Semantic contrast with uncertainty-aware pseudo label for lumbar semi-supervised classification.

BACKGROUND: Lumbar disc herniation (LDH) is a prevalent spinal disease that can result in severe pain, with Magnetic resonance imaging (MRI) serving as a commonly diagnostic tool. However, annotating numerous MRI images, necessary for deep learning based LDH diagnosis, can be challenging and labor-intensive. Semi-supervised learning, mainly utilizing pseudo labeling and consistency regularization, can leverage limited labeled images and abundant unlabeled images. However, consistency regularization solely focuses on maintaining the semantic consistency of transformed unlabeled data but fails to utilize the semantic information from labeled data to guide the unlabeled data, and additionally, pseudo labeling is prone to confirmation bias. METHOD: We propose SeCoFixMatch, an innovative approach that seamlessly integrates semantic contrast and uncertainty-aware pseudo labeling into semi-supervised learning. Semantic contrast constraints the semantic consistency between labeled and unlabeled images. Pseudo labels are generated by combining predictive confidence and uncertainty, with uncertainty computing by optimizing the Kullback-Leibler (KL) loss between predictive and target Dirichlet distribution. RESULTS: Comparison with other semi-supervised models and ablation experiment with varying labeled data demonstrate the effectiveness and generalization of proposed model. Notably, SeCoFixMatch, trained with just 40 labels, outperforms the baseline model trained with 200 labels, reducing the annotation effort by a remarkable 80%. CONCLUSIONS: Proposed pseudo labeling algorithm generates more precise pseudo labels for semantic contrastive learning and semantic contrastive learning facilitates better feature representation, thereby further improving the prediction accuracy of pseudo label. The mutual reinforcement of pseudo labeling and semantic contrast constraints boosts the performance of semi-supervised algorithm.

Deep learning-based detection and classification of lumbar disc herniation on magnetic resonance images.

Background: The severity assessment of lumbar disc herniation (LDH) on MR images is crucial for selecting suitable surgical candidates. However, the interpretation of MR images is time-consuming and requires repetitive work. This study aims to develop and evaluate a deep learning-based diagnostic model for automated LDH detection and classification on lumbar axial T2-weighted MR images. Methods: A total of 1115 patients were analyzed in this retrospective study; both a development dataset (1015 patients, 15 249 images) and an external test dataset (100 patients, 1273 images) were utilized. According to the Michigan State University (MSU) classification criterion, experts labeled all images with consensus, and the final labeled results were regarded as the reference standard. The automated diagnostic model comprised Faster R-CNN and ResNeXt101 as the detection and classification network, respectively. The deep learning-based diagnostic performance was evaluated by calculating mean intersection over union (IoU), accuracy, precision, sensitivity, specificity, F1 score, the area under the receiver operating characteristics curve (AUC), and intraclass correlation coefficient (ICC) with 95% confidence intervals (CIs). Results: High detection consistency was obtained in the internal test dataset (mean IoU = 0.82, precision = 98.4%, sensitivity = 99.4%) and external test dataset (mean IoU = 0.70, precision = 96.3%, sensitivity = 97.8%). Overall accuracy for LDH classification was 87.70% (95% CI: 86.59%-88.86%) and 74.23% (95% CI: 71.83%-76.75%) in the internal and external test datasets, respectively. For internal testing, the proposed model achieved a high agreement in classification (ICC = 0.87, 95% CI: 0.86-0.88, P < 0.001), which was higher than that of external testing (ICC = 0.79, 95% CI: 0.76-0.81, P < 0.001). The AUC for model classification was 0.965 (95% CI: 0.962-0.968) and 0.916 (95% CI: 0.908-0.925) in the internal and external test datasets, respectively. Conclusions: The automated diagnostic model achieved high performance in detecting and classifying LDH and exhibited considerable consistency with experts' classification.

Multi-view features integrated 2D\3D Net for glomerulopathy histologic types classification using ultrasound images.

BACKGROUND AND OBJECTIVE: Early diagnoses and rational therapeutics of glomerulopathy can control progression and improve prognosis. The gold standard for the diagnosis of glomerulopathy is pathology by renal biopsy, which is invasive and has many contraindications. We aim to use renal ultrasonography for histologic classification of glomerulopathy. METHODS: Ultrasonography can present multi-view sections of kidney, thus we proposed a multi-view and cross-domain integration strategy (CD-ConcatNet) to obtain more effective features and improve diagnosis accuracy. We creatively apply 2D group convolution and 3D convolution to process multiple 2D ultrasound images and extract multi-view features of renal ultrasound images. Cross-domain concatenation in each spatial resolution of feature maps is applied for more informative feature learning. RESULTS: A total of 76 adult patients were collected and divided into training dataset (56 cases with 515 images) and validation dataset (20 cases with 180 images). We obtained the best mean accuracy of 0.83 and AUC of 0.8667 in the validation dataset. CONCLUSION: Comparison experiments demonstrate that our designed CD-ConcatNet achieves the best classification performance and has great superiority on histologic types diagnosis. Results also prove that the integration of multi-view ultrasound images is beneficial for histologic classification and ultrasound images can indeed provide discriminating information for histologic diagnosis.

Preliminary study on the application of renal ultrasonography radiomics in the classification of glomerulopathy.

BACKGROUND: The aim of this study was to investigate the potential use of renal ultrasonography radiomics features in the histologic classification of glomerulopathy. METHODS: A total of 623 renal ultrasound images from 46 membranous nephropathy (MN) and 22 IgA nephropathy patients were collected. The cases and images were divided into a training group (51 cases with 470 images) and a test group (17 cases with 153 images). A total of 180 dimensional features were designed and extracted from the renal parenchyma in the ultrasound images. Least absolute shrinkage and selection operator (LASSO) logistic regression was then applied to these normalized radiomics features to select the features with the highest correlations. Four machine learning classifiers, including logistic regression, a support vector machine (SVM), a random forest, and a K-nearest neighbour classifier, were deployed for the classification of MN and IgA nephropathy. Subsequently, the results were assessed according to accuracy and receiver operating characteristic (ROC) curves. RESULTS: Patients with MN were older than patients with IgA nephropathy. MN primarily manifested in patients as nephrotic syndrome, whereas IgA nephropathy presented mainly as nephritic syndrome. Analysis of the classification performance of the four classifiers for IgA nephropathy and MN revealed that the random forest achieved the highest area under the ROC curve (AUC) (0.7639) and the highest specificity (0.8750). However, logistic regression attained the highest accuracy (0.7647) and the highest sensitivity (0.8889). CONCLUSIONS: Quantitative radiomics imaging features extracted from digital renal ultrasound are fully capable of distinguishing IgA nephropathy from MN. Radiomics analysis, a non-invasive method, is helpful for histological classification of glomerulopathy.

Multicenter Computer-Aided Diagnosis for Lymph Nodes Using Unsupervised Domain-Adaptation Networks Based on Cross-Domain Confounding Representations.

To achieve the robust high-performance computer-aided diagnosis systems for lymph nodes, CT images may be typically collected from multicenter data, which cause the isolated performance of the model based on different data source centers. The variability adaptation problem of lymph node data which is related to the problem of domain adaptation in deep learning differs from the general domain adaptation problem because of the typically larger CT image size and more complex data distributions. Therefore, domain adaptation for this problem needs to consider the shared feature representation and even the conditioning information of each domain so that the adaptation network can capture significant discriminative representations in a domain-invariant space. This paper extracts domain-invariant features based on a cross-domain confounding representation and proposes a cycle-consistency learning framework to encourage the network to preserve class-conditioning information through cross-domain image translations. Compared with the performance of different domain adaptation methods, the accurate rate of our method achieves at least 4.4% points higher under multicenter lymph node data. The pixel-level cross-domain image mapping and the semantic-level cycle consistency provided a stable confounding representation with class-conditioning information to achieve effective domain adaptation under complex feature distribution.

Application of CT radiomics in prediction of early recurrence in hepatocellular carcinoma.

PURPOSE: To appraise the ability of the computed tomography (CT) radiomics signature for prediction of early recurrence (ER) in patients with hepatocellular carcinoma (HCC). METHODS: A set of 325 HCC patients were enrolled in this retrospective study and the whole dataset was divided into 2 cohorts, including "training set" (225 patients) and "test set" (100 patients). All patients who underwent partial hepatectomy were followed up at least within 1 year. 656 Radiomics features were extracted from arterial-phase and portal venous-phase CT images. Lasso regression model was used for data dimension reduction, feature selection, and radiomics signature building. Univariate analysis was used to identify clinical and radiomics significant features. Models (radiomics signature, clinical model, and combined model) were evaluated by area under the curve (AUC) of receiver operating characteristic curve. The models' performances for prediction of ER were assessed. RESULTS: The radiomics signature was built by 14 selected radiomics features and was significantly associated with ER (P < 0.001); the AUCs of the "train set" and the "test set" were 0.818 (95% CI 0.760-0.865) and 0.719 (95% CI 0.621-0.805), respectively. The tumor size, tumor capsule, and γ-glutamyl transferase (GGT) were significantly associated with ER in the clinical model (P < 0.05). The combined model showed incremental prognostic value, with the AUCs of "training dataset" and "test dataset" were 0.846 (95% CI 0.792-0.890) and 0.737 (95% CI 0.640-0.820), respectively. The radiomics signature, tumor size, and the level of GGT were independent predictors of ER (P < 0.05). CONCLUSIONS: The CT radiomics signature can be conveniently used to predict the ER in patient with HCC. The combined model performed better for prediction of ER than radiomics signature or clinical model.

[Quantitative analysis of hepatocellular carcinomas pathological grading in non-contrast magnetic resonance images].

In order to solve the pathological grading of hepatocellular carcinomas (HCC) which depends on biopsy or surgical pathology invasively, a quantitative analysis method based on radiomics signature was proposed for pathological grading of HCC in non-contrast magnetic resonance imaging (MRI) images. The MRI images were integrated to predict clinical outcomes using 328 radiomics features, quantifying tumour image intensity, shape and text, which are extracted from lesion by manual segmentation. Least absolute shrinkage and selection operator (LASSO) were used to select the most-predictive radiomics features for the pathological grading. A radiomics signature, a clinical model, and a combined model were built. The association between the radiomics signature and HCC grading was explored. This quantitative analysis method was validated in 170 consecutive patients (training dataset: n = 125; validation dataset, n = 45), and cross-validation with receiver operating characteristic (ROC) analysis was performed and the area under the ROC curve (AUC) was employed as the prediction metric. Through the proposed method, AUC was 0.909 in training dataset and 0.800 in validation dataset, respectively. Overall, the prediction performances by radiomics features showed statistically significant correlations with pathological grading. The results showed that radiomics signature was developed to be a significant predictor for HCC pathological grading, which may serve as a noninvasive complementary tool for clinical doctors in determining the prognosis and therapeutic strategy for HCC.

Fully Convolutional DenseNet with Multiscale Context for Automated Breast Tumor Segmentation.

Breast tumor segmentation plays a crucial role in subsequent disease diagnosis, and most algorithms need interactive prior to firstly locate tumors and perform segmentation based on tumor-centric candidates. In this paper, we propose a fully convolutional network to achieve automatic segmentation of breast tumor in an end-to-end manner. Considering the diversity of shape and size for malignant tumors in the digital mammograms, we introduce multiscale image information into the fully convolutional dense network architecture to improve the segmentation precision. Multiple sampling rates of atrous convolution are concatenated to acquire different field-of-views of image features without adding additional number of parameters to avoid over fitting. Weighted loss function is also employed during training according to the proportion of the tumor pixels in the entire image, in order to weaken unbalanced classes problem. Qualitative and quantitative comparisons demonstrate that the proposed algorithm can achieve automatic tumor segmentation and has high segmentation precision for various size and shapes of tumor images without preprocessing and postprocessing.

Predicting the grade of hepatocellular carcinoma based on non-contrast-enhanced MRI radiomics signature.

PURPOSE: This study was conducted in order to investigate the value of magnetic resonance imaging (MRI)-based radiomics signatures for the preoperative prediction of hepatocellular carcinoma (HCC) grade. METHODS: Data from 170 patients confirmed to have HCC by surgical pathology were divided into a training group (n = 125) and a test group (n = 45). The radiomics features of tumours based on both T1-weighted imaging (WI) and T2WI were extracted by using Matrix Laboratory (MATLAB), and radiomics signatures were generated using the least absolute shrinkage and selection operator (LASSO) logistic regression model. The predicted values of pathological HCC grades using radiomics signatures, clinical factors (including age, sex, tumour size, alpha fetoprotein (AFP) level, history of hepatitis B, hepatocirrhosis, portal vein tumour thrombosis, portal hypertension and pseudocapsule) and the combined models were assessed. RESULTS: Radiomics signatures could successfully categorise high-grade and low-grade HCC cases (p < 0.05) in both the training and test datasets. Regarding the performances of clinical factors, radiomics signatures and the combined clinical and radiomics signature (from the combined T1WI and T2WI images) models for HCC grading prediction, the areas under the curve (AUCs) were 0.600, 0.742 and 0.800 in the test datasets, respectively. Both the AFP level and radiomics signature were independent predictors of HCC grade (p < 0.05). CONCLUSIONS: Radiomics signatures may be important for discriminating high-grade and low-grade HCC cases. The combination of the radiomics signatures with clinical factors may be helpful for the preoperative prediction of HCC grade. KEY POINTS: • The radiomics signature based on non-contrast-enhanced MR images was significantly associated with the pathological grade of HCC. • The radiomics signatures based on T1WI or T2WI images performed similarly at predicting the pathological grade of HCC. • Combining the radiomics signature and clinical factors (including age, sex, tumour size, AFP level, history of hepatitis B, hepatocirrhosis, portal vein tumour thrombosis, portal hypertension and pseudocapsule) may be helpful for the preoperative prediction of HCC grade.

Multi-level features combined end-to-end learning for automated pathological grading of breast cancer on digital mammograms.

We propose to discriminate the pathological grades directly on digital mammograms instead of pathological images. An end-to-end learning algorithm based on the combined multi-level features is proposed. Low-level features are extracted and selected by supervised LASSO logistic regression. Convolutional Neural Network (CNN) is designed to extract high-level semantic features. These extracted multi-level features are combined to optimize the new CNN end to end to make different parts of the network learn to pay attention to different level of features. Results demonstrate that our proposed algorithm is superior to other CNN models and obtain comparable performance compared with pathological images.