, employing confocal fluorescent imaging to visualize the morphology of the mitochondria. Furthermore, we applied the DLS microscopy incorporating the two-stream deep learning method to differentiate the TNBC subtype and HER2 positive breast cancer subtype, with the classification accuracy achieving 0.89.

The Effects of Bacillus subtilis QST713 and ß-mannanase on growth performance, intestinal barrier function, and the gut microbiota in weaned piglets.

We investigated the effects of different Bacillus subtilis QST713 doses and a B. subtilis QST713 and ß-mannanase mix on growth performance, intestinal barrier function, and gut microbiota in weaned piglets. In total, 320 healthy piglets were randomly assigned to four groups: 1) control group (basal diet), 2) BS100 group (basal diet plus 100 mg/kg B. subtilis QST713), 3) BS200 group (basal diet plus 200 mg/kg B. subtilis QST713), and 4) a BS100XT group (basal diet plus 100 mg/kg B. subtilis QST713 and 150 mg/kg ß-mannanase). The study duration was 42 d. We showed that feed intake in weaned piglets on days 1 to 21 was increased in group BS100 (P < 0.05), and that the feed conversion ratio in group BS100XT animals decreased throughout the study (P < 0.05). In terms of microbial counts, the BS100XT group showed reduced Escherichia coli and Clostridium perfringens numbers on day 21 (P < 0.05). Moreover, no significant α-diversity differences were observed across all groups during the study (P > 0.05). However, principal coordinates analysis indicated clear separations in bacterial community structures across groups (analysis of similarities: P < 0.05) on days 21 and 42. Additionally, E-cadherin, occludin, and zonula occludens-1 (ZO-1) expression in piglet feces increased (P < 0.05) by adding B. subtilis QST713 and ß-mannanase to diets. Notably, this addition decreased short-chain fatty acid concentrations. In conclusion, B. subtilis QST713 addition or combined B. subtilis QST713 plus ß-mannanase effectively improved growth performance, intestinal barrier function, and microbial balance in weaned piglets.

The use of antibiotics in pig farming raises serious concerns in terms of antibiotic resistance. Consequently, alternative approaches such as probiotics, including Bacillus subtilis, and enzymes such as ß-mannanase, have been proposed to improve pig health and performance. In particular, B. subtilis improves gut microbiota and reduces the prevalence of harmful bacteria such as Escherichia coli and Clostridium perfringens. Similarly, ß-mannanase enhances feed digestibility and improves nutrient use in pigs. Thus, combined B. subtilis and ß-mannanase may provide synergistic effects toward pig performance and gut health. In this study, we showed that adding B. subtilis to a weaned piglet diet improved feed intake, while a B. subtilis and ß-mannanase mix reduced feed conversion ratios in weaned piglets.

Combining 3D skeleton data and deep convolutional neural network for balance assessment during walking.

Introduction: Balance impairment is an important indicator to a variety of diseases. Early detection of balance impairment enables doctors to provide timely treatments to patients, thus reduce their fall risk and prevent related disease progression. Currently, balance abilities are usually assessed by balance scales, which depend heavily on the subjective judgement of assessors. Methods: To address this issue, we specifically designed a method combining 3D skeleton data and deep convolutional neural network (DCNN) for automated balance abilities assessment during walking. A 3D skeleton dataset with three standardized balance ability levels were collected and used to establish the proposed method. To obtain better performance, different skeleton-node selections and different DCNN hyperparameters setting were compared. Leave-one-subject-out-cross-validation was used in training and validation of the networks. Results and Discussion: Results showed that the proposed deep learning method was able to achieve 93.33% accuracy, 94.44% precision and 94.46% F1 score, which outperformed four other commonly used machine learning methods and CNN-based methods. We also found that data from body trunk and lower limbs are the most important while data from upper limbs may reduce model accuracy. To further validate the performance of the proposed method, we migrated and applied a state-of-the-art posture classification method to the walking balance ability assessment task. Results showed that the proposed DCNN model improved the accuracy of walking balance ability assessment. Layer-wise Relevance Propagation (LRP) was used to interpret the output of the proposed DCNN model. Our results suggest that DCNN classifier is a fast and accurate method for balance assessment during walking.

