Simultaneous detection of dental caries and fissure sealant in intraoral photos by deep learning: a pilot study.

BACKGROUND: Deep learning, as an artificial intelligence method has been proved to be powerful in analyzing images. The purpose of this study is to construct a deep learning-based model (ToothNet) for the simultaneous detection of dental caries and fissure sealants in intraoral photos. METHODS: A total of 1020 intraoral photos were collected from 762 volunteers. Teeth, caries and sealants were annotated by two endodontists using the LabelMe tool. ToothNet was developed by modifying the YOLOX framework for simultaneous detection of caries and fissure sealants. The area under curve (AUC) in the receiver operating characteristic curve (ROC) and free-response ROC (FROC) curves were used to evaluate model performance in the following aspects: (i) classification accuracy of detecting dental caries and fissure sealants from a photograph (image-level); and (ii) localization accuracy of the locations of predicted dental caries and fissure sealants (tooth-level). The performance of ToothNet and dentist with 1year of experience (1-year dentist) were compared at tooth-level and image-level using Wilcoxon test and DeLong test. RESULTS: At the image level, ToothNet achieved an AUC of 0.925 (95% CI, 0.880-0.958) for caries detection and 0.902 (95% CI, 0.853-0.940) for sealant detection. At the tooth level, with a confidence threshold of 0.5, the sensitivity, precision, and F1-score for caries detection were 0.807, 0.814, and 0.810, respectively. For fissure sealant detection, the values were 0.714, 0.750, and 0.731. Compared with ToothNet, the 1-year dentist had a lower F1 value (0.599, p < 0.0001) and AUC (0.749, p < 0.0001) in caries detection, and a lower F1 value (0.727, p = 0.023) and similar AUC (0.829, p = 0.154) in sealant detection. CONCLUSIONS: The proposed deep learning model achieved multi-task simultaneous detection in intraoral photos and showed good performance in the detection of dental caries and fissure sealants. Compared with 1-year dentist, the model has advantages in caries detection and is equivalent in fissure sealants detection.

A novel multidisciplinary machine learning approach based on clinical, imaging, colonoscopy, and pathology features for distinguishing intestinal tuberculosis from Crohn's disease.

OBJECTIVES: Differentiating intestinal tuberculosis (ITB) from Crohn's disease (CD) remains a diagnostic dilemma. Misdiagnosis carries potential grave implications. We aim to establish a multidisciplinary-based model using machine learning approach for distinguishing ITB from CD. METHODS: Eighty-two patients including 25 patients with ITB and 57 patients with CD were retrospectively recruited (54 in training cohort and 28 in testing cohort). The region of interest (ROI) for the lesion was delineated on magnetic resonance enterography (MRE) and colonoscopy images. Radiomic features were extracted by least absolute shrinkage and selection operator regression. Pathological feature was extracted automatically by deep-learning method. Clinical features were filtered by logistic regression analysis. Diagnostic performance was evaluated by receiver operating characteristic (ROC) curve and decision curve analysis (DCA). Delong's test was applied to compare the efficiency between the multidisciplinary-based model and the other four single-disciplinary-based models. RESULTS: The radiomics model based on MRE features yielded an AUC of 0.87 (95% confidence interval [CI] 0.68-0.96) on the test data set, which was similar to the clinical model (AUC, 0.90 [95% CI 0.71-0.98]) and higher than the colonoscopy radiomics model (AUC, 0.68 [95% CI 0.48-0.84]) and pathology deep-learning model (AUC, 0.70 [95% CI 0.49-0.85]). Multidisciplinary model, integrating 3 clinical, 21 MRE radiomic, 5 colonoscopy radiomic, and 4 pathology deep-learning features, could significantly improve the diagnostic performance (AUC of 0.94, 95% CI 0.78-1.00) on the bases of single-disciplinary-based models. DCA confirmed the clinical utility. CONCLUSIONS: Multidisciplinary-based model integrating clinical, MRE, colonoscopy, and pathology features was useful in distinguishing ITB from CD.

