Combination Of Image Features Research Articles

The Contribution of White Matter Diffusion and Cortical Perfusion Pathology to Vascular Cognitive Impairment: A Multimode Imaging-Based Machine Learning Study.

Widespread impairments in white matter and cerebrovascular integrity have been consistently implicated in the pathophysiology of patients with small vessel disease (SVD). However, the neural circuit mechanisms that underlie the developing progress of clinical cognitive symptoms remain largely elusive. Here, we conducted cross-modal MRI scanning including diffusion tensor imaging and arterial spin labeling in a cohort of 113 patients with SVD, which included 74 patients with vascular mild cognitive impairment (vMCI) and 39 patients without vMCI symptoms, and hence developed multimode imaging-based machine learning models to identify markers that discriminated SVD subtypes. Diffusion and perfusion features, respectively, extracted from individual white matter and gray matter regions were used to train three sets of classifiers in a nested 10-fold fashion: diffusion-based, perfusion-based, and combined diffusion-perfusion-based classifiers. We found that the diffusion-perfusion combined classifier achieved the highest accuracy of 72.57% with leave-one-out cross-validation, with the diffusion features largely spanning the capsular lateral pathway of the cholinergic tracts, and the perfusion features mainly distributed in the frontal-subcortical-limbic areas. Furthermore, diffusion-based features within vMCI group were associated with performance on executive function tests. We demonstrated the superior accuracy of using diffusion-perfusion combined multimode imaging features for classifying vMCI subtype out of a cohort of patients with SVD. Disruption of white matter integrity might play a critical role in the progression of cognitive impairment in patients with SVD, while malregulation of coritcal perfusion needs further study.

Improved heterogeneous data fusion and multi‐scale feature selection method for lung cancer subtype classification

SummaryThe diagnosis of the disease requires a variety of data and indicators. In order to make the computer perform intelligent computation and diagnosis from a doctor's perspective, complementary information between different modal data needs to be taken into account. Meanwhile, the redundancy features with key information should be selected in order to reduce the complexity of calculation. In this study, an adaptive dynamic loss function is proposed to weight different scales in the multi‐scale expansion network of pathological images according to the doctor's diagnosis process. And an ant colony algorithm based on maximum information coefficient correlation was designed for unsupervised feature selection of fusion features combined image feature and patient differential genes. Experimental results show that the addition of pathological image information and genetic information plays an important role in the classification of lung cancer subtypes. Compared with other feature selection methods, the proposed algorithm can quickly converge. Combining pathological image and gene expression matrix for cancer diagnosis can improve the diagnostic accuracy of specific patients, with an accuracy of 95.62 and AUC achieves 0.897. The proposed method has high effectiveness and superior performance in the classification of lung cancer.

Implementation of early and late fusion methods for content-based image retrieval

Content-Based Image Retrieval (CBIR) systems retrieve images from the image repository or database in which they are visually similar to the query image. CBIR plays an important role in various fields such as medical diagnosis, crime prevention, web-based searching, and architecture. CBIR consists mainly of two stages: The first is the extraction of features and the second is the matching of similarities. There are several ways to improve the efficiency and performance of CBIR, such as segmentation, relevance feedback, expansion of queries, and fusion-based methods. The literature has suggested several methods for combining and fusing various image descriptors. In general, fusion strategies are typically divided into two groups, namely early and late fusion strategies. Early fusion is the combination of image features from more than one descriptor into a single vector before the similarity computation, while late fusion refers either to the combination of outputs produced by various retrieval systems or to the combination of different rankings of similarity. In this study, a group of color and texture features is proposed to be used for both methods of fusion strategies. Firstly, an early combination of eighteen color features and twelve texture features are combined into a single vector representation and secondly, the late fusion of three of the most common distance measures are used in the late fusion stage. Our experimental results on two common image datasets show that our proposed method has good performance retrieval results compared to the traditional way of using single features descriptor and also has an acceptable retrieval performance compared to some of the state-of-the-art methods. The overall accuracy of our proposed method is 60.6% and 39.07% for Corel-1K and GHIM-10K ‎datasets, respectively.

A machine learning approach to distinguish between knees without and with osteoarthritis using MRI-based radiomic features from tibial bone

ObjectivesOur aim was to assess the ability of semi-automatically extracted magnetic resonance imaging (MRI)–based radiomic features from tibial subchondral bone to distinguish between knees without and with osteoarthritis.MethodsThe right knees of 665 females from the population-based Rotterdam Study scanned with 1.5T MRI were analyzed. A fast imaging employing steady-state acquisition sequence was used for the quantitative bone analyses. Tibial bone was segmented using a method that combines multi-atlas and appearance models. Radiomic features related to the shape and texture were calculated from six volumes of interests (VOIs) in the proximal tibia. Machine learning–based Elastic Net models with 10-fold cross-validation were used to distinguish between knees without and with MRI Osteoarthritis Knee Score (MOAKS)–based tibiofemoral osteoarthritis. Performance of the covariate (age and body mass index), image features, and combined covariate + image features models were assessed using the area under the receiver operating characteristic curve (ROC AUC).ResultsOf 665 analyzed knees, 76 (11.4%) had osteoarthritis. An ROC AUC of 0.68 (95% confidence interval (CI): 0.60–0.75) was obtained using the covariate model. The image features model yielded an ROC AUC of 0.80 (CI: 0.73–0.87). The model that combined image features from all VOIs and covariates yielded an ROC AUC of 0.80 (CI: 0.73–0.87).ConclusionOur results suggest that radiomic features are useful imaging biomarkers of subchondral bone for the diagnosis of osteoarthritis. An advantage of assessing bone on MRI instead of on radiographs is that other tissues can be assessed simultaneously.Key Points• Subchondral bone plays a role in the osteoarthritis disease processes.• MRI radiomics is a potential method for quantifying changes in subchondral bone.• Semi-automatically extracted radiomic features of tibia differ between subjects without and with osteoarthritis.

Machine learning for predicting knee osteoarthritis progression over 8 years using combined MR imaging features, demographics, and clinical factors: data from the osteoarthritis initiative

Purpose: Osteoarthritis (OA) is a heterogeneous joint disease that affects approximately 250 million people and causes severe disability. The ability to predict which patients will develop knee OA would enable preventive measures (such as weight loss or lifestyle changes) prior to disease development, with prospects of preventing disability and pain associated with OA progression. Established risk factors for OA include obesity, genetic predisposition, and joint injury; however, imaging features have not been well integrated.

Combined clinical and imaging features better predict the critical outcomes of patients with SARS-COV-2.

The purpose of this study was to investigate the predictive value of combined clinical and imaging features, compared with the clinical or radiological risk factors only. Moreover, the expected results aimed to improve the identification of severe acute respiratory syndrome coronavirus-2 (SARS-COV-2) patients who may have critical outcomes.This retrospective study included laboratory-confirmed SARS-COV-2 cases between January 18, 2020, and February 16, 2020. The patients were divided into 2 groups with noncritical illness and critical illness regarding severity status within the hospitalization. Univariable and multivariable logistic regression models were used to explore the risk factors associated with clinical and radiological outcomes in patients with SARS-COV-2. The ROC curves were performed to compare the prediction performance of different factors.A total of 180 adult patients in this study included 20 critical patients and 160 noncritical patients. In univariate logistic regression analysis, 15 risk factors were significantly associated with critical outcomes. Of importance, C-reactive protein (1.051, 95% confidence interval 1.024-1.078), D-dimer (1.911, 95% CI, 1.050–3.478), and CT score (1.29, 95% CI, 1.053–1.529) on admission were independent risk factors in multivariate analysis. The combined model achieved a better performance in disease severity prediction (P = .05).CRP, D-dimer, and CT score on admission were independent risk factors for critical illness in adults with SARS-COV-2. The combined clinical and radiological model achieved better predictive performance than clinical or radiological factors alone.

Brain imaging features of children with Hoyeraal-Hreidarsson syndrome.

ObjectiveThis study aimed to summarize the magnetic resonance imaging (MRI) and computed tomography (CT) features of the central nervous system (CNS) in children with Hoyeraal‐Hreidarsson syndrome.MethodsThe imaging and clinical data of four children diagnosed with Hoyeraal‐Hreidarsson syndrome by clinical and laboratory tests in the Guangzhou Women and Children's Medical Center were gathered and analyzed retrospectively. The clinical manifestations and CNS imaging features of Hoyeraal‐Hreidarsson syndrome were summarized based on our results and a literature review.ResultsOur results showed that delayed development, skin pigmentation, nail/toenail dystrophy, thrombocytopenia, and anemia are the most observed clinical presentations of Hoyeraal‐Hreidarsson syndrome. Important findings on CNS imaging showed that all patients had cerebellar hypoplasia, delayed myelination, hydrocephalus, brain atrophy, and calcification. The gene mutations in all cases were consistent with those of dyskeratosis congenita, including TINF2 mutations in three cases and DKC1 mutations in one case.ConclusionHoyeraal‐Hreidarsson syndrome is a severe variant of dyskeratosis congenita. Both DKC1 and TINF2 mutations can lead to the phenotypes of Hoyeraal‐Hreidarsson syndrome. In our study, CNS imaging revealed that cerebellar hypoplasia has an important diagnostic value for Hoyeraal‐Hreidarsson syndrome while delayed myelination, calcification of the parenchyma, brain atrophy, and hydrocephalus are also important findings on CNS imaging. Combining imaging features with clinical and laboratory indicators can assist the diagnosis of Hoyeraal‐Hreidarsson syndrome.

Imaging-Based Staging of Hepatic Fibrosis in Patients with Hepatitis B: A Dynamic Radiomics Model Based on Gd-EOB-DTPA-Enhanced MRI.

Accurate grading of liver fibrosis can effectively assess the severity of liver disease and help doctors make an appropriate diagnosis. This study aimed to perform the automatic staging of hepatic fibrosis on patients with hepatitis B, who underwent gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid (Gd-EOB-DTPA)-enhanced magnetic resonance imaging with dynamic radiomics analysis. The proposed dynamic radiomics model combined imaging features from multi-phase dynamic contrast-enhanced (DCE) images and time-domain information. Imaging features were extracted from the deep learning-based segmented liver volume, and time-domain features were further explored to analyze the variation in features during contrast enhancement. Model construction and evaluation were based on a 132-case data set. The proposed model achieved remarkable performance in significant fibrosis (fibrosis stage S1 vs. S2–S4; accuracy (ACC) = 0.875, area under the curve (AUC) = 0.867), advanced fibrosis (S1–S2 vs. S3–S4; ACC = 0.825, AUC = 0.874), and cirrhosis (S1–S3 vs. S4; ACC = 0.850, AUC = 0.900) classifications in the test set. It was more dominant compared with the conventional single-phase or multi-phase DCE-based radiomics models, normalized liver enhancement, and some serological indicators. Time-domain features were found to play an important role in the classification models. The dynamic radiomics model can be applied for highly accurate automatic hepatic fibrosis staging.

Detection, Attribution and Localization of GAN Generated Images

Recent advances in Generative Adversarial Networks (GANs) have led to the creation of realistic-looking digital images that pose a major challenge to their detection by humans or computers. GANs are used in a wide range of tasks, from modifying small attributes of an image (StarGAN [14]), transferring attributes between image pairs (CycleGAN [92]), as well as generating entirely new images (ProGAN [37], StyleGAN [38], SPADE/GauGAN [65]). In this paper, we propose a novel approach to detect, attribute and localize GAN generated images that combines image features with deep learning methods. For every image, co-occurrence matrices are computed on neighborhood pixels of RGB channels in different directions (horizontal, vertical and diagonal). A deep learning network is then trained on these features to detect, attribute and localize these GAN generated/manipulated images. A large scale evaluation of our approach on 5 GAN datasets comprising over 2.76 million images (ProGAN, StarGAN, CycleGAN, StyleGAN and SPADE/GauGAN) shows promising results in detecting GAN generated images.

Chinese description generation of dual attention images based on multi-modal fusion

In view of the lack of attention to image features in most of the current research on image Chinese description generation, the low quality of image details, and the low accuracy of generated language description, a multi-modal fusion dual attention image description generation model is proposed. The model uses the pre-trained VGG19 network to extract image features, LSTM extracts key information features of text, and combines image features and text features through attention stitching to obtain the attention weight of each spatial position of the image, and then input it into the LSTM language model Generate descriptive sentences. In the AI Challenger 2017 test set, the BLEU-1 and CIDEr scores of this algorithm can reach 71.2 and 106.1 respectively, which are better than baseline models based on a single attention structure, such as NIC, and perform manual observation and comparison of the effects of several models. The model can generate natural language descriptions that can better represent the details of the picture.

FlyIT: Drosophila Embryogenesis Image Annotation based on Image Tiling and Convolutional Neural Networks.

With the rise of image-based transcriptomics, spatial gene expression data has become increasingly important for understanding gene regulations from the tissue level down to the cell level. Especially, the gene expression images of Drosophila embryos provide a new data source in the study of Drosophila embryogenesis. It is imperative to develop automatic annotation tools since manual annotation is labor-intensive and requires professional knowledge. Although a lot of image annotation methods have been proposed in the computer vision field, they may not work well for gene expression images, due to the great difference between these two annotation tasks. Besides the apparent difference on images, the annotation is performed at the gene level rather than the image level, where the expression patterns of a gene are recorded in multiple images. Moreover, the annotation terms often correspond to local expression patterns of images, yet they are assigned collectively to groups of images and the relations between the terms and single images are unknown. In order to learn the spatial expression patterns comprehensively for genes, we propose a new method, called FlyIT (image annotation based on Image Tiling and convolutional neural networks for fruit Fly). We implement two versions of FlyIT, learning at image-level and gene-level, respectively. The gene-level version employs an image tiling strategy to get a combined image feature representation for each gene. FlyIT uses a pre-trained ResNet model to obtain feature representation and a new loss function to deal with the class imbalance problem. As the annotation of Drosophila images is a multi-label classification problem, the new loss function considers the difficulty levels for recognizing different labels of the same sample and adjusts the sample weights accordingly. The experimental results on the FlyExpress database show that both the image tiling strategy and the deep architecture lead to the great enhancement of the annotation performance. FlyIT outperforms the existing annotators by a large margin (over 9 percent on AUC and 12 percent on macro F1 for predicting the top 10 terms). It also shows advantages over other deep learning models, including both single-instance and multi-instance learning frameworks.

Few-shot learning with saliency maps as additional visual information

Few-shot learning aims to learn to recognize new object categories from few training examples. Recently, few-shot learning methods have made significant progress. However, most of these methods are based on the concept of learning relations between only the image features in order to recognize objects and this alone may not be sufficient due to the training data scarcity. Therefore, this study focuses on providing saliency maps as additional visual information that describes the shape of the objects and supports few-shot visual learning. In this paper, we propose a simple few-shot learning method called Few-shot Learning with Saliency Maps as Additional Visual Information (SMAVI). Our method encodes the images and the saliency maps, then it learns the deep relations between the combined image features and saliency map features of the objects, where the saliency maps are extracted from the images using a saliency network. The experimental results show that the proposed method outperforms the related state of the art methods on standard few-shot learning datasets.

A computed tomography-based nomogram to predict pneumothorax caused by preoperative localization of ground glass nodules using hook wire.

To develop and validate a CT-based nomogram to predict the occurrence of loculated pneumothorax due to hook wire placement. Patients (n = 177) were divided into pneumothorax (n = 72) and non-pneumothorax (n = 105) groups. Multivariable logistic regression analysis was applied to build a clinical prediction model using significant predictors identified by univariate analysis of imaging features and clinical factors. Receiver operating characteristic (ROC) was applied to evaluate the discrimination of the nomogram, which was calibrated using calibration curve. Based on the results of multivariable regression analysis, transfissure approach [odds ratio (OR): 757.94; 95% confidence interval CI (21.20-27099.30) p < 0.0001], transemphysema [OR: 116.73; 95% CI (12.34-1104.04) p < 0.0001], localization of multiple nodules [OR: 8.04; 95% CI (2.09-30.89) p = 0.002], and depth of nodule [OR: 0.77; 95% CI (0.71-0.85) p < 0.0001] were independent risk factors for pneumothorax and were included in the predictive model (p < 0.05). The area under the ROC curve value for the nomogram was 0.95 [95% CI (0.92-0.98)] and the calibration curve indicated good consistency between risk predicted using the model and actual risk. A CT-based nomogram combining imaging features and clinical factors can predict the probability of pneumothorax before localization of ground-glass nodules. The nomogram is a decision-making tool to prevent pneumothorax and determine whether to proceed with further treatment. A nomogram composed of transfissure, transemphysema, multiple nodule localization, and depth of nodule has been developed to predict the probability of pneumothorax before localization of GGNs.

Contrast-enhanced cone beam breast CT features of breast cancers: correlation with immunohistochemical receptors and molecular subtypes.

To investigate the association of contrast-enhanced cone beam breast CT (CE-CBBCT) features, immunohistochemical (IHC) receptors, and molecular subtypes in breast cancer. In this retrospective study, patients who underwent preoperative CE-CBBCT and received complete IHC results were analyzed. CE-CBBCT features were evaluated by two radiologists. Observer reproducibility and feature reliability were assessed. The association between CE-CBBCT features, IHC receptors, and molecular subtypes was analyzed using the chi-square, Mann-Whitney, and Kruskal-Wallis tests. Multivariate logistic regression was performed to assess the ability of combined imaging features to discriminate molecular subtypes. ROC curve was used to evaluate prediction performance. A total of 240 invasive cancers identified in 211 women were enrolled. Molecular subtypes of breast cancer were significantly associated with focality number of lesions, lesion type, tumor size, lesion density, internal enhancement pattern, degree of lesion enhancement (ΔHU), mass shape, spiculation, calcifications, calcification distribution, and increased peripheral vascularity of lesion (all p < 0.005), some of which also helped to differentiate IHC receptor status. A multivariate logistic regression model showed that tumor size (odds ratio, OR = 1.244), mass shape (OR = 0.311), spiculation (OR = 0.159), and internal enhancement pattern (OR = 0.227) were associated with differentiation between luminal and non-luminal subtypes (AUC = 0.809). Combined CE-CBBCT features, including lesion type (OR = 0.118), calcifications (OR = 0.181), and ΔHU (OR = 0.962), could be significant indicators of triple-negative versus HER-2-enriched subtypes (AUC = 0.913). CE-CBBCT features have the potential to help predict IHC receptor status and distinguish molecular subtypes of breast cancer, which could in turn help to develop individual treatment decisions and prognosis predictions. • A total of 11 CE-CBBCT features were associated with molecular subtypes, some of which also helped to differentiate IHC receptor status. • Tumor size, irregular mass shape, spiculation, and internal enhancement pattern could help identify luminal subtype. • Lesion type, calcification, and ΔHU could be significant indicators of HER-2- enriched versus triple-negative breast cancers.

Temporal relationship between serial RT-PCR results and serial chest CT imaging, and serial CT changes in coronavirus 2019 (COVID-19) pneumonia: a descriptive study of 155 cases in China.

ObjectiveTo determine CT’s role in the early detection of COVID-19 infection and serial CT changes in the disease course in patients with COVID-19 pneumonia.MethodsFrom January 21 to February 18, 2020, all of the patients who were suspected of novel coronavirus infection and verified by RT-PCR tests were retrospectively enrolled in our study. All of the patients underwent serial RT-PCR tests and serial CT imaging. The temporal relationship between the serial RT-PCR results (negative conversion to positive, positive to negative) and serial CT imaging was investigated, and serial CT changes were evaluated.ResultsA total of 155 patients with confirmed COVID-19 pneumonia were evaluated. Chest CT detection time of COVID-19 pneumonia was 2.61 days earlier than RT-PCR test (p = 0.000). The lung CT improvement time was significantly shorter than that of RT-PCR conversion to negative (p = 0.000). Three stages were identified from the onset of the initial symptoms: stage 1 (0–3 days), stage 2 (4–7 days), and stage 3 (8–14 days and later). Ground glass opacity (GGO) was predominant in stage 1, then consolidation and crazy paving signs were dramatically increased in stage 2. In stage 3, fibrotic lesions were rapidly increased. There were significant differences in the main CT features (p = 0.000), number of lobes involved (p = 0.001), and lesion distribution (p = 0.000) among the different stages.ConclusionChest CT detected COVID-19 pneumonia earlier than the RT-PCR results and can be used to monitor disease course. Combining imaging features with epidemiology history and clinical information could facilitate the early diagnosis of COVID-19 pneumonia.Key Points • The chest CT detection time of COVID-19 pneumonia was 2.61 days earlier than that of an initial RT-PCR positive result (t = − 7.31, p = 0.000). • The lung CT improvement time was significantly shorter than that of RT-PCR conversion to negative (t = − 4.72, p = 0.000). • At the early stage (0–3 days), the CT features of COVID-19 were predominantly GGO and small-vessel thickening; at stage 2 (4–7 days), GGO evolved to consolidation and crazy paving signs. At stage 3 (8–14 days and later), fibrotic lesions significantly increased, accompanied by consolidation, GGO, and crazy paving signs.

Diabetic Peripheral Neuropathy Risk Assessment using Digital Fundus Photographs and Machine Learning.

In this work, we demonstrate a novel approach to assessing the risk of Diabetic Peripheral Neuropathy (DPN) using only the retinal images of the patients. Our methodology consists of convolutional neural network feature extraction, dimensionality reduction and feature selection with random projections, combination of image features to case-level representations, and the training and testing of a support vector machine classifier. Using clinical diagnosis as ground truth for DPN, we achieve an overall accuracy of 89% on a held-out test set, with sensitivity reaching 78% and specificity reaching 95%.

Combination of Imaging Features and Clinical Biomarkers Predicts Positive Pathology and Microbiology Findings Suggestive of Spondylodiscitis in Patients Undergoing Image-Guided Percutaneous Biopsy.

Pathology and microbiology results for suspected spondylodiscitis on MR imaging are often negative in up to 70% of cases. We aimed to predict whether MR imaging features will add diagnostic value when combined with clinical biomarkers to predict positive findings of spondylodiscitis on pathology and/or microbiology from percutaneous biopsy. In this retrospective single-center institutional review board-approved study, patients with radiologically suspected spondylodiscitis and having undergone percutaneous biopsies were assessed. Demographic characteristics, laboratory values, and tissue and blood cultures were collected. Pathology and microbiology results were used as end points. Three independent observers provided MR imaging-based scoring for typical MR imaging features for spondylodiscitis. Multivariate logistic regression and receiver operating characteristic analysis were performed to determine an optimal combination of imaging and clinical biomarkers in predicting positive findings on pathology and/or microbiology from percutaneous biopsy suggestive of spondylodiscitis. Our patient cohort consisted of 72 patients, of whom 33.3% (24/72) had spondylodiscitis. The mean age was 63 ± 16 years with a male/female ratio of 41:31. Logistic regression revealed a combination with an area under the curve of 0.72 for pathology and 0.68 for pathology and/or microbiology. Epidural enhancement on MR imaging improved predictive performance to 0.87 for pathology and 0.78 for pathology and/or microbiology. Our findings demonstrate that epidural enhancement on MR imaging added diagnostic value when combined with clinical biomarkers to help predict which patients undergoing percutaneous biopsy will have positive findings for spondylodiscitis on pathology and/or microbiology.

CT differential diagnosis of COVID-19 and non-COVID-19 in symptomatic suspects: a practical scoring method

BackgroundAlthough typical and atypical CT image findings of COVID-19 are reported in current studies, the CT image features of COVID-19 overlap with those of viral pneumonia and other respiratory diseases. Hence, it is difficult to make an exclusive diagnosis.MethodsThirty confirmed cases of COVID-19 and forty-three cases of other aetiology or clinically confirmed non-COVID-19 in a general hospital were included. The clinical data including age, sex, exposure history, laboratory parameters and aetiological diagnosis of all patients were collected. Seven positive signs (posterior part/lower lobe predilection, bilateral involvement, rounded GGO, subpleural bandlike GGO, crazy-paving pattern, peripheral distribution, and GGO +/− consolidation) from significant COVID-19 CT image features and four negative signs (only one lobe involvement, only central distribution, tree-in-bud sign, and bronchial wall thickening) from other non-COVID-19 pneumonia were used. The scoring analysis of CT features was compared between the two groups (COVID-19 and non-COVID-19).ResultsOlder age, symptoms of diarrhoea, exposure history related to Wuhan, and a lower white blood cell and lymphocyte count were significantly suggestive of COVID-19 rather than non-COVID-19 (p < 0.05). The receiver operating characteristic (ROC) curve of the combined CT image features analysis revealed that the area under the curve (AUC) of the scoring system was 0.854. These cut-off values yielded a sensitivity of 56.67% and a specificity of 95.35% for a score > 4, a sensitivity of 100% and a specificity of 23.26% for a score > 0, and a sensitivity of 86.67% and a specificity of 67.44% for a score > 2.ConclusionsWith a simple and practical scoring system based on CT imaging features, we can make a hierarchical diagnosis of COVID-19 and non-COVID-19 with different management suggestions.

Prognostic Implications of Combined Imaging and Histologic Criteria in Squamous Cell Carcinoma with Mandibular Invasion.

Prognosis prediction of squamous cell carcinoma (SCC) with mandibular invasion is controversial, and a more sophisticated staging system to aid prognosis could be developed with imaging characteristics of bone invasion. Imaging-feature analysis provides practical, stratified results for survival prognosis in oral SCC (OSCC) of the mandible, and imaging advances enable more detailed tumor visualization. We retrospectively evaluated significant bone-invasion features associated with poor outcomes in mandibular OSCC to assess the predictive value of staging criteria that combined imaging features and histologic grade (combined imaging–histology (IH) grade) in 65 patients (39 men, 26 women) with mandibular SCC diagnosed from 2006 to 2016. Clinicopathologic features, including T-stage and histologic grade, and prognosis were retrieved. Tumors were classified into three types by extent of mandibular invasion on pretreatment imaging studies. Moreover, we assessed the involvement of the mandibular canal. We examined the correlation of factors associated with locoregional recurrence and overall mortality. The Harrell Concordance Index (C-index) determined prognostic performance of predictors. Nineteen (29%) patients showed locoregional recurrence and 13 (20%) died. For locoregional recurrence and mortality rates, imaging-detected mandibular canal (MC) involvement is a stronger prognostic factor for recurrence (C-index = 0.61 > 0.58) and survival (C-index = 0.58 > 0.63) than histopathologically confirmed perineural invasion, as was the IH grade, especially IH Grade 3, which was significantly associated with worse locoregional recurrence (p < 0.02). Imaging-based staging showed higher prognostic performance than T-staging (C-index = 0.57 (recurrence), 0.60 (death)), when combined with histologic grading (C-index = 0.69 for both) or used alone (C-index = 0.63 (locoregional recurrence), 0.69 (death)). Overall survival was significantly stratified by Imaging type and IH grade. Therefore, analysis of imaging features provided more specific, practical results for survival prognosis in mandibular OSCC. Imaging advances can potentially provide detailed gross views of tumor masses to facilitate development of prognostic criteria for OSCC.

Prediction of Pancreatic Cancer Based on Imaging Features in Patients With Duct Abnormalities.

Abnormalities of the main pancreatic duct may be an early indicator of pancreatic ductal adenocarcinoma (PDAC). We develop and validate algorithms that predict the risk of PDAC using features identified on cross-sectional imaging and other clinical characteristics collected through electronic medical records. Adult patients with abdominal computed tomography or magnetic resonance imaging in January 2006 to June 2016 demonstrating dilatation of main pancreatic duct were identified. Pancreas-related morphologic features were extracted from radiology reports using natural language processing. The cumulative incidence of PDAC with death as a competing risk was estimated using multistate models. Model discrimination was assessed using c-index. The models were internally validated using bootstrapping. The cohort consisted of 7819 patients (mean age, 71 years; 65% female). A total of 781 patients (10%) developed PDAC within 3 years after the first eligible imaging study. The final models achieved reasonable discrimination (c-index, 0.825-0.833). The 3-year average risk of PDAC in the top 5% of the total eligible patients was 56.0%, more than 20 times of the average risk among the bottom 50% of patients. Prediction models combining imaging features and clinical measures can be used to further stratify the risk of pancreatic cancer among patients with pancreas ductal dilatation.