Deep learning combined with radiomics for the classification of enlarged cervical lymph nodes

Using T2-weighted magnetic resonance imaging-derived radiomics to classify cervical lymphadenopathy in children

BACKGROUND

Cervical lymphadenopathy is common in children and has diverse causes varying from benign to malignant, their similar manifestations making differential diagnosis difficult.

OBJECTIVE

This study aimed to investigate whether radiomic models using conventional magnetic resonance imaging (MRI) could classify pediatric cervical lymphadenopathy.

METHODS

A total of 419 cervical lymph nodes from 146 patients, and encompassing four common etiologies (Kikuchi disease, reactive hyperplasia, suppurative lymphadenitis and malignancy), were randomly divided into training and testing sets in a ratio of 7:3. For each lymph node, 1,218 features were extracted from T2-weighted images. Then, the least absolute shrinkage and selection operator (LASSO) models were used to select the most relevant ones. Two models were built using a support vector machine classifier, one was to classify benign and malignant lymph nodes and the other further distinguished four different diseases. The performance was assessed by receiver operating characteristic curves and decision curve analysis.

RESULTS

By LASSO, 20 features were selected to construct a model to distinguish benign and malignant lymph nodes, which achieved an area under the curve (AUC) of 0.89 and 0.80 in the training and testing sets, respectively. Sixteen features were selected to construct a model to distinguish four different cervical lymphadenopathies. For each etiology, Kikuchi disease, reactive hyperplasia, suppurative lymphadenitis, and malignancy, an AUC of 0.97, 0.91, 0.88, and 0.87 was achieved in the training set, and an AUC of 0.96, 0.80, 0.82, and 0.82 was achieved in the testing set, respectively.

CONCLUSION

MRI-derived radiomic analysis provides a promising non-invasive approach for distinguishing causes of cervical lymphadenopathy in children.

Diagnosis of cervical lymphoma using a YOLO-v7-based model with transfer learning

To investigate the ability of an auxiliary diagnostic model based on the YOLO-v7-based model in the classification of cervical lymphadenopathy images and compare its performance against qualitative visual evaluation by experienced radiologists. Three types of lymph nodes were sampled randomly but not uniformly. The dataset was randomly divided into for training, validation, and testing. The model was constructed with PyTorch. It was trained and weighting parameters were tuned on the validation set. Diagnostic performance was compared with that of the radiologists on the testing set. The mAP of the model was 96.4% at the 50% intersection-over-union threshold. The accuracy values of it were 0.962 for benign lymph nodes, 0.982 for lymphomas, and 0.960 for metastatic lymph nodes. The precision values of it were 0.928 for benign lymph nodes, 0.975 for lymphomas, and 0.927 for metastatic lymph nodes. The accuracy values of radiologists were 0.659 for benign lymph nodes, 0.836 for lymphomas, and 0.580 for metastatic lymph nodes. The precision values of radiologists were 0.478 for benign lymph nodes, 0.329 for lymphomas, and 0.596 for metastatic lymph nodes. The model effectively classifies lymphadenopathies from ultrasound images and outperforms qualitative visual evaluation by experienced radiologists in differential diagnosis.

Magnetic resonance imaging and deoxyribonucleic acid methylation-based radiogenomic models for survival risk stratification of glioblastoma

Glioblastoma multiforme (GBM) is one of the deadliest tumours. This study aimed to construct radiogenomic prognostic models of glioblastoma overall survival (OS) based on magnetic resonance imaging (MRI) Gd-T1WI images and deoxyribonucleic acid (DNA) methylation-seq and to understand the related biological pathways. The ResNet3D-18 model was used to extract radiomic features, and Lasso-Cox regression analysis was utilized to establish the prognostic models. A nomogram was constructed by combining the radiogenomic features and clinicopathological variables. The DeLong test was performed to compare the area under the curve (AUC) of the models. We screened differentially expressed genes (DEGs) with original ribonucleic acid (RNA)-seq in risk stratification and used Gene Ontology (GO) and Kyoto Encyclopaedia of Genes and Genomes (KEGG) annotations for functional enrichment analysis. For the 1-year OS models, the AUCs of the radiogenomic set, methylation set and deep learning set in the training cohort were 0.864, 0.804 and 0.787, and those in the validation cohort were 0.835, 0.768 and 0.651, respectively. The AUCs of the 0.5-, 1- and 2-year nomograms in the training cohort were 0.943, 0.861 and 0.871, and those in the validation cohort were 0.864, 0.885 and 0.805, respectively. A total of 245 DEGs were screened; functional enrichment analysis showed that these DEGs were associated with cell immunity. The survival risk-stratifying radiogenomic models for glioblastoma OS had high predictability and were associated with biological pathways related to cell immunity.