Development and validation of a deep learning radiomics model with clinical-radiological characteristics for the identification of occult peritoneal metastases in patients with pancreatic ductal adenocarcinoma.

BACKGROUND: Occult peritoneal metastases (OPM) in patients with pancreatic ductal adenocarcinoma (PDAC) are frequently overlooked during imaging. The authors aimed to develop and validate a computed tomography (CT)-based deep learning-based radiomics (DLR) model to identify OPM in PDAC before treatment. METHODS: This retrospective, bicentric study included 302 patients with PDAC (training: n =167, OPM-positive, n =22; internal test: n =72, OPM-positive, n =9: external test, n =63, OPM-positive, n =9) who had undergone baseline CT examinations between January 2012 and October 2022. Handcrafted radiomics (HCR) and DLR features of the tumor and HCR features of peritoneum were extracted from CT images. Mutual information and least absolute shrinkage and selection operator algorithms were used for feature selection. A combined model, which incorporated the selected clinical-radiological, HCR, and DLR features, was developed using a logistic regression classifier using data from the training cohort and validated in the test cohorts. RESULTS: Three clinical-radiological characteristics (carcinoembryonic antigen 19-9 and CT-based T and N stages), nine HCR features of the tumor, 14 DLR features of the tumor, and three HCR features of the peritoneum were retained after feature selection. The combined model yielded satisfactory predictive performance, with an area under the curve (AUC) of 0.853 (95% CI: 0.790-0.903), 0.845 (95% CI: 0.740-0.919), and 0.852 (95% CI: 0.740-0.929) in the training, internal test, and external test cohorts, respectively (all P <0.05). The combined model showed better discrimination than the clinical-radiological model in the training (AUC=0.853 vs. 0.612, P <0.001) and the total test (AUC=0.842 vs. 0.638, P <0.05) cohorts. The decision curves revealed that the combined model had greater clinical applicability than the clinical-radiological model. CONCLUSIONS: The model combining CT-based DLR and clinical-radiological features showed satisfactory performance for predicting OPM in patients with PDAC.

Deep Learning Combined with Radiologist's Intervention Achieves Accurate Segmentation of Hepatocellular Carcinoma in Dual-Phase Magnetic Resonance Images.

Purpose: Segmentation of hepatocellular carcinoma (HCC) is crucial; however, manual segmentation is subjective and time-consuming. Accurate and automatic lesion contouring for HCC is desirable in clinical practice. In response to this need, our study introduced a segmentation approach for HCC combining deep convolutional neural networks (DCNNs) and radiologist intervention in magnetic resonance imaging (MRI). We sought to design a segmentation method with a deep learning method that automatically segments using manual location information for moderately experienced radiologists. In addition, we verified the viability of this method to assist radiologists in accurate and fast lesion segmentation. Method: In our study, we developed a semiautomatic approach for segmenting HCC using DCNN in conjunction with radiologist intervention in dual-phase gadolinium-ethoxybenzyl-diethylenetriamine penta-acetic acid- (Gd-EOB-DTPA-) enhanced MRI. We developed a DCNN and deep fusion network (DFN) trained on full-size images, namely, DCNN-F and DFN-F. Furthermore, DFN was applied to the image blocks containing tumor lesions that were roughly contoured by a radiologist with 10 years of experience in abdominal MRI, and this method was named DFN-R. Another radiologist with five years of experience (moderate experience) performed tumor lesion contouring for comparison with our proposed methods. The ground truth image was contoured by an experienced radiologist and reviewed by an independent experienced radiologist. Results: The mean DSC of DCNN-F, DFN-F, and DFN-R was 0.69 ± 0.20 (median, 0.72), 0.74 ± 0.21 (median, 0.77), and 0.83 ± 0.13 (median, 0.88), respectively. The mean DSC of the segmentation by the radiologist with moderate experience was 0.79 ± 0.11 (median, 0.83), which was lower than the performance of DFN-R. Conclusions: Deep learning using dual-phase MRI shows great potential for HCC lesion segmentation. The radiologist-aided semiautomated method (DFN-R) achieved improved performance compared to manual contouring by the radiologist with moderate experience, although the difference was not statistically significant.

