Search

Objectives: Develop and evaluate a deep learning-based automatic meningioma segmentation method for preoperative meningioma differentiation using radiomic features., Methods: A retrospective multicentre inclusion of MR examinations (T1/T2-weighted and contrast-enhanced T1-weighted imaging) was conducted. Data from centre 1 were allocated to training (n = 307, age = 50.94 ± 11.51) and internal testing (n = 238, age = 50.70 ± 12.72) cohorts, and data from centre 2 external testing cohort (n = 64, age = 48.45 ± 13.59). A modified attention U-Net was trained for meningioma segmentation. Segmentation accuracy was evaluated by five quantitative metrics. The agreement between radiomic features from manual and automatic segmentations was assessed using intra class correlation coefficient (ICC). After univariate and minimum-redundancy-maximum-relevance feature selection, L1-regularized logistic regression models for differentiating between low-grade (I) and high-grade (II and III) meningiomas were separately constructed using manual and automatic segmentations; their performances were evaluated using ROC analysis., Results: Dice of meningioma segmentation for the internal testing cohort were 0.94 ± 0.04 and 0.91 ± 0.05 for tumour volumes in contrast-enhanced T1-weighted and T2-weighted images, respectively; those for the external testing cohort were 0.90 ± 0.07 and 0.88 ± 0.07. Features extracted using manual and automatic segmentations agreed well, for both the internal (ICC = 0.94, interquartile range: 0.88-0.97) and external (ICC = 0.90, interquartile range: 0.78-70.96) testing cohorts. AUC of radiomic model with automatic segmentation was comparable with that of the model with manual segmentation for both the internal (0.95 vs. 0.93, p = 0.176) and external (0.88 vs. 0.91, p = 0.419) testing cohorts., Conclusions: The developed deep learning-based segmentation method enables automatic and accurate extraction of meningioma from multiparametric MR images and can help deploy radiomics for preoperative meningioma differentiation in clinical practice., Key Points: • A deep learning-based method was developed for automatic segmentation of meningioma from multiparametric MR images. • The automatic segmentation method enabled accurate extraction of meningiomas and yielded radiomic features that were highly consistent with those that were obtained using manual segmentation. • High-grade meningiomas were preoperatively differentiated from low-grade meningiomas using a radiomic model constructed on features from automatic segmentation., (© 2022. The Author(s), under exclusive licence to European Society of Radiology.)

Purpose: Diagnosis of lymph node metastasis (LNM) is critical for patients with pancreatic ductal adenocarcinoma (PDAC). We aimed to build deep learning radiomics (DLR) models of dual-energy computed tomography (DECT) to classify LNM status of PDAC and to stratify the overall survival before treatment., Methods: From August 2016 to October 2020, 148 PDAC patients underwent regional lymph node dissection and scanned preoperatively DECT were enrolled. The virtual monoenergetic image at 40 keV was reconstructed from 100 and 150 keV of DECT. By setting January 1, 2021, as the cut-off date, 113 patients were assigned into the primary set, and 35 were in the test set. DLR models using VMI 40 keV, 100 keV, 150 keV, and 100 + 150 keV images were developed and compared. The best model was integrated with key clinical features selected by multivariate Cox regression analysis to achieve the most accurate prediction., Results: DLR based on 100 + 150 keV DECT yields the best performance in predicting LNM status with the AUC of 0.87 (95% confidence interval [CI]: 0.85-0.89) in the test cohort. After integrating key clinical features (CT-reported T stage, LN status, glutamyl transpeptadase, and glucose), the AUC was improved to 0.92 (95% CI: 0.91-0.94). Patients at high risk of LNM portended significantly worse overall survival than those at low risk after surgery (P = 0.012)., Conclusions: The DLR model showed outstanding performance for predicting LNM in PADC and hold promise of improving clinical decision-making., (© 2021. The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.)

