Combination Of Image Features Research Articles

Application and progress of X‐ray, computed tomography, and magnetic resonance imaging radiomics in osteosarcoma

AbstractLooking back on the development of radiomics in osteosarcoma over recent years, in addition to distinguishing osteosarcoma from other malignant bone tumors (mainly Ewing's sarcoma), more research directions are using radiomics to evaluate and predict the efficacy and survival of patients undergoing neoadjuvant chemotherapy. Among the three commonly used examination methods of X‐ray, CT, and magnetic resonance imaging (MRI), more and more studies have been conducted on MRI‐based radiomics, which fully reflects the advantages of MRI's high soft tissue contrast, multi‐sequence imaging, and most of the studies used a combination of imaging features and clinical features to make predictions. Some articles also considered relevant laboratory examination results and more and more studies are performing external verification. The field is now gradually developing toward multidimensional data and multicenter cooperation and data sharing.

Retrieving Surface Deformation of Mining Areas Using ZY-3 Stereo Imagery and DSMs

Measuring surface deformation is crucial for a better understanding of spatial-temporal evolution and the mechanism of mining-induced deformation, thus effectively assessing the mining-related geohazards, such as landslides or damage to surface infrastructures. This study proposes a method of retrieving surface deformation by combining multi-temporal digital surface models (DSMs) with image homonymous features using China’s ZY-3 satellite stereo imagery. DSM is generated from three-line-array images of ZY-3 satellite using a rational function model (RFM) as the imaging geometric model. Then, elevation changes in deformation are extracted using the difference of DSMs acquired at different times, while planar displacements of deformation are calculated using image homonymous features extracted from multi-temporal digital orthographic maps (DOMs). Scale invariant feature transform (SIFT) points and line band descriptor (LBD) lines are selected as two kinds of salient features for image homonymous features generation. Cross profiles are also extracted for deformation in typical regions. Four sets of stereo imagery acquired in 2012 to 2022 are used for deformation extraction and analysis in the Fushun coalfield of China, where surface deformation is quite distinct and coupled with rising and descending elevation together. The results show that 21.60% of the surface in the study area was deformed from 2012 to 2017, while a decline from 2017 to 2022 meant that 17.19% of the surface was deformed with a 95% confidence interval. Moreover, the ratio of descending area was reduced to 6.44% between 2017 and 2022, which is lower than the ratios in other years. The slip deformation area in the west open pit mine is about 1.22 km2 and the displacement on the south slope is large, reaching an average of 26.89 m and sliding from south to north to the bottom of the pit between 2012 and 2017, but elevations are increased by an average of about 16.35 m, involving an area of about 0.86 km2 between 2017 and 2022 due to the restoration of the open pit. The results demonstrate that more quantitative features and specific surface deformation can be retrieved in mining areas by combining image features with DSMs derived from ZY-3 satellite stereo imagery.

Clinical Characteristics of Tracheobronchopathia Osteochondroplastica: A Retrospective Study of 33 Patients.

Tracheobronchopathia osteochondroplastica (TO) is a relatively rare benign tracheobronchial disease, which is often misdiagnosed or missed. To date, there is no specific treatment for TO. The aim of this study was to investigate the clinical manifestations, imaging features, bronchoscopy results, pathological findings, and diagnostic points of TO. A total of 33 patients diagnosed with TO were enrolled. Clinical data were collected using retrospective methods in the period from January 2021 and November 2022. Descriptive analysis was used. Patients included 17 (51.5%) male and 16 (48.5%) female, with a median age of 54 years. The main clinical manifestations included cough in 15 cases, fever in 6 cases, chest tightness in 4 cases, haemoptysis in 3 cases, and chest pain in 4 cases. The time from the onset of symptoms to diagnosis was 1 week to 96 months. Some patients were diagnosed with other lung diseases, including 16 patients with tuberculosis, 2 patients with lung cancer, 3 patients with nontuberculous mycobacteriosis, 3 patients with tuberculous pleurisy, 2 patients with bronchiectasis, and 1 patient with pneumonia. Chest computed tomography (CT) scan demonstrated calcified nodules in 10 (30.3%) patients. In bronchoscopy, entire tracheal involvement was found in 21 (63.6%) patients, 12 (36.4%) patients were found to have involvement of only part of the trachea. The patients were divided into three groups according to the bronchoscopic presentation, the largest proportion was stage II (19/33), followed by stage I (8/33) and stage III (6/33). Histopathological findings showed squamous metaplasia, cartilaginous, and bony tissues. TO is a slowly progressing disease with non-specific clinical symptoms and a low positive rate of imaging diagnosis, making it susceptible to misdiagnosis and missed diagnosis. The disease needs to be diagnosed by combining imaging features, fiberoptic bronchoscopy, and pathological findings.