Predicting brain age using partition modeling strategy and atlas-based attentional enhancement in the Chinese population.

As a biomarker of human brain health during development, brain age is estimated based on subtle differences in brain structure from those under typical developmental. Magnetic resonance imaging (MRI) is a routine diagnostic method in neuroimaging. Brain age prediction based on MRI has been widely studied. However, few studies based on Chinese population have been reported. This study aimed to construct a brain age predictive model for the Chinese population across its lifespan. We developed a partition prediction method based on transfer learning and atlas attention enhancement. The participants were separated into four age groups, and a deep learning model was trained for each group to identify the brain regions most critical for brain age prediction. The Atlas attention-enhancement method was also used to help the models focus only on critical brain regions. The proposed method was validated using 354 participants from domestic datasets. For prediction performance in the testing sets, the mean absolute error was 2.218 ± 1.801 years, and the Pearson correlation coefficient (r) was 0.969, exceeding previous results for wide-range brain age prediction. In conclusion, the proposed method could provide brain age estimation to assist in assessing the status of brain health.

Deep learning for differentiation of osteolytic osteosarcoma and giant cell tumor around the knee joint on radiographs: a multicenter study.

OBJECTIVES: To develop a deep learning (DL) model for differentiating between osteolytic osteosarcoma (OS) and giant cell tumor (GCT) on radiographs. METHODS: Patients with osteolytic OS and GCT proven by postoperative pathology were retrospectively recruited from four centers (center A, training and internal testing; centers B, C, and D, external testing). Sixteen radiologists with different experiences in musculoskeletal imaging diagnosis were divided into three groups and participated with or without the DL model's assistance. DL model was generated using EfficientNet-B6 architecture, and the clinical model was trained using clinical variables. The performance of various models was compared using McNemar's test. RESULTS: Three hundred thirty-three patients were included (mean age, 27 years ± 12 [SD]; 186 men). Compared to the clinical model, the DL model achieved a higher area under the curve (AUC) in both the internal (0.97 vs. 0.77, p = 0.008) and external test set (0.97 vs. 0.64, p < 0.001). In the total test set (including the internal and external test sets), the DL model achieved higher accuracy than the junior expert committee (93.1% vs. 72.4%; p < 0.001) and was comparable to the intermediate and senior expert committee (93.1% vs. 88.8%, p = 0.25; 87.1%, p = 0.35). With DL model assistance, the accuracy of the junior expert committee was improved from 72.4% to 91.4% (p = 0.051). CONCLUSION: The DL model accurately distinguished osteolytic OS and GCT with better performance than the junior radiologists, whose own diagnostic performances were significantly improved with the aid of the model, indicating the potential for the differential diagnosis of the two bone tumors on radiographs. CRITICAL RELEVANCE STATEMENT: The deep learning model can accurately distinguish osteolytic osteosarcoma and giant cell tumor on radiographs, which may help radiologists improve the diagnostic accuracy of two types of tumors. KEY POINTS: • The DL model shows robust performance in distinguishing osteolytic osteosarcoma and giant cell tumor. • The diagnosis performance of the DL model is better than junior radiologists'. • The DL model shows potential for differentiating osteolytic osteosarcoma and giant cell tumor.

Deep Learning Radiomic Analysis of MRI Combined with Clinical Characteristics Diagnoses Placenta Accreta Spectrum and its Subtypes.