Xylo-oligosaccharides improve the adverse effects of plant-based proteins on weaned piglet health by maintaining the intestinal barrier and inhibiting harmful bacterial growth.

Introduction: Piglets are more susceptible to weaning stress syndrome when fed high levels of plant-based proteins that contain abundant food antigens and anti-nutritional factors. Xylo-oligosaccharides (XOS) are a potential prebiotic that may improve the tolerance of weaned piglets to plant-based proteins. The aim of this study was to investigate the effects of XOS supplementation in high and low plant-based protein diets on growth performance, gut morphology, short-chain fatty acid (SCFA) production, and gut microbiota of weaned piglets. Methods: A total of 128 weanling piglets with an average body weight (BW) of 7.63 ± 0.45 kg were randomly allocated to one of the four dietary treatments in a 2 × 2 factorial arrangement, with two levels of plant-based proteins (d 1-14: 68.3 or 81.33%, d 15-28: 81.27 or 100%) and XOS complex (0 or 0.43%) over a 28-day trial. Results: The growth performance of piglets did not differ significantly among groups (P > 0.05). However, the diarrhea index of weaned piglets fed a high plant-based protein diet (HP) was significantly higher than that of those fed a low plant-based protein diet (LP) at days 1-14 and throughout the experimental period (P < 0.05). XOS treatment tended to reduce the diarrhea index at days 1-14 (P = 0.062) and during the whole experiment period (P = 0.083). However, it significantly increased the digestibility of organic matter at days 15-28 (P < 0.05). Moreover, dietary XOS supplementation increased ileal mucosa mRNA expression of occludin and ZO-1 (P < 0.05). Furthermore, the concentration of butyric acid (BA) in the cecal contents and in the concentrations of BA and valeric acid (VA) in colon contents were significantly elevated in the XOS groups (P < 0.05). Additionally, XOS optimized the gut flora by lowering the number of pathogenic bacteria such as p_Campylobacterota, thereby stabilizing the gut ecosystem. Discussion: In conclusion, the HP diet aggravated diarrhea in weaned piglets while the XOS diet alleviated it by improving nutrient digestibility, protecting intestinal morphology, and optimizing the gut flora.

Chemical composition, energy content and apparent total tract digestibility of extruded full fat soybean from different sources fed to non-gestating, gestating and lactating sows.

This experiment was conducted to determine the chemical composition, digestible energy (DE), metabolizable energy (ME) and the apparent total tract (ATTD) of nutrients in six extruded full fat soybean (EFSB) samples from different sources fed to non-gestating, gestating and lactating sows. Forty-two non-gestating sows (Landrace × Yorkshire; parity 3 to 5), 42 gestating sows (Landrace × Yorkshire; parity 3 to 5; day 90 of gestation) and 42 lactating sows (Landrace × Yorkshire; parity 3 to 5; day 6 of lactation) were assigned to seven dietary treatments including a corn-based diet and six diets containing 30.24% EFSB from different sources in a completely randomized design with six replicate sows per dietary treatment. Total fecal and urine collection method was used during non-gestation and gestation, and the index method was used during lactation (0.3% chromic oxide). Differences in the chemical composition of the six EFSB samples from different sources were mainly reflected in ether extract, ash, crude fiber, neutral detergent fiber (NDF), acid detergent fiber, total dietary fiber, insoluble dietary fiber, soluble dietary fiber, and vitamin and micro minerals content, with a coefficient of variation ≥8.37%. The potassium hydroxide solubility of the six EFSB samples varied from 66.60% to 85.55%. There were no differences in ATTD of NDF between different EFSB samples. Additionally, there were no differences in ME values and ME/DE ratios between different physiological stages, but ATTD of NDF were higher for non-gestating and gestating sows than lactating sows (P < 0.01). In conclusion, EFSB can be used as a high-quality energy ingredient with high DE and ME values when fed to sows. DE values of EFSB in non-gestating, gestating, and lactating sows were 20.50, 20.70, and 20.02 MJ/kg, respectively, while ME values of EFSB was 19.76 MJ/kg in both non-gestating and gestating sows.