CNN Approach for Predicting Survival Outcome of Patients With COVID-19

Coronavirus disease 2019 (COVID-19) has been challenged specifically with the new variant. The number of patients seeking treatment has increased significantly, putting tremendous pressure on hospitals and healthcare systems. With the potential of artificial intelligence (AI) to leverage clinicians to improve personalized medicine for COVID-19, we propose a deep learning model based on 1D and 3D convolutional neural networks (CNNs) to predict the survival outcome of COVID-19 patients. Our model consists of two CNN channels that operate with CT scans and the corresponding clinical variables. Specifically, each patient data set consists of CT images and the corresponding 44 clinical variables used in the 3D CNN and 1D CNN input, respectively. This model aims to combine imaging and clinical features to predict short-term from long-term survival. Our models demonstrate higher performance metrics compared to state-of-the-art models with AUC-ROC of 91.44 – 91.60% versus 84.36 – 88.10% and Accuracy of 83.39 – 84.47% versus 79.06 – 81.94% in predicting the survival groups of patients with COVID-19. Based on the findings, the combined clinical and imaging features in the deep CNN model can be used as a prognostic tool and help to distinguish censored and uncensored cases of COVID-19.

Deep learning model for distinguishing Mayo endoscopic subscore 0 and 1 in patients with ulcerative colitis

The aim of this study was to address the issue of differentiating between Mayo endoscopic subscore (MES) 0 and MES 1 using a deep learning model. A dataset of 492 ulcerative colitis (UC) patients who demonstrated MES improvement between January 2018 and December 2019 at Samsung Medical Center was utilized. Specifically, two representative images of the colon and rectum were selected from each patient, resulting in a total of 984 images for analysis. The deep learning model utilized in this study consisted of a convolutional neural network (CNN)-based encoder, with two auxiliary classifiers for the colon and rectum, as well as a final MES classifier that combined image features from both inputs. In the internal test, the model achieved an F1-score of 0.92, surpassing the performance of seven novice classifiers by an average margin of 0.11, and outperforming their consensus by 0.02. The area under the receiver operating characteristic curve (AUROC) was calculated to be 0.97 when considering MES 1 as positive, with an area under the precision-recall curve (AUPRC) of 0.98. In the external test using the Hyperkvasir dataset, the model achieved an F1-score of 0.89, AUROC of 0.86, and AUPRC of 0.97. The results demonstrate that the proposed CNN-based model, which integrates image features from both the colon and rectum, exhibits superior performance in accurately discriminating between MES 0 and MES 1 in patients with UC.

Cervical myelitis: a practical approach to its differential diagnosis on MR imaging.

Differential diagnosis of non-compressive cervical myelopathy encompasses a broad spectrum of inflammatory, infectious, vascular, neoplastic, neurodegenerative, and metabolic etiologies. Although the speed of symptom onset and clinical course seem to be specific for certain neurological diseases, lesion pattern on MR imaging is a key player to confirm diagnostic considerations. The differentiation between acute complete transverse myelitis and acute partial transverse myelitis makes it possible to distinguish between certain entities, with the latter often being the onset of multiple sclerosis. Typical medullary MRI lesion patterns include a) longitudinal extensive transverse myelitis, b) short-range ovoid and peripheral lesions, c) polio-like appearance with involvement of the anterior horns, and d) granulomatous nodular enhancement prototypes. Cerebrospinal fluid analysis, blood culture tests, and autoimmune antibody testing are crucial for the correct interpretation of imaging findings. The combination of neuroradiological features and neurological and laboratory findings including cerebrospinal fluid analysis improves diagnostic accuracy. · The differentiation of medullary lesion patterns, i. e., longitudinal extensive transverse, short ovoid and peripheral, polio-like, and granulomatous nodular, facilitates the diagnosis of myelitis.. · Discrimination of acute complete and acute partial transverse myelitis makes it possible to categorize different entities, with the latter frequently being the overture of multiple sclerosis (MS).. · Neuromyelitis optica spectrum disorders (NMOSD) may start as short transverse myelitis and should not be mistaken for MS.. · The combination of imaging features and neurological and laboratory findings including cerebrospinal fluid analysis improves diagnostic accuracy.. · Additional brain imaging is mandatory in suspected demyelinating, systemic autoimmune, infectious, paraneoplastic, and metabolic diseases..