BACKGROUND: Different placenta accreta spectrum (PAS) subtypes pose varying surgical risks to the parturient. Machine learning model has the potential to diagnose PAS disorder. PURPOSE: To develop a cascaded deep semantic-radiomic-clinical (DRC) model for diagnosing PAS and its subtypes based on T2-weighted MRI. STUDY TYPE: Retrospective. POPULATION: 361 pregnant women (mean age: 33.10 ± 4.37 years), suspected of PAS, divided into segment training cohort (N = 40), internal training cohort (N = 139), internal testing cohort (N = 60), and external testing cohort (N = 122). FIELD STRENGTH/SEQUENCE: Coronal T2-weighted sequence at 1.5 T and 3.0 T. ASSESSMENT: Clinical characteristics such as history of uterine surgery and the presence of placenta previa, complete placenta previa and dangerous placenta previa were extracted from clinical records. The DRC model (incorporating radiomics, deep semantic features, and clinical characteristics), a cumulative radiological score method performed by radiologists, and other models (including a radiomics and clinical, the clinical, radiomics and deep learning models) were developed for PAS disorder diagnosing (existence of PAS and its subtypes). STATISTICAL TESTS: AUC, ACC, Student's t-test, the Mann-Whitney U test, chi-squared test, dice coefficient, intraclass correlation coefficients, least absolute shrinkage and selection operator regression, receiver operating characteristic curve, calibration curve with the Hosmer-Lemeshow test, decision curve analysis, DeLong test, and McNemar test. P < 0.05 indicated a significant difference. RESULTS: In PAS diagnosis, the DRC-1 outperformed than other models (AUC = 0.850 and 0.841 in internal and external testing cohorts, respectively). In PAS subtype classification (abnormal adherent placenta and abnormal invasive placenta), DRC-2 model performed similarly with radiologists (P = 0.773 and 0.579 in the internal testing cohort and P = 0.429 and 0.874 in the external testing cohort, respectively). DATA CONCLUSION: The DRC model offers efficiency and high diagnostic sensitivity in diagnosis, aiding in surgical planning. LEVEL OF EVIDENCE: 3 TECHNICAL EFFICACY: Stage 2.

Deep learning model to differentiate Crohn's disease from intestinal tuberculosis using histopathological whole slide images from intestinal specimens.

Crohn's disease (CD) and intestinal tuberculosis (ITB) share similar histopathological characteristics, and differential diagnosis can be a dilemma for pathologists. This study aimed to apply deep learning (DL) to analyze whole slide images (WSI) of surgical resection specimens to distinguish CD from ITB. Overall, 1973 WSI from 85 cases from 3 centers were obtained. The DL model was established in internal training and validated in external test cohort, evaluated by area under receiver operator characteristic curve (AUC). Diagnostic results of pathologists were compared with those of the DL model using DeLong's test. DL model had case level AUC of 0.886, 0.893 and slide level AUC of 0.954, 0.827 in training and test cohorts. Attention maps highlighted discriminative areas and top 10 features were extracted from CD and ITB. DL model's diagnostic efficiency (AUC = 0.886) was better than junior pathologists (*1 AUC = 0.701, P = 0.088; *2 AUC = 0.861, P = 0.788) and inferior to senior GI pathologists (*3 AUC = 0.910, P = 0.800; *4 AUC = 0.946, P = 0.507) in training cohort. In the test cohort, model (AUC = 0.893) outperformed senior non-GI pathologists (*5 AUC = 0.782, P = 0.327; *6 AUC = 0.821, P = 0.516). We developed a DL model for the classification of CD and ITB, improving pathological diagnosis accuracy effectively.

CT-based radiomics signature of visceral adipose tissue and bowel lesions for identifying patients with Crohn's disease resistant to infliximab.