Can deep learning replace histopathological examinations in the differential diagnosis of cervical lymphadenopathy?

INTRODUCTION

We aimed to develop a diagnostic deep learning model using contrast-enhanced CT images and to investigate whether cervical lymphadenopathies can be diagnosed with these deep learning methods without radiologist interpretations and histopathological examinations.

MATERIAL METHOD

A total of 400 patients who underwent surgery for lymphadenopathy in the neck between 2010 and 2022 were retrospectively analyzed. They were examined in four groups of 100 patients: the granulomatous diseases group, the lymphoma group, the squamous cell tumor group, and the reactive hyperplasia group. The diagnoses of the patients were confirmed histopathologically. Two CT images from all the patients in each group were used in the study. The CT images were classified using ResNet50, NASNetMobile, and DenseNet121 architecture input.

RESULTS

The classification accuracies obtained with ResNet50, DenseNet121, and NASNetMobile were 92.5%, 90.62, and 87.5, respectively.

CONCLUSION

Deep learning is a useful diagnostic tool in diagnosing cervical lymphadenopathy. In the near future, many diseases could be diagnosed with deep learning models without radiologist interpretations and invasive examinations such as histopathological examinations. However, further studies with much larger case series are needed to develop accurate deep-learning models.

Comparing three-dimensional and two-dimensional deep-learning, radiomics, and fusion models for predicting occult lymph node metastasis in laryngeal squamous cell carcinoma based on CT imaging: a multicentre, retrospective, diagnostic study

Background

The occult lymph node metastasis (LNM) of laryngeal squamous cell carcinoma (LSCC) affects the treatment and prognosis of patients. This study aimed to comprehensively compare the performance of the three-dimensional and two-dimensional deep learning models, radiomics model, and the fusion models for predicting occult LNM in LSCC.

Methods

In this retrospective diagnostic study, a total of 553 patients with clinical N0 stage LSCC, who underwent surgical treatment without distant metastasis and multiple primary cancers, were consecutively enrolled from four Chinese medical centres between January 01, 2016 and December 30, 2020. The participant data were manually retrieved from medical records, imaging databases, and pathology reports. The study cohort was divided into a training set (n = 300), an internal test set (n = 89), and two external test sets (n = 120 and 44, respectively). The three-dimensional deep learning (3D DL), two-dimensional deep learning (2D DL), and radiomics model were developed using CT images of the primary tumor. The clinical model was constructed based on clinical and radiological features. Two fusion strategies were utilized to develop the fusion model: the feature-based DLRad_FB model and the decision-based DLRad_DB model. The discriminative ability and correlation of 3D DL, 2D DL and radiomics features were analysed comprehensively. The performances of the predictive models were evaluated based on the pathological diagnosis.

Findings

The 3D DL features had superior discriminative ability and lower internal redundancy compared to 2D DL and radiomics features. The DLRad_DB model achieved the highest AUC (0.89-0.90) among all the study sets, significantly outperforming the clinical model (AUC = 0.73-0.78, P = 0.0001-0.042, Delong test). Compared to the DLRad_DB model, the AUC values for the DLRad_FB, 3D DL, 2D DL, and radiomics models were 0.82-0.84 (P = 0.025-0.46), 0.86-0.89 (P = 0.75-0.97), 0.83-0.86 (P = 0.029-0.66), and 0.79-0.82 (P = 0.0072-0.10), respectively in the study sets. Additionally, the DLRad_DB model exhibited the best sensitivity (82-88%) and specificity (79-85%) in the test sets.

Interpretation

The decision-based fusion model DLRad_DB, which combines 3D DL, 2D DL, radiomics, and clinical data, can be utilized to predict occult LNM in LSCC. This has the potential to minimize unnecessary lymph node dissection and prophylactic radiotherapy in patients with cN0 disease.

Funding

National Natural Science Foundation of China, Natural Science Foundation of Shandong Province.

A new prognostic nomogram in patients with mucosa-associated lymphoid tissue lymphoma: a multicenter retrospective study

Background

The present study sought to understand how clinical factors and inflammatory biomarkers affected the prognosis of mucosa-associated lymphoid tissue (MALT) lymphoma and develop a predictive nomogram to assist in clinical practice.

Methods

We conducted a retrospective study on 183 cases of newly diagnosed MALT lymphoma from January 2011 to October 2021, randomly divided into two groups: a training cohort (75%); and a validation cohort (25%). The least absolute shrinkage and selection operator (LASSO) regression analysis was combined with multivariate Cox regression analysis to construct a nomogram for predicting the progression-free survival (PFS) in patients with MALT lymphoma. To evaluate the accuracy of the nomogram model, the area under the receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA) were used.

Results

The PFS was significantly associated with the Ann Arbor Stage, targeted therapy, radiotherapy, and platelet-to-lymphocyte ratio (PLR) in MALT lymphoma. These four variables were combined to establish a nomogram to predict the PFS rates at three and five years. Importantly, our nomogram yielded good predictive value with area under the ROC curve (AUC) values of 0.841 and 0.763 in the training cohort and 0.860 and 0.879 in the validation cohort for the 3-year and 5-year PFS, respectively. Furthermore, the 3-year and 5-year PFS calibration curves revealed a high degree of consistency between the prediction and the actual probability of relapse. Additionally, DCA demonstrated the net clinical benefit of this nomogram and its ability to identify high-risk patients accurately.

Conclusion

The new nomogram model could accurately predict the prognosis of MALT lymphoma patients and assist clinicians in designing individualized treatments.

Clinical features combined with ultrasound-based radiomics nomogram for discrimination between benign and malignant lesions in ultrasound suspected supraclavicular lymphadenectasis

Background

Conventional ultrasound (CUS) is the first choice for discrimination benign and malignant lymphadenectasis in supraclavicular lymph nodes (SCLNs), which is important for the further treatment. Radiomics provide more comprehensive and richer information than radiographic images, which are imperceptible to human eyes.

Objective

This study aimed to explore the clinical value of CUS-based radiomics analysis in preoperative differentiation of malignant from benign lymphadenectasis in CUS suspected SCLNs.

Methods

The characteristics of CUS images of 189 SCLNs were retrospectively analyzed, including 139 pathologically confirmed benign SCLNs and 50 malignant SCLNs. The data were randomly divided (7:3) into a training set (n=131) and a validation set (n=58). A total of 744 radiomics features were extracted from CUS images, radiomics score (Rad-score) built were using least absolute shrinkage and selection operator (LASSO) logistic regression. Rad-score model, CUS model, radiomics-CUS (Rad-score + CUS) model, clinic-radiomics (Clin + Rad-score) model, and combined CUS-clinic-radiomics (Clin + CUS + Rad-score) model were built using logistic regression. Diagnostic accuracy was assessed by receiver operating characteristic (ROC) curve analysis.

Results

A total of 20 radiomics features were selected from 744 radiomics features and calculated to construct Rad-score. The AUCs of Rad-score model, CUS model, Clin + Rad-score model, Rad-score + CUS model, and Clin + CUS + Rad-score model were 0.80, 0.72, 0.85, 0.83, 0.86 in the training set and 0.77, 0.80, 0.82, 0.81, 0.85 in the validation set. There was no statistical significance among the AUC of all models in the training and validation set. The calibration curve also indicated the good predictive performance of the proposed nomogram.

Conclusions

The Rad-score model, derived from supraclavicular ultrasound images, showed good predictive effect in differentiating benign from malignant lesions in patients with suspected supraclavicular lymphadenectasis.

A multidomain fusion model of radiomics and deep learning to discriminate between PDAC and AIP based on 18F-FDG PET/CT images

PURPOSE

To explore a multidomain fusion model of radiomics and deep learning features based on ¹⁸F-fluorodeoxyglucose positron emission tomography/computed tomography (¹⁸F-FDG PET/CT) images to distinguish pancreatic ductal adenocarcinoma (PDAC) and autoimmune pancreatitis (AIP), which could effectively improve the accuracy of diseases diagnosis.

MATERIALS AND METHODS

This retrospective study included 48 patients with AIP (mean age, 65 ± 12.0 years; range, 37-90 years) and 64 patients with PDAC patients (mean age, 66 ± 11.3 years; range, 32-88 years). Three different methods were discussed to identify PDAC and AIP based on ¹⁸F-FDG PET/CT images, including the radiomics model (RAD_model), the deep learning model (DL_model), and the multidomain fusion model (MF_model). We also compared the classification results of PET/CT, PET, and CT images in these three models. In addition, we explored the attributes of deep learning abstract features by analyzing the correlation between radiomics and deep learning features. Five-fold cross-validation was used to calculate receiver operating characteristic (ROC), area under the roc curve (AUC), accuracy (Acc), sensitivity (Sen), and specificity (Spe) to quantitatively evaluate the performance of different classification models.

RESULTS

The experimental results showed that the multidomain fusion model had the best comprehensive performance compared with radiomics and deep learning models, and the AUC, accuracy, sensitivity, specificity were 96.4% (95% CI 95.4-97.3%), 90.1% (95% CI 88.7-91.5%), 87.5% (95% CI 84.3-90.6%), and 93.0% (95% CI 90.3-95.6%), respectively. And our study proved that the multimodal features of PET/CT were superior to using either PET or CT features alone. First-order features of radiomics provided valuable complementary information for the deep learning model.

CONCLUSION

The preliminary results of this paper demonstrated that our proposed multidomain fusion model fully exploits the value of radiomics and deep learning features based on ¹⁸F-FDG PET/CT images, which provided competitive accuracy for the discrimination of PDAC and AIP.

One 3D VOI-based deep learning radiomics strategy, clinical model and radiologists for predicting lymph node metastases in pancreatic ductal adenocarcinoma based on multiphasic contrast-enhanced computer tomography

Purpose

We designed to construct one 3D VOI-based deep learning radiomics strategy for identifying lymph node metastases (LNM) in pancreatic ductal adenocarcinoma on the basis of multiphasic contrast-enhanced computer tomography and to assist clinical decision-making.

Methods

This retrospective research enrolled 139 PDAC patients undergoing pre-operative arterial phase and venous phase scanning examination between 2015 and 2021. A primary group (training group and validation group) and an independent test group were divided. The DLR strategy included three sections. (1) Residual network three dimensional-18 (Resnet 3D-18) architecture was constructed for deep learning feature extraction. (2) Least absolute shrinkage and selection operator model was used for feature selection. (3) Fully connected network served as the classifier. The DLR strategy was applied for constructing different 3D CNN models using 5-fold cross-validation. Radiomics scores (Rad score) were calculated for distinguishing the statistical difference between negative and positive lymph nodes. A clinical model was constructed by combining significantly different clinical variables using univariate and multivariable logistic regression. The manifestation of two radiologists was detected for comparing with computer-developed models. Receiver operating characteristic curves, the area under the curve, accuracy, precision, recall, and F1 score were used for evaluating model performance.

Results

A total of 45, 49, and 59 deep learning features were selected via LASSO model. No matter in which 3D CNN model, Rad score demonstrated the deep learning features were significantly different between non-LNM and LNM groups. The AP+VP DLR model yielded the best performance in predicting status of lymph node in PDAC with an AUC of 0.995 (95% CI:0.989-1.000) in training group; an AUC of 0.940 (95% CI:0.910-0.971) in validation group; and an AUC of 0.949 (95% CI:0.914-0.984) in test group. The clinical model enrolled the histological grade, CA19-9 level and CT-reported tumor size. The AP+VP DLR model outperformed AP DLR model, VP DLR model, clinical model, and two radiologists.

Conclusions

The AP+VP DLR model based on Resnet 3D-18 demonstrated excellent ability for identifying LNM in PDAC, which could act as a non-invasive and accurate guide for clinical therapeutic strategies. This 3D CNN model combined with 3D tumor segmentation technology is labor-saving, promising, and effective.