Hybrid Visual-Ranging Servoing for Positioning Based on Image and Measurement Features.

In this article, a hybrid visual-ranging servoing method is proposed to realize high-precision positioning tasks with a 6-degree of freedom (DOF) manipulator. This method utilizes the image and measurement features directly in the control loop. Without the need of complex image feature design and attitude estimation, this method realizes the 6-DOF control of a robot. A vital challenge in traditional vision-based systems is avoiding local minima and singularity problems. To tackle this issue, a full-rank interaction matrix hybrid visual servo (FRHVS) design criterion is proposed, which guarantees that the hybrid interaction matrix and its pseudoinverse matrix are both full rank. Moreover, the interaction matrix for these hybrid strategies, which combines image features with other sensors features, is derived in an analytical form. Experiments on a 6-DOF manipulator show that the proposed method is effective and has global asymptotic stability and high precision.

Non-invasive prediction of microvascular invasion in patients with hepatocellular carcinoma: is there any added value in combining imaging features and radiomics?

Non-invasive prediction of microvascular invasion in patients with hepatocellular carcinoma: is there any added value in combining imaging features and radiomics?

MetaAttention model: a new approach for skin lesion diagnosis using AB features and attention mechanism

Ozone depletion has always been a hot crisis around the globe. Its consequence is the increase in ultraviolet radiation at the surface in many regions and countries, which then causes danger to the human immune system, eyes, and especially skin - the part that is directly exposed most to the sunlight. According to the World Health Organization, the number of cases of skin cancer is higher than that of breast, prostate, and lung cancer combined. Therefore, there has been a lot of research to apply deep learning models in solving the problem of skin cancer classification. This paper proposes a novel approach, named MetaAttention, aimed at improving the performance of transfer learning models for skin lesion classification. The method combines image features with patients’ metadata features using the attention mechanism, incorporating clinical knowledge related to ABCD signals to better distinguish melanoma cell carcinoma disease, which has long been a significant challenge for researchers. The experimental results indicate that the proposed approach outperforms the state-of-the-art method, EfficientNet-B4, achieving an accuracy of 89.9% with Scale-dot product MetaAttention and 90.63% with Additive MetaAttention. The method has the potential to support dermatologists in diagnosing skin lesions effectively and efficiently. Furthermore, with larger datasets, our method could be further fine-tuned to achieve even better performance on a broader range of labels.

Deep learning radio-clinical signature for predicting neoadjuvant chemotherapy response and prognosis from pretreatment CT images of locally advanced gastric cancer patients.

Early noninvasive screening of patients who would benefit from neoadjuvant chemotherapy (NCT) is essential for personalized treatment of locally advanced gastric cancer (LAGC). The aim of this study was to identify a radioclinical signatures from pretreatment oversampled CT images to predict the response to NCT and prognosis of LAGC patients. LAGC patients were retrospectively recruited from six hospitals from January 2008 to December 2021. An SE-ResNet50-based chemotherapy response prediction system was developed from pretreatment CT images preprocessed with an imaging oversampling method (i.e., DeepSMOTE). Then, the Deep learning (DL) signature and clinic-based features were fed into the deep learning radioclinical signature (DLCS). The predictive performance of the model was evaluated based on discrimination, calibration and clinical usefulness. An additional model was built to predict overall survival (OS) and explore the survival benefit of the proposed DL signature and clinicopathological characteristics. A total of 1060 LAGC patients were recruited from six hospitals, the training cohort (TC) and internal validation cohort (IVC) patients were randomly selected from center I. An external validation cohort (EVC) of 265 patients from 5 other centers was also included. The DLCS exhibited excellent performance in predicting the response to NCT in the IVC (AUC, 0.86) and EVC (AUC, 0.82), with good calibration in all cohorts (P>0.05). Moreover, the DLCS model outperformed the clinical model (P<0.05). Additionally, we found that the DL signature could serve as an independent factor for prognosis (HR, 0.828, P=0.004). The C-index, iAUC, and IBS for the OS model were 0.64, 1.24 and 0.71 in the test set. We proposed a DLCS model that combined imaging features with clinical risk factors to accurately predict tumor response and identify the risk of OS in LAGC patients priors to NCT that can then be used to guide personalized treatment plans with the help of computerized tumor-level characterization.

Abstract CT113: Safety and efficacy of Tumor Treating Fields (TTFields) combined with bevacizumab and systemic chemotherapy in recurrent GBM

Abstract Background: The treatment of recurrent glioblastoma faces many challenges. The development of new therapeutic approaches and the exploration of biomarkers of therapeutic effectiveness will provide a basis for better tumor control and individualized diagnosis and treatment. TTFields therapy, bevacizumab and second-line chemotherapy are the alternative treatment methods for recurrent gliomas. Especially, the introduction of TTFields has prolonged the tumor control of patients. Clinical studies have shown that bevacizumab combined with chemotherapy can improve the prognosis of patients. In this study, the randomized controlled principle was used to compare TTFields plus second-line chemotherapy and combined chemotherapy on the basis of both, and to explore the safety and effectiveness of combined treatment for patients with recurrent glioblastoma. Objectives: 1. To evaluate the safety and efficacy of TTFields, bevacizumab, and second-line chemotherapy in the treatment of relapsed GBM. 2. To evaluate the effects of TTFields, bevacizumab and their combination on the permeability of the blood-brain barrier. 3. To explore the correlation between serum VEGF, plasma MMP-2, MMP-3 and other factors and the efficacy of combination therapy. Methods: A total of 40 patients who met the inclusion criteria were planned to be recruited and randomly assigned to two experimental groups, with 20 patients in each group. On the basis of second-line chemotherapy, patients in each group were treated with TTFields alone, TTFields combined with BEV, respectively. The 6-month survival rate was used as the primary outcome, and the median OS and median PFS were used as the secondary outcomes. The effectiveness and safety of each regimen was evaluated and compared. Krans, CBV, CBF, MTT and other imaging indicators were used to evaluate vascular permeability and blood flow. To compare the differences between BEV alone and BEV combined with TTFields, and to evaluate the efficacy prediction of these hematologic molecules in the combined regimen. Moreover, imaging and hematological indicators were combined to construct a prognostic model that could predict the efficacy of the combined regimen. Key Issues to be Resolved: 1. To evaluate the comprehensive effect of TTFields combined with BEV treatment on blood-brain barrier permeability and blood perfusion. 2. A prognostic prediction model will be established by combining imaging features and hematological indicators. 3. Preliminarily determine the direction of optimization of the combined treatment program. Citation Format: Xun Kang, Zehao Cai, Feng Chen, Botao Zhang, Bo Jiang, Wenbin Li. Safety and efficacy of Tumor Treating Fields (TTFields) combined with bevacizumab and systemic chemotherapy in recurrent GBM [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 2 (Clinical Trials and Late-Breaking Research); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(8_Suppl):Abstract nr CT113.

OpenCorr: An open source library for research and development of digital image correlation

OpenCorr is an open source C++ library made for research and development of digital image correlation (DIC) technology. It provides full-function modules of 2D and stereo DIC, as well as digital volume correlation (DVC). In contrast with most of current open source DIC software which is application-oriented, the sophisticated design of OpenCorr makes it developer-friendly and facilitates the method study. Users can create DIC software quickly for specific applications by assembling the modules in this library and realize new algorithms readily by modifying the modules or incorporating additional modules. Three examples are given as the showcases for the capability of OpenCorr. The first one is an image feature guided DIC method, which can accurately measure large strain over 35%. The second is a stereo DIC method combining image feature guided DIC with an epipolar constraint aided algorithm to achieve enhanced robustness. The third is a novel self-adaptive DIC method, which dynamically optimize the size and shape of subset at each point of interest (POI) according to the nearby image features. The computation speed of DIC and DVC modules is evaluated through a few tests, which demonstrate superior running efficiency of the library. OpenCorr has attracted rapidly growing interest from both academic and industrial communities. It is expected to stimulate the advance and spread of DIC technology.

Prediction of New-Onset Diabetes after Pancreatectomy with Subspace Clustering based Multi-view Feature Selection.

The pancreas plays an important role in glucose metabolism, and developing diabetes or long-term glucose metabolism disturbance may be a prevalent sequela after pancreatectomy. Nevertheless, relative factors of new-onset diabetes after pancreatectomy stay unclear. Radiomics analysis is potential to identify image markers for disease prediction or prognosis. Meanwhile, combination of imaging and electronic medical record (EMR) showed superior performance than imaging or EMR alone in previous studies. One critical step is to identity predictors from high-dimensional features, and it is even more challenging to select and fuse imaging and EMR features. In this work, we develop a radiomics pipeline to assess postoperative new-onset diabetes risk of patients undergoing distal pancreatectomy. Specifically, we extract multiscale image features with 3D wavelet transformation, and include patients' characteristics, body composition and pancreas volume information as clinical features. Then, we propose a multi-view subspace clustering guided feature selection method (MSCUFS) for the selection and fusion of image and clinical features. Finally, a prediction model is constructed with classical machine learning classifier. Experimental results on an established distal pancreatectomy cohort showed that the SVM model with combined imaging and EMR features demonstrated good discrimination, with an AUC value of 0.824, which improved the model with image features alone by 0.037 AUC. Compared with state-of-the-art feature selection methods, the proposed MSCUFS has superior performance in fusing image and clinical features.

A survival prediction model via interpretable machine learning for patients with oropharyngeal cancer following radiotherapy.

To explore interpretable machine learning (ML) methods, with the hope of adding more prognosis value, for predicting survival for patients with Oropharyngeal-Cancer (OPC). A cohort of 427 OPC patients (Training 341, Test 86) from TCIA databasewas analyzed. Radiomic features of gross-tumor-volume (GTV) extracted from planning CT using Pyradiomics, and HPV p16 status, etc. patient characteristics were considered as potential predictors. A multi-level dimension reduction algorithm consisting of Least-Absolute-Selection-Operator (Lasso) and Sequential-Floating-Backward-Selection (SFBS) was proposed to effectively remove redundant/irrelevant features. The interpretable model was constructed by quantifying the contribution of each feature to the Extreme-Gradient-Boosting (XGBoost) decision by Shapley-Additive-exPlanations (SHAP) algorithm. The Lasso-SFBS algorithm proposed in this study finally selected 14 features, and our prediction model achieved an area-under-ROC-curve (AUC) of 0.85 on the test dataset based on this feature set. The ranking of the contribution values calculated by SHAP shows that the top predictors that were most correlated with survival were ECOG performance status, wavelet-LLH_firstorder_Mean, chemotherapy, wavelet-LHL_glcm_InverseVariance, tumor size. Those patients who had chemotherapy, with positive HPV p16 status, and lower ECOG performance status, tended to have higher SHAP scores and longer survival; who had an older age at diagnosis, heavy drinking and smoking pack year history, tended to lower SHAP scores and shorter survival. We demonstrated predictive values of combined patient characteristics and imaging features for the overall survival of OPC patients. The multi-level dimension reduction algorithm can reliably identify the most plausible predictors that are mostly associated with overall survival. The interpretable patient-specific survival prediction model, capturing correlations of each predictor and clinical outcome, was developed to facilitate clinical decision-making for personalized treatment.

Prognostic value of diffuse reduction of spleen density on postoperative survival of pancreatic ductal adenocarcinoma: A retrospective study.

It is difficult to predict the prognosis of patients with pancreatic ductal adenocarcinoma (PDAC) before radical operation. The purpose of this study was to explore the connection between the diffuse reduction of spleen density on computed tomography (DROSD) and the postoperative prognosis of patients with PDAC. A total of 160 patients with PDAC who underwent radical surgery in the First Affiliated Hospital of Wenzhou Medical University were enrolled. Cox regression analysis was used to cast the overall survival (OS) and evaluate the prognostic factors. Nomogram was used to forecast the possibility of 1-year, 3-year, and 5-year OS. The prediction accuracy and clinical net benefit are performed by concordance index (C-index), calibration curve, time-dependent receiver operating characteristics (tdROC), and decision curve analysis. In multivariable Cox analysis, DROSD is independently related to OS. Advanced age, TNM stage, neutrophil/lymphocyte ratio, and severe complications were also independent prognostic factors. The calibration curves of nomogram showed optimal agreement between prediction and observation. The C-index of nomogram is 0.662 (95%CI, 0.606-0.754). The area under tdROC curve for a 3-year OS of nomogram is 0.770. DROSD is an independent risk factor for an OS of PDAC. We developed a nomogram that combined imaging features, clinicopathological factors, and systemic inflammatory response to provide a personalized risk assessment for patients with PDAC.

MRI Prediction Model for Tenosynovial Giant Cell Tumor with Risk of Diffuse-type

To propose a magnetic resonance imaging (MRI) prediction model for diffuse-type tenosynovial giant cell tumors (D-TSGCTs). Anatomic locations were classified and then nodularity, margin, peripheral and internal hypointensity, and bone and cartilage involvement were evaluated on MRI. Student's t-test, chi-square test, diagnostic performance, logistic regression analysis, and decision tree were performed. Nineteen intra-articular (11 localized; eight diffuse) and 55 extra-articular (44 localized; 11 diffuse) TSGCTs were included. Extra-articular D-TSGCTs showed significantly more frequent multinodular (72.7% vs. 25.0%, p=0.009), and infiltrative lesions (90.9% vs. 34.1%, p=0.002), without peripheral hypointensity (90.9% vs. 18.2%, p < 0.001), and contained granular internal hypointensity (72.7% vs. 31.8%; p=0.003) with more frequent bone (81.8% vs. 27.3%; p=0.003) and cartilage (50.0% vs. 0.0%; p=0.038) involvement than localized-type. Intra-articular D-TSGCT also showed significance in all MRI features (100.0% vs. 9.1%, p=0.001; 100.0% vs. 27.3%, p=0.007; 100.0% vs. 36.4%, p=0.018; 100.0% vs. 27.3%, p=0.007; 50.0% vs. 0.0%, p=0.038), except bone involvement (37.5% vs. 9.1%, p=0.352) than localized-type. Cartilage involvement revealed the highest specificity (88.6-100.0%), regardless of location. Nodularity (100.0%; odds-ratio [OR]: 70.000) and peripheral hypointensity (90.9%; OR: 62.250) demonstrated the highest sensitivities ORs for D-TSGCT in intra-articular and extra-articular cases, respectively. MRI models for D-TSGCG beginning with the cartilage involvement in both anatomic locations and next on nodularity and peripheral hypointensity in intra-articular and extra-articular locations, respectively, exhibited sensitivity and specificity of 100% and 90.9% for intra-articular and 100% and 77.2% for extra-articular TSGCTs, respectively. MRI can suggest the risk of D-TSGCT by combining imaging features with anatomic locations.

P254 Deep Learning Model for Distinguishing MES 0 and MES 1 in Patients with Ulcerative Colitis

Abstract Background Endoscopic remission has recently been defined as a Mayo endoscopic sub-score (MES) of 0. Therefore, patients with MES 1 need to step up to achieve an MES of 0 on the endoscopic score. In discriminating between MES 0 and 1, the inter-observer variation was very severe among the endoscopists. This study aimed to narrow the gap in distinguishing between MES 0 and MES1 using a deep learning model. Methods From the endoscopic images of 492 ulcerative colitis (UC) patients with MES improvement (MES 0 or MES 1) from January 2018 to December 2019 at this center, two representative images of the colon and rectum were selected, and a total of 984 images were analyzed. Our model was composed of a convolutional neural network (CNN)-based encoder, two auxiliary classifiers for the colon and rectum, and a final MES classifier for the combined image features of the two inputs. The following three experiments were conducted to validate and test the proposed model: 12-fold cross-validation for representative model selection, performance comparison with a novice group through an internal test dataset, and an external test with Hyperkvasir, a public gastrointestinal endoscopic image dataset. Results In the internal test, our model showed an F1-score of 0.92. This was, on average, 0.11 higher than each of the seven novices and 0.02 higher than their consensus. Considering MES 1 as positive, the area under the receiver operating characteristic curve (AUROC) was 0.97, and the area under the precision-recall curve (AUPRC) was 0.98. In the external test with Hyperkvasir, our model showed an F1-score of 0.89, AUROC of 0.86, and AUPRC of 0.97. Conclusion We found that the proposed CNN-based model that integrates the image features of the colon and rectum has superior performance in discriminating between MES 0 and MES 1 in patients with UC. Further prospective studies are required to prove the clinical utility of our model in clinical practice.

CT and MR imaging features of mixed neuroendocrine–non-neuroendocrine neoplasm of the pancreas compared with pancreatic ductal adenocarcinoma and neuroendocrine tumor

ObjectiveThis study aimed to assess the computed tomography (CT) and magnetic resonance imaging (MRI) features of pancreatic mixed neuroendocrine–non-neuroendocrine neoplasm (MiNEN) and compare them with those of pancreatic ductal adenocarcinoma (PDAC) and neuroendocrine tumor (NET).MethodsTwelve patients with pancreatic MiNEN, 24 patients with PDAC, and 24 patients with NET, who underwent both contrast-enhanced CT and MRI, were included. Clinical data and the key imaging features were retrospectively evaluated by two independent readers and compared between MiNEN and PDAC or NET. Univariate and multivariable logistic regression analyses were performed to obtain predictors for pancreatic MiNEN.ResultsPatients with pancreatic MiNEN more frequently presented with large size and heterogeneous and cystic components compared with PDAC (p < 0.031) and ill-defined irregular margins, progressive enhancement, and adjacent organ involvement compared with NET (p < 0.036). However, vascular invasion was less commonly seen in MiNEN than PDAC (p = 0.010). Moderate enhancement was observed more frequently in MiNEN than in PDAC or NET (p < 0.001). Multivariate logistic analyses demonstrated that moderate enhancement and ill-defined irregular margin were the most valuable features for the prediction of pancreatic MiNEN (p ≤ 0.044). The combination of the two features resulted in a specificity of 93.8%, sensitivity of 83.3%, and accuracy of 91.7%.ConclusionsWe have mainly described the radiological findings of pancreatic MiNEN with ill-defined irregular margin and moderate enhancement compared with PDAC and NET. The combination of imaging features could improve diagnostic efficiency and help in the selection of the correct treatment method.

Image Dynamics-Based Visual Servo Control for Unmanned Aerial Manipulatorl With a Virtual Camera

This article proposes a visual servo control method based on image dynamics for an unmanned aerial manipulator (UAM) combining an unmanned aerial vehicle (UAV) with a multidegree-of-freedom onboard manipulator. First, taking into account the dynamic characteristics of the coupled UAM, the dynamic model of the coupled system is derived. Then, based on the perspective projection model, the appropriate image features are defined on the virtual image plane to obtain decoupled image feature dynamics. Combining image features dynamics and UAM dynamics, the position and attitude controller of the UAM flying platform with the center of gravity compensation is derived. Furthermore, the image feature motion is further applied to the positioning control of the end-effector of the onboard manipulator. In addition, the UAM motion distribution strategy is designed to coordinate the movement between the UAM flying platform and the onboard manipulator. The simulation results illustrate the performance of the proposed visual servo control based on image dynamics. As a result, the trajectories of the feature points are smooth in the real image plane, which means the UAM movement is stable. Finally, the experimental results in an outdoor environment verify the practicability and effectiveness of the proposed visual servo control method for autonomous operations. The proposed control method manages the UAM from an unknown initial position to the desired position, which can be applied to various types of aerial operations missions such as object transmission.

Artificial Intelligence in Lung Cancer Imaging: Unfolding the Future.

Lung cancer is one of the malignancies with higher morbidity and mortality. Imaging plays an essential role in each phase of lung cancer management, from detection to assessment of response to treatment. The development of imaging-based artificial intelligence (AI) models has the potential to play a key role in early detection and customized treatment planning. Computer-aided detection of lung nodules in screening programs has revolutionized the early detection of the disease. Moreover, the possibility to use AI approaches to identify patients at risk of developing lung cancer during their life can help a more targeted screening program. The combination of imaging features and clinical and laboratory data through AI models is giving promising results in the prediction of patients' outcomes, response to specific therapies, and risk for toxic reaction development. In this review, we provide an overview of the main imaging AI-based tools in lung cancer imaging, including automated lesion detection, characterization, segmentation, prediction of outcome, and treatment response to provide radiologists and clinicians with the foundation for these applications in a clinical scenario.