PURPOSE: To develop a CT-based radiomics model combining with VAT and bowel features to improve the predictive efficacy of IFX therapy on the basis of bowel model. METHODS: This retrospective study included 231 CD patients (training cohort, n = 112; internal validation cohort, n = 48; external validation cohort, n = 71) from two tertiary centers. Machine-learning VAT model and bowel model were developed separately to identify CD patients with primary nonresponse to IFX. A comprehensive model incorporating VAT and bowel radiomics features was further established to verify whether CT features extracted from VAT would improve the predictive efficacy of bowel model. Area under the curve (AUC) and decision curve analysis were used to compare the prediction performance. Clinical utility was assessed by integrated differentiation improvement (IDI). RESULTS: VAT model and bowel model exhibited comparable performance for identifying patients with primary nonresponse in both internal (AUC: VAT model vs bowel model, 0.737 (95% CI, 0.590-0.854) vs. 0.832 (95% CI, 0.750-0.896)) and external validation cohort [AUC: VAT model vs. bowel model, 0.714 (95% CI, 0.595-0.815) vs. 0.799 (95% CI, 0.687-0.885)), exhibiting a relatively good net benefit. The comprehensive model incorporating VAT into bowel model yielded a satisfactory predictive efficacy in both internal (AUC, 0.840 (95% CI, 0.706-0.930)) and external validation cohort (AUC, 0.833 (95% CI, 0.726-0.911)), significantly better than bowel alone (IDI = 4.2% and 3.7% in internal and external validation cohorts, both p < 0.05). CONCLUSION: VAT has an effect on IFX treatment response. It improves the performance for identification of CD patients at high risk of primary nonresponse to IFX therapy with selected features from RM. CRITICAL RELEVANCE STATEMENT: Our radiomics model (RM) for VAT-bowel analysis captured the pathophysiological changes occurring in VAT and whole bowel lesion, which could help to identify CD patients who would not response to infliximab at the beginning of therapy. KEY POINTS: • Radiomics signatures with VAT and bowel alone or in combination predicting infliximab efficacy. • VAT features contribute to the prediction of IFX treatment efficacy. • Comprehensive model improved the performance compared with the bowel model alone.

Deep learning-based whole-body characterization of prostate cancer lesions on [68Ga]Ga-PSMA-11 PET/CT in patients with post-prostatectomy recurrence.

PURPOSE: The automatic segmentation and detection of prostate cancer (PC) lesions throughout the body are extremely challenging due to the lesions' complexity and variability in appearance, shape, and location. In this study, we investigated the performance of a three-dimensional (3D) convolutional neural network (CNN) to automatically characterize metastatic lesions throughout the body in a dataset of PC patients with recurrence after radical prostatectomy. METHODS: We retrospectively collected [68 Ga]Ga-PSMA-11 PET/CT images from 116 patients with metastatic PC at two centers: center 1 provided the data for fivefold cross validation (n = 78) and internal testing (n = 19), and center 2 provided the data for external testing (n = 19). PET and CT data were jointly input into a 3D U-Net to achieve whole-body segmentation and detection of PC lesions. The performance in both the segmentation and the detection of lesions throughout the body was evaluated using established metrics, including the Dice similarity coefficient (DSC) for segmentation and the recall, precision, and F1-score for detection. The correlation and consistency between tumor burdens (PSMA-TV and TL-PSMA) calculated from automatic segmentation and artificial ground truth were assessed by linear regression and Bland‒Altman plots. RESULTS: On the internal test set, the DSC, precision, recall, and F1-score values were 0.631, 0.961, 0.721, and 0.824, respectively. On the external test set, the corresponding values were 0.596, 0.888, 0.792, and 0.837, respectively. Our approach outperformed previous studies in segmenting and detecting metastatic lesions throughout the body. Tumor burden indicators derived from deep learning and ground truth showed strong correlation (R2 ≥ 0.991, all P < 0.05) and consistency. CONCLUSION: Our 3D CNN accurately characterizes whole-body tumors in relapsed PC patients; its results are highly consistent with those of manual contouring. This automatic method is expected to improve work efficiency and to aid in the assessment of tumor burden.

Development and Validation of a Deep Learning and Radiomics Combined Model for Differentiating Complicated From Uncomplicated Acute Appendicitis.

RATIONALE AND OBJECTIVES: This study aimed to develop and validate a deep learning and radiomics combined model for differentiating complicated from uncomplicated acute appendicitis (AA). MATERIALS AND METHODS: This retrospective multicenter study included 1165 adult AA patients (training cohort, 700 patients; validation cohort, 465 patients) with available abdominal pelvic computed tomography (CT) images. The reference standard for complicated/uncomplicated AA was the surgery and pathology records. We developed our combined model with CatBoost based on the selected clinical characteristics, CT visual features, deep learning features, and radiomics features. We externally validated our combined model and compared its performance with that of the conventional combined model, the deep learning radiomics (DLR) model, and the radiologist's visual diagnosis using receiver operating characteristic (ROC) curve analysis. RESULTS: In the training cohort, the area under the ROC curve (AUC) of our combined model in distinguishing complicated from uncomplicated AA was 0.816 (95% confidence interval [CI]: 0.785-0.844). In the validation cohort, our combined model showed robust performance across the data from three centers, with AUCs of 0.836 (95% CI: 0.785-0.879), 0.793 (95% CI: 0.695-0.872), and 0.723 (95% CI: 0.632-0.802). In the total validation cohort, our combined model (AUC = 0.799) performed better than the conventional combined model, DLR model, and radiologist's visual diagnosis (AUC = 0.723, 0.755, and 0.679, respectively; all P < 0.05). Decision curve analysis showed that our combined model provided greater net benefit in predicting complicated AA than the other three models. CONCLUSION: Our combined model allows the accurate differentiation of complicated and uncomplicated AA.