Background: Image registration technology has become an important medical image preprocessing step with the wide application of computer‐aided diagnosis technology in various medical image analysis tasks. Purpose: We propose a multiscale feature fusion registration based on deep learning to achieve the accurate registration and fusion of head magnetic resonance imaging (MRI) and solve the problem that general registration methods cannot handle the complex spatial information and position information of head MRI. Methods: Our proposed multiscale feature fusion registration network consists of three sequentially trained modules. The first is an affine registration module that implements affine transformation; the second is to realize non‐rigid transformation, a deformable registration module composed of top‐down and bottom‐up feature fusion subnetworks in parallel; and the third is a deformable registration module that also realizes non‐rigid transformation and is composed of two feature fusion subnetworks in series. The network decomposes the deformation field of large displacement into multiple deformation fields of small displacement by multiscale registration and registration, which reduces the difficulty of registration. Moreover, multiscale information in head MRI is learned in a targeted manner, which improves the registration accuracy, by connecting the two feature fusion subnetworks. Results: We used 29 3D head MRIs for training and seven volumes for testing and calculated the values of the registration evaluation metrics for the new algorithm to register anterior and posterior lateral pterygoid muscles. The Dice similarity coefficient was 0.745 ± 0.021, the Hausdorff distance was 3.441 ± 0.935 mm, the Average surface distance was 0.738 ± 0.098 mm, and the Standard deviation of the Jacobian matrix was 0.425 ± 0.043. Our new algorithm achieved a higher registration accuracy compared with state‐of‐the‐art registration methods. Conclusions: Our proposed multiscale feature fusion registration network can realize end‐to‐end deformable registration of 3D head MRI, which can effectively cope with the characteristics of large deformation displacement and the rich details of head images and provide reliable technical support for the diagnosis and analysis of head diseases. [ABSTRACT FROM AUTHOR]

• Rulefit enables the utilization of logical associations behind structured MRI report for nasopharyngeal carcinoma in the forms of prognostic rules. • The Rules model using Rulefit achieved a better accuracy and generalizability than common models for the prediction of overall survival among advanced nasopharyngeal carcinoma, with the highest C-index of 0.711 in external testing cohort. • Significant survival benefit was achieved in high-risk group stratified by Rules model with additional induction chemotherapy treatment. Structured MRI report facilitate prognostic prediction for nasopharyngeal carcinoma (NPC). However, the intrinsic association among structured variables is not fully utilised. This study aimed to investigate the performance of a Rulefit-based model in feature integration behind structured MRI report and prognostic prediction in advanced NPC. We retrospectively enrolled 1207 patients diagnosed with non-metastatic advanced NPC from two centres, and divided into training (N = 544), internal testing (N = 367), and external testing (N = 296) cohorts. Machine learning algorithms including multivariate analysis, deep learning, Lasso, and Rulefit were used to establish corresponding prognostic models. The concordance indices (C- indices) of three clinical and six combined models with different algorithms for overall survival (OS) prediction were compared. Survival benefits of induction chemotherapy (IC) were calculated among risk groups stratified by different models. A website was established for individualised survival visualisation. Incorporating structured variables into Stage model significantly improved the prognostic prediction performance. Six prognostic rules with structured variables were identified by Rulefit. OS prediction of Rules model was comparable to Lasso model in internal testing cohort (C-index: 0.720 vs. 0.713, P = 0.100) and achieved the highest C-index of 0.711 in external testing cohort, indicating better generalisability. The Rules model stratified patients into risk groups with significant 5-year OS differences in each cohort, and revealed significant survival benefits from additional IC in high-risk group. The Rulefit-based Rules model, with the revelation of intrinsic associations behind structured variables, is promising in risk stratification and guiding individualised IC treatment for advanced NPC. [ABSTRACT FROM AUTHOR]

• The SeqSeg addresses the problem of background dominance to achieve accurate NPC detection and segmentation. • Intelligent deep Q-learning-based nasopharyngeal carcinoma detection model is proposed. • Experiments on a large dataset from multi-device and multi-center demonstrate the reliability of the proposed method. • Efficient reward function and exploration strategy boost NPC detectio performance. • Recurrent attention and dilated border weighted loss function foster NPC segmentation. [Display omitted] Reliable nasopharyngeal carcinoma (NPC) segmentation plays an important role in radiotherapy planning. However, recent deep learning methods fail to achieve satisfactory NPC segmentation in magnetic resonance (MR) images, since NPC is infiltrative and typically has a small or even tiny volume with indistinguishable border, making it indiscernible from tightly connected surrounding tissues from immense and complex backgrounds. To address such background dominance problems, this paper proposes a sequential method (SeqSeg) to achieve accurate NPC segmentation. Specifically, the proposed SeqSeg is devoted to solving the problem at two scales: the instance level and feature level. At the instance level, SeqSeg is forced to focus attention on the tumor and its surrounding tissue through the deep Q-learning (DQL)-based NPC detection model by prelocating the tumor and reducing the scale of the segmentation background. Next, at the feature level, SeqSeg uses high-level semantic features in deeper layers to guide feature learning in shallower layers, thus directing the channel-wise and region-wise attention to mine tumor-related features to perform accurate segmentation. The performance of our proposed method is evaluated by extensive experiments on the large NPC dataset containing 1101 patients. The experimental results demonstrated that the proposed SeqSeg not only outperforms several state-of-the-art methods but also achieves better performance in multi-device and multi-center datasets. [ABSTRACT FROM AUTHOR]