Extruded full fat soybean (EFSB) is used as a high-quality ingredient in swine diets. However, the database of feedstuffs including EFSB is mainly based on growing pig models (NRC, 2012) because the nutritional value of EFSB in sows is poorly understood. The digestibility of feedstuffs differs between sows and growing pigs, and even between sows at different physiological stages. Therefore, we evaluated the digestibility of six EFSB samples from different sources in non-gestating, gestating and lactating sows. The results confirmed differences in digestibility of EFSB samples from different sources, and differences in the digestibility of EFSB between sows at different physiological stages.

Effects of feeding levels on ileal amino acid digestibility of extruded full fat soybeans in nongestating sows.

The objectives of this study were to investigate the effect of feeding levels on amino acid (AA) digestibility of extruded full fat soybeans (EFSB) fed to nongestating sows and to provide a reference for setting feed intake level when evaluating the quality of nutrients in the feed ingested by sows. Twelve nongestating sows (parity 3 to 5) were fitted with a T-cannula at the distal ileum. After recovery, sows were assigned to a replicated 6 × 3 incomplete Latin square design using two diets (nitrogen-free and EFSB) and three levels of feed intake (1.3, 2.0, and 3.4 times the maintenance requirement for metabolizable energy (ME)). The design included six dietary treatments and three periods, and each period contained two replicates for a total of six replicate sows per treatment. All diets contained 0.3% chromic oxide as an indigestible marker. In each period, ileal digesta samples were collected continuously for 12 h on days 6 and 7 after 5 d of acclimation to the experimental diet. Results of the experiment indicated that different feeding levels (1.3, 2.0, and 3.4 times the maintenance requirement for ME) had no effects on apparent ileal digestibility (AID) of AA and standardized ileal digestibility (SID) of AA, but feeding level did affect the endogenous AA loss estimated using the nitrogen-free diet method. Endogenous phenylalanine excretion was greater with 1.3 times than with 3.4 times the maintenance requirement for ME (P = 0.03), and endogenous tyrosine excretion was greater with 1.3 and 2.0 times than with 3.4 times the maintenance requirement for ME (P = 0.01). Increasing feed consumption resulted in greater loss of total endogenous AAs and crude protein (CP). In conclusion, feeding levels of 1.3 to 3.4 times the maintenance requirement for ME did not affect the AID and SID of AA of EFSB fed to nongestating sows.

Accurate evaluation of nutrient digestibility of feedstuffs by sows is the basis of accurate diet formulation, which can reduce waste of feed resources and decrease cost of breeding. The methodology of feedstuff evaluation for sows is based mainly on the growing pig model, but the digestive physiology differs significantly between sows and growing pigs. Additionally, the amount of feed (2 to 3 kg/d) given to nongestating or gestating sows differs significantly among studies, and the effect of feed intake on standardized ileal amino acid digestibility of sows has not been reported. Results indicate that feeding level did not affect standardized ileal amino acid digestibility of extruded full fat soybeans in nongestating sows.

Atypical architectural distortion detection in digital breast tomosynthesis: a computer-aided detection model with adaptive receptive field.

Objective. In digital breast tomosynthesis (DBT), architectural distortion (AD) is a breast lesion that is difficult to detect. Compared with typical ADs, which have radial patterns, identifying a typical ADs is more difficult. Most existing computer-aided detection (CADe) models focus on the detection of typical ADs. This study focuses on atypical ADs and develops a deep learning-based CADe model with an adaptive receptive field in DBT.Approach. Our proposed model uses a Gabor filter and convergence measure to depict the distribution of fibroglandular tissues in DBT slices. Subsequently, two-dimensional (2D) detection is implemented using a deformable-convolution-based deep learning framework, in which an adaptive receptive field is introduced to extract global features in slices. Finally, 2D candidates are aggregated to form the three-dimensional AD detection results. The model is trained on 99 positive cases with ADs and evaluated on 120 AD-positive cases and 100 AD-negative cases.Main results. A convergence-measure-based model and deep-learning model without an adaptive receptive field are reproduced as controls. Their mean true positive fractions (MTPF) ranging from 0.05 to 4 false positives per volume are 0.3846 ± 0.0352 and 0.6501 ± 0.0380, respectively. Our proposed model achieves an MTPF of 0.7148 ± 0.0322, which is a significant improvement (p< 0.05) compared with the other two methods. In particular, our model detects more atypical ADs, primarily contributing to the performance improvement.Significance. The adaptive receptive field helps the model improve the atypical AD detection performance. It can help radiologists identify more ADs in breast cancer screening.

The nutritive value of soybean meal from different sources for sows during mid- and late gestation.

A precise understanding of the nutritive value of soybean meal (SBM) for pregnant sow is required for accurate feeding. Hence, we evaluated the nutritive value of 11 SBM samples from different sources for sows during mid and late gestation. In total, 24 mid-gestating sows (parity three; 230.3 ± 12.0 kg on day 37 of gestation) and 24 late-gestating sows (parity three; 238.8 ± 20.9 kg on day 72 of gestation) were assigned to a replicated 12 × 3 Youden square design with 12 diets and 3 periods. The 12 diets included a corn-based diet and 11 diets containing 25.50% SBMs from different sources. After 5-d adaptation, urine and feces were collected for 5 d. Although the chemical characteristics of SBM varied between samples, no differences were observed in digestible energy (DE), metabolizable energy (ME), apparent total tract digestibility (ATTD) of dry matter, gross energy, crude fiber, and neutral detergent fiber values in SBMs fed to both animal groups. However, de-hulled SBM 4 from Brazil displayed greater ATTD for nitrogen (N) in late-gestating sows (P < 0.05); animals displayed significantly (P < 0.01) greater ME, ME:DE ratio, and N net utilization values when compared with mid-gestating sows. The chemical composition of SBMs can be used to predict DE and ME values. In conclusion, ME, ME:DE ratio, and N net utilization SBM values for late-gestating sows were greater than in mid-gestating sows. Therefore, we should consider differences in ME values for SBMs when formulating diets for sows in mid and late gestation periods.

Soybean meal (SBM) is the most commonly used protein source in swine diets, with high available energy. Sows have crucial roles in the pig industry, therefore precise knowledge of actual SBM nutritive values at different gestation stages is vital for efficient livestock production and management. In our study, we evaluated the nutritive value of SBMs from different sources in mid- and late-gestating sows, and generated prediction equations for digestible energy (DE) and metabolizable energy (ME) values. We identified no differences in DE and ME values in SBMs from different sources when fed to sows during mid- and late gestation. However, regardless of the pregnancy stage, DE and ME SBM values for sows, identified in this study, were greater than values published by NRC (2012). Also, ME, ME:DE ratio, and nitrogen (N) net utilization SBM values for late-gestating sows were greater than in mid-gestating sows. The chemical composition of SBMs can be used to predict DE and ME values. Our study data can be used to accurately formulate diets for pregnant sows.

Architectural distortion detection based on superior-inferior directional context and anatomic prior knowledge in digital breast tomosynthesis.

BACKGROUND: In 2020, breast cancer becomes the most leading diagnosed cancer all over the world. The burden is increasing in the prevention and treatment of breast cancer. Accurately detecting breast lesions in screening images is important for early detection of cancer. Architectural distortion (AD) is one of the breast lesions that need to be detected. PURPOSE: To develop a deep-learning-based computer-aided detection (CADe) model for AD in digital breast tomosynthesis (DBT). This model uses the superior-inferior directional context of DBT and anatomic prior knowledge to reduce false positive (FP). It can identify some negative samples that cannot be distinguished by deep learning features. METHODS: The proposed CADe model consists of three steps. In the first step, a deep learning detection network detects two-dimensional (2D) candidates of ADs in DBT slices with the inputs preprocessed by Gabor filters and convergence measure. In the second step, three-dimensional (3D) candidates are obtained by stacking 2D candidates along superior-inferior direction. In the last step, FP reduction for 3D candidates is implemented based on superior-inferior directional context and anatomic prior knowledge of breast. DBT data from 99 cases with AD were used as the training set to train the CADe model, and data from 208 cases were used as an independent test set (including 108 cases with AD and 100 cases without AD as the control group). The free-response receiver operating characteristic and mean true positive fraction (MTPF) in the range of 0.05-2.0 FPs per volume are used to evaluate the model. RESULTS: Compared with the baseline model based on convergence measure, our proposed method demonstrates significant improvement (MTPF: 0.2826 ± 0.0321 vs. 0.6640 ± 0.0399). Results of an ablation study show that our proposed context- and anatomy-based FP reduction methods improve the detection performance. The number of FPs per DBT volume reduces from 2.47 to 1.66 at 80% sensitivity after employing these two schemes. CONCLUSIONS: The deep learning model demonstrates practical value for AD detection. The results indicate that introducing superior-inferior directional context and anatomic prior knowledge into model can indeed reduce FPs and improve the performance of CADe model.

China's ecosystem service value in 1992-2018: Pattern and anthropogenic driving factors detection using Bayesian spatiotemporal hierarchy model.

Maintaining ecosystem services (ESs) and reducing ecosystem degradation are important goals for achieving sustainable development. However, under the influence of various anthropogenic factors, the total ecosystem service value (ESV) of China continues to decline, and the detailed processes involved in this decline are unclear. In this paper, a new long-term annual land cover dataset (the Climate Change Initiative Land Cover or CCI-LC dataset) with a spatial resolution of 300 m was employed to estimate the ESV of China, and Bayesian spatiotemporal hierarchy models were built to examine the detailed patterns and anthropogenic driving factors. From 1992 to 2018, the total ESV of China fluctuated and decreased from 3265.3 to 3253.29 billion US$ at an average rate of 0.55 billion US$ per year. Furthermore, the model revealed the spatiotemporal variations in the ESV pattern, and simultaneously detected the influences of 9 variables related to economic factors, population, infrastructure, energy, agriculture and ecological restoration, providing a convenient and effective method for ESV spatiotemporal analysis. The results enrich our understanding of the detailed spatiotemporal variation and anthropogenic driving factors underlying the declining ESV in China. These findings have substantial guiding implications for adjusting ecological regulation policies.

Weakly-supervised lesion analysis with a CNN-based framework for COVID-19.

Objective.Lesions of COVID-19 can be clearly visualized using chest CT images, and hence provide valuable evidence for clinicians when making a diagnosis. However, due to the variety of COVID-19 lesions and the complexity of the manual delineation procedure, automatic analysis of lesions with unknown and diverse types from a CT image remains a challenging task. In this paper we propose a weakly-supervised framework for this task requiring only a series of normal and abnormal CT images without the need for annotations of the specific locations and types of lesions.Approach.A deep learning-based diagnosis branch is employed for classification of the CT image and then a lesion identification branch is leveraged to capture multiple types of lesions.Main Results.Our framework is verified on publicly available datasets and CT data collected from 13 patients of the First Affiliated Hospital of Shantou University Medical College, China. The results show that the proposed framework can achieve state-of-the-art diagnosis prediction, and the extracted lesion features are capable of distinguishing between lesions showing ground glass opacity and consolidation.Significance.The proposed approach integrates COVID-19 positive diagnosis and lesion analysis into a unified framework without extra pixel-wise supervision. Further exploration also demonstrates that this framework has the potential to discover lesion types that have not been reported and can potentially be generalized to lesion detection of other chest-based diseases.

Explaining the longevity characteristics in China from a geographical perspective: A multi-scale geographically weighted regression analysis.

Longevity is a near-universal human aspiration that can affect moral progress and economic development at the social level. In rapidly developing China, questions about the geographical distribution and environmental factors of longevity phenomenon need to be answered more clearly. This study calculated the longevity index (LI), longevity index for females (LIF) and longevity index for males (LIM) based on the percentage of the long-lived population among the total number of elderly people to investigate regional and gender characteristics at the county level in China. A new multi-scale geographically weighted regression (MGWR) model and four possible geographical environmental factors were applied to explore environmental effects. The results indicate that the LIs of 2838 counties ranged from 1.3% to 16.3%, and the distribution showed obvious regional and gender differences. In general, the LI was high in the East and low in the West, and the LIF was higher than the LIM in 2614 counties (92.1%). The MGWR model performed well explaining that geographical environmental factors, including topographic features, vegetation conditions, human social activity and air pollution factors have a variable influence on longevity at different spatial scales and in different regions. These findings enrich our understanding of the spatial distribution, gender differences and geographical environmental effects on longevity in China, which provides an important reference for people interested in the variations in the associations between different geographical factors.

Genomic Profiling Reveals the Molecular Landscape of Gastrointestinal Tract Cancers in Chinese Patients.

Gastrointestinal tract cancers have high incidence and mortality in China, but their molecular characteristics have not been fully investigated. We sequenced 432 tumor samples from the colorectum, stomach, pancreas, gallbladder, and biliary tract to investigate cancer-related mutations and detail the landscape of microsatellite instability (MSI), tumor mutation burden (TMB), and chromosomal instability (CIN). We observed the highest TMB in colorectal and gastric cancers and the lowest TMB in gastrointestinal stromal tumors (GISTs). Twenty-four hyper-mutated tumors were identified only in colorectal and gastric cancers, with a significant enrichment of mutations in the polymerase genes (POLE, POLD1, and POLH) and mismatch repair (MMR) genes. Additionally, CIN preferentially occurred in colorectal and gastric cancers, while pancreatic, gallbladder, and biliary duct cancers had a much lower CIN. High CIN was correlated with a higher prevalence of malfunctions in chromosome segregation and cell cycle genes, including the copy number loss of WRN, NAT1, NF2, and BUB1B, and the copy number gain of MYC, ERBB2, EGFR, and CDK6. In addition, TP53 mutations were more abundant in high-CIN tumors, while PIK3CA mutations were more frequent in low-CIN tumors. In colorectal and gastric cancers, tumors with MSI demonstrated much fewer copy number changes than microsatellite stable (MSS) tumors. In colorectal and gastric cancers, the molecular characteristics of tumors revealed the mutational diversity between the different anatomical origins of tumors. This study provides novel insights into the molecular landscape of Chinese gastrointestinal cancers and the genetic differences between tumor locations, which could be useful for future clinical patient stratification and targeted interventions.

Influence of low tumor content on tumor mutational burden estimation by whole-exome sequencing and targeted panel sequencing.

BACKGROUND: Tumor mutational burden (TMB) is a promising biomarker for stratifying patient subpopulation who would benefit from immune checkpoint blockade (ICB) therapies. Although great efforts have been made for standardizing TMB measurement, mutation calling and TMB quantification can be challenging in samples with low tumor content including liquid biopsies. The effect of varying tumor content on TMB estimation by different assay methods has never been systematically investigated. METHOD: We established a series of reference standard DNA samples derived from 11 pairs of tumor-normal matched human cell lines across different cancer types. Each tumor cell line was mixed with its matched normal at 0% (control), 1%, 2%, 5%, and 10% mass-to-mass ratio to mimic the clinical samples with low tumor content. TMB of these reference standards was evaluated by both ∼1000× whole-exome sequencing (wesTMB) and targeted panel sequencing (psTMB) at four different vendors. Both regression and classification analyses of TMB were performed for theoretical investigation and clinical practice purposes. RESULTS: Linear regression model was established that demonstrated in silico psTMB determined by regions of interest (ROI) as a great representative of wesTMB based on TCGA dataset. It was also true in our reference standard samples as the predicted psTMB interval based on the observed wesTMB captured the intended 90% of the in silico psTMB values. Although ∼1000× deep WES was applied, reference standard samples with less than 5% of tumor proportions are below the assay limit of detection (LoD) of wesTMB quantification. However, predicted wesTMB based on observed psTMB accurately classify (>0.97 AUC) for TMB high and low patient stratification even in samples with 2% of tumor content, which is more clinically relevant, as TMB determination should be a qualitative assay for TMB high and low patient classification. One targeted panel sequencing vendor using an optimized blood psTMB pipeline can further classify TMB status accurately (>0.82 AUC) in samples with only 1% of tumor content. CONCLUSIONS: We developed a linear model to establish the quantitative correlation between wesTMB and psTMB. A set of DNA reference standards was produced in aid to standardize TMB measurements in samples with low tumor content across different targeted sequencing panels. This study is a significant contribution aiming to harmonize TMB estimation and extend its future application in clinical samples with low tumor content including liquid biopsy.

Can a Computer-Aided Mass Diagnosis Model Based on Perceptive Features Learned From Quantitative Mammography Radiology Reports Improve Junior Radiologists' Diagnosis Performance? An Observer Study.

Radiologists' diagnostic capabilities for breast mass lesions depend on their experience. Junior radiologists may underestimate or overestimate Breast Imaging Reporting and Data System (BI-RADS) categories of mass lesions owing to a lack of diagnostic experience. The computer-aided diagnosis (CAD) method assists in improving diagnostic performance by providing a breast mass classification reference to radiologists. This study aims to evaluate the impact of a CAD method based on perceptive features learned from quantitative BI-RADS descriptions on breast mass diagnosis performance. We conducted a retrospective multi-reader multi-case (MRMC) study to assess the perceptive feature-based CAD method. A total of 416 digital mammograms of patients with breast masses were obtained from 2014 through 2017, including 231 benign and 185 malignant masses, from which we randomly selected 214 cases (109 benign, 105 malignant) to train the CAD model for perceptive feature extraction and classification. The remaining 202 cases were enrolled as the test set for evaluation, of which 51 patients (29 benign and 22 malignant) participated in the MRMC study. In the MRMC study, we categorized six radiologists into three groups: junior, middle-senior, and senior. They diagnosed 51 patients with and without support from the CAD model. The BI-RADS category, benign or malignant diagnosis, malignancy probability, and diagnosis time during the two evaluation sessions were recorded. In the MRMC evaluation, the average area under the curve (AUC) of the six radiologists with CAD support was slightly higher than that without support (0.896 vs. 0.850, p = 0.0209). Both average sensitivity and specificity increased (p = 0.0253). Under CAD assistance, junior and middle-senior radiologists adjusted the assessment categories of more BI-RADS 4 cases. The diagnosis time with and without CAD support was comparable for five radiologists. The CAD model improved the radiologists' diagnostic performance for breast masses without prolonging the diagnosis time and assisted in a better BI-RADS assessment, especially for junior radiologists.

Deep learning of mammary gland distribution for architectural distortion detection in digital breast tomosynthesis.

Computer aided detection (CADe) for breast lesions can provide an important reference for radiologists in breast cancer screening. Architectural distortion (AD) is a type of breast lesion that is difficult to detect. A majority of CADe methods focus on detecting the radial pattern, which is a main characteristic of typical ADs. However, a few atypical ADs do not exhibit such a pattern. To improve the performance of CADe for typical and atypical ADs, we propose a deep-learning-based model that used mammary gland distribution as prior information to detect ADs in digital breast tomosynthesis (DBT). First, information about gland distribution, including the Gabor magnitude, the Gabor orientation field, and a convergence map, were produced using a bank of Gabor filters and convergence measures. Then, this prior information and an original slice were input into a Faster R-CNN detection network to obtain the 2-D candidates for each slice. Finally, a 3-D aggregation scheme was employed to fuse these 2-D candidates as 3-D candidates for each DBT volume. Retrospectively, 64 typical AD volumes, 74 atypical AD volumes, and 127 normal volumes were collected. Six-fold cross-validation and mean true positive fraction (MTPF) were used to evaluate the model. Compared to an existing convergence-based model, our proposed model achieved an MTPF of 0.53 ± 0.04, 0.61 ± 0.05, and 0.45 ± 0.04 for all DBT volumes, typical + normal volumes, and atypical + normal volumes, respectively. These results were significantly better than those of 0.36 ± 0.03, 0.46 ± 0.04, and 0.28 ± 0.04 for a convergence-based model (p ≪ 0.01). These results indicate that employing the prior information of gland distribution and a deep learning method can improve the performance of CADe for AD.

Quantitative characterization of tumor cell-free DNA shortening.

BACKGROUND: Previous studies found that cell-free DNA (cfDNA) generated from tumors was shorter than that from healthy cells, and selecting short cfDNA could enrich for tumor cfDNA and improve its usage in early cancer diagnosis and treatment monitoring; however, the underlying mechanism of shortened tumor cfDNA was still unknown, which potentially limits its further clinical application. RESULTS: Using targeted sequencing of cfDNA in a large cohort of solid tumor patient, sequencing reads harboring tumor-specific somatic mutations were isolated to examine the exact size distribution of tumor cfDNA. For the majority of studied cases, 166 bp remained as the peak size of tumor cfDNA, with tumor cfDNA showing an increased proportion of short fragments (100-150 bp). Less than 1% of cfDNA samples were found to be peaked at 134/144 bp and independent of tumor cfDNA purity. Using whole-genome sequencing of cfDNA, we discovered a positive correlation between cfDNA shortening and the magnitude of chromatin inaccessibility, as measured by transcription, DNase I hypersensitivity, and histone modifications. Tumor cfDNA shortening occurred simultaneously at both 5' and 3' ends of the DNA wrapped around nucleosomes. CONCLUSIONS: Tumor cfDNA shortening exhibited two distinctive modes. Tumor cfDNA purity and chromatin inaccessibility were contributing factors but insufficient to trigger a global transition from 166 bp dominant to 134/144 bp dominant phenotype.

Automated fibroglandular tissue segmentation in breast MRI using generative adversarial networks.

Fibroglandular tissue (FGT) segmentation is a crucial step for quantitative analysis of background parenchymal enhancement (BPE) in magnetic resonance imaging (MRI), which is useful for breast cancer risk assessment. In this study, we develop an automated deep learning method based on a generative adversarial network (GAN) to identify the FGT region in MRI volumes and evaluate its impact on a specific clinical application. The GAN consists of an improved U-Net as a generator to generate FGT candidate areas and a patch deep convolutional neural network (DCNN) as a discriminator to evaluate the authenticity of the synthetic FGT region. The proposed method has two improvements compared to the classical U-Net: (1) the improved U-Net is designed to extract more features of the FGT region for a more accurate description of the FGT region; (2) a patch DCNN is designed for discriminating the authenticity of the FGT region generated by the improved U-Net, which makes the segmentation result more stable and accurate. A dataset of 100 three-dimensional (3D) bilateral breast MRI scans from 100 patients (aged 22-78 years) was used in this study with Institutional Review Board (IRB) approval. 3D hand-segmented FGT areas for all breasts were provided as a reference standard. Five-fold cross-validation was used in training and testing of the models. The Dice similarity coefficient (DSC) and Jaccard index (JI) values were evaluated to measure the segmentation accuracy. The previous method using classical U-Net was used as a baseline in this study. In the five partitions of the cross-validation set, the GAN achieved DSC and JI values of 87.0 ± 7.0% and 77.6 ± 10.1%, respectively, while the corresponding values obtained through by the baseline method were 81.1 ± 8.7% and 69.0 ± 11.3%, respectively. The proposed method is significantly superior to the previous method using U-Net. The FGT segmentation impacted the BPE quantification application in the following manner: the correlation coefficients between the quantified BPE value and BI-RADS BPE categories provided by the radiologist were 0.46 ± 0.15 (best: 0.63) based on GAN segmented FGT areas, while the corresponding correlation coefficients were 0.41 ± 0.16 (best: 0.60) based on baseline U-Net segmented FGT areas. BPE can be quantified better using the FGT areas segmented by the proposed GAN model than using the FGT areas segmented by the baseline U-Net.