Noninvasive identification of HER2-low-positive status by MRI-based deep learning radiomics predicts the disease-free survival of patients with breast cancer.

OBJECTIVE: This study aimed to establish a MRI-based deep learning radiomics (DLR) signature to predict the human epidermal growth factor receptor 2 (HER2)-low-positive status and further verified the difference in prognosis by the DLR model. METHODS: A total of 481 patients with breast cancer who underwent preoperative MRI were retrospectively recruited from two institutions. Traditional radiomics features and deep semantic segmentation feature-based radiomics (DSFR) features were extracted from segmented tumors to construct models separately. Then, the DLR model was constructed to assess the HER2 status by averaging the output probabilities of the two models. Finally, a Kaplan‒Meier survival analysis was conducted to explore the disease-free survival (DFS) in patients with HER2-low-positive status. The multivariate Cox proportional hazard model was constructed to further determine the factors associated with DFS. RESULTS: First, the DLR model distinguished between HER2-negative and HER2-overexpressing patients with AUCs of 0.868 and 0.763 in the training and validation cohorts, respectively. Furthermore, the DLR model distinguished between HER2-low-positive and HER2-zero patients with AUCs of 0.855 and 0.750, respectively. Cox regression analysis showed that the prediction score obtained using the DLR model (HR, 0.175; p = 0.024) and lesion size (HR, 1.043; p = 0.009) were significant, independent predictors of DFS. CONCLUSIONS: We successfully constructed a DLR model based on MRI to noninvasively evaluate the HER2 status and further revealed prospects for predicting the DFS of patients with HER2-low-positive status. CLINICAL RELEVANCE STATEMENT: The MRI-based DLR model could noninvasively identify HER2-low-positive status, which is considered a novel prognostic predictor and therapeutic target. KEY POINTS: • The DLR model effectively distinguished the HER2 status of breast cancer patients, especially the HER2-low-positive status. • The DLR model was better than the traditional radiomics model or DSFR model in distinguishing HER2 expression. • The prediction score obtained using the model and lesion size were significant independent predictors of DFS.

Fast and robust pulsed chemical exchange saturation transfer (CEST) MRI using a quasi-steady-state (QUASS) algorithm at 3 T.

Chemical exchange saturation transfer (CEST) has emerged as a powerful technique to image dilute labile protons. However, its measurement depends on the RF saturation duration (Tsat) and relaxation delay (Trec). Although the recently developed quasi-steady-state (QUASS) solution can reconstruct equilibrium CEST effects under continuous-wave RF saturation, it does not apply to pulsed-CEST MRI on clinical scanners with restricted hardware or specific absorption rate limits. This study proposed a QUASS algorithm for pulsed-CEST MRI and evaluated its performance in muscle CEST measurement. An approximated expression of a steady-state pulsed-CEST signal was incorporated in the off-resonance spin-lock model, from which the QUASS pulsed-CEST effect was derived. Numerical simulation, creatine phantom, and healthy volunteer scans were conducted at 3 T. The CEST effect was quantified with asymmetry analysis in the simulation and phantom experiments. CEST effects of creatine, amide proton transfer, phosphocreatine, and combined magnetization transfer and nuclear Overhauser effects were isolated from a multi-pool Lorentzian model in muscles. Apparent and QUASS CEST measurements were compared under different Tsat/Trec and duty cycles. Paired Student's t-test was employed with P < 0.05 as statistically significant. The simulation, phantom, and human studies showed the strong impact of Tsat/Trec on apparent CEST measurements, which were significantly smaller than the corresponding QUASS CEST measures, especially under short Tsat/Trec times. In comparison, the QUASS algorithm mitigates such impact and enables accurate CEST measurements under short Tsat/Trec times. In conclusion, the QUASS algorithm can accelerate robust pulsed-CEST MRI, promising the efficient detection and evaluation of muscle diseases in clinical settings.

Radiomics Prediction of Muscle Invasion in Bladder Cancer Using Semi-Automatic Lesion Segmentation of MRI Compared with Manual Segmentation.

Conventional radiomics analysis requires the manual segmentation of lesions, which is time-consuming and subjective. This study aimed to assess the feasibility of predicting muscle invasion in bladder cancer (BCa) with radiomics using a semi-automatic lesion segmentation method on T2-weighted images. Cases of non-muscle-invasive BCa (NMIBC) and muscle-invasive BCa (MIBC) were pathologically identified in a training cohort and in internal and external validation cohorts. For bladder tumor segmentation, a deep learning-based semi-automatic model was constructed, while manual segmentation was performed by a radiologist. Semi-automatic and manual segmentation results were respectively used in radiomics analyses to distinguish NMIBC from MIBC. An equivalence test was used to compare the models' performance. The mean Dice similarity coefficients of the semi-automatic segmentation method were 0.836 and 0.801 in the internal and external validation cohorts, respectively. The area under the receiver operating characteristic curve (AUC) were 1.00 (0.991) and 0.892 (0.894) for the semi-automated model (manual) on the internal and external validation cohort, respectively (both p < 0.05). The average total processing time for semi-automatic segmentation was significantly shorter than that for manual segmentation (35 s vs. 92 s, p < 0.001). The BCa radiomics model based on semi-automatic segmentation method had a similar diagnostic performance as that of manual segmentation, while being less time-consuming and requiring fewer manual interventions.

Dual-Energy CT Deep Learning Radiomics to Predict Macrotrabecular-Massive Hepatocellular Carcinoma.

Background It is unknown whether the additional information provided by multiparametric dual-energy CT (DECT) could improve the noninvasive diagnosis of the aggressive macrotrabecular-massive (MTM) subtype of hepatocellular carcinoma (HCC). Purpose To evaluate the diagnostic performance of dual-phase contrast-enhanced multiparametric DECT for predicting MTM HCC. Materials and Methods Patients with histopathologic examination-confirmed HCC who underwent contrast-enhanced DECT between June 2019 and June 2022 were retrospectively recruited from three independent centers (center 1, training and internal test data set; centers 2 and 3, external test data set). Radiologic features were visually analyzed and combined with clinical information to establish a clinical-radiologic model. Deep learning (DL) radiomics models were based on DL features and handcrafted features extracted from virtual monoenergetic images and material composition images on dual phase using binary least absolute shrinkage and selection operators. A DL radiomics nomogram was developed using multivariable logistic regression analysis. Model performance was evaluated with the area under the receiver operating characteristic curve (AUC), and the log-rank test was used to analyze recurrence-free survival. Results A total of 262 patients were included (mean age, 54 years ± 12 [SD]; 225 men [86%]; training data set, n = 146 [56%]; internal test data set, n = 35 [13%]; external test data set, n = 81 [31%]). The DL radiomics nomogram better predicted MTM than the clinical-radiologic model (AUC = 0.91 vs 0.77, respectively, for the training set [P < .001], 0.87 vs 0.72 for the internal test data set [P = .04], and 0.89 vs 0.79 for the external test data set [P = .02]), with similar sensitivity (80% vs 87%, respectively; P = .63) and higher specificity (90% vs 63%; P < .001) in the external test data set. The predicted positive MTM groups based on the DL radiomics nomogram had shorter recurrence-free survival than predicted negative MTM groups in all three data sets (training data set, P = .04; internal test data set, P = .01; and external test data set, P = .03). Conclusion A DL radiomics nomogram derived from multiparametric DECT accurately predicted the MTM subtype in patients with HCC. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Chu and Fishman in this issue.

Classification of temporal lobe epilepsy based on neuropsychological tests and exploration of its underlying neurobiology.

Objective: To assist improving long-term postoperative seizure-free rate, we aimed to use machine learning algorithms based on neuropsychological data to differentiate temporal lobe epilepsy (TLE) from extratemporal lobe epilepsy (extraTLE), as well as explore the relationship between magnetic resonance imaging (MRI) and neuropsychological tests. Methods: Twenty-three patients with TLE and 23 patients with extraTLE underwent neuropsychological tests and MRI scans before surgery. The least absolute shrinkage and selection operator were firstly employed for feature selection, and a machine learning approach with neuropsychological tests was employed to classify TLE using leave-one-out cross-validation. A generalized linear model was used to analyze the relationship between brain alterations and neuropsychological tests. Results: We found that logistic regression with the selected neuropsychological tests generated classification accuracies of 87.0%, with an area under the receiver operating characteristic curve (AUC) of 0.89. Three neuropsychological tests were acquired as significant neuropsychological signatures for the diagnosis of TLE. We also found that the Right-Left Orientation Test difference was related to the superior temporal and the banks of the superior temporal sulcus (bankssts). The Conditional Association Learning Test (CALT) was associated with the cortical thickness difference in the lateral orbitofrontal area between the two groups, and the Component Verbal Fluency Test was associated with the cortical thickness difference in the lateral occipital cortex between the two groups. Conclusion: These results showed that machine learning-based classification with the selected neuropsychological data can successfully classify TLE with high accuracy compared to previous studies, which could provide kind of warning sign for surgery candidate of TLE patients. In addition, understanding the mechanism of cognitive behavior by neuroimaging information could assist doctors in the presurgical evaluation of TLE.

Improving Tumor Classification by Reusing Self-predicted Segmentation of Medical Images as Guiding Knowledge.

Differential diagnosis of tumors is important for computer-aided diagnosis. In computer-aided diagnosis systems, expert knowledge of lesion segmentation masks is limited as it is only used during preprocessing or as supervision to guide feature extraction. To improve the utilization of lesion segmentation masks, this study proposes a simple and effective multitask learning network that improves medical image classification using self-predicted segmentation as guiding knowledge; we call this network RS 2-net. In RS 2-net, the predicted segmentation probability map obtained from the initial segmentation inference is added to the original image to form a new input, which is then reinput to the network for the final classification inference. We validated the proposed RS 2-net using three datasets: the pNENs-Grade dataset, which tested the prediction of pancreatic neuroendocrine neoplasm grading, and the HCC-MVI dataset, which tested the prediction of microvascular invasion of hepatocellular carcinoma, and ISIC 2017 public skin lesion dataset. The experimental results indicate that the proposed strategy of reusing self-predicted segmentation is effective, and RS 2-net outperforms other popular networks and existing state-of-the-art studies. Interpretive analytics based on feature visualization demonstrates that the improved classification performance of our reuse strategy is due to the semantic information that can be acquired in advance in a shallow network.

Serine-arginine protein kinase 1 (SRPK1) promotes EGFR-TKI resistance by enhancing GSK3ß Ser9 autophosphorylation independent of its kinase activity in non-small-cell lung cancer.

Resistance to epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors (TKIs) is a major challenge for clinicians and patients with non-small cell lung cancer (NSCLC). Serine-arginine protein kinase 1 (SRPK1) is a key oncoprotein in the EGFR/AKT pathway that participates in tumorigenesis. We found that high SRPK1 expression was significantly associated with poor progression-free survival (PFS) in patients with advanced NSCLC undergoing gefitinib treatment. Both in vitro and in vivo assays suggested that SRPK1 reduced the ability of gefitinib to induce apoptosis in sensitive NSCLC cells independently of its kinase activity. Moreover, SRPK1 facilitated binding between LEF1, ß-catenin and the EGFR promoter region to increase EGFR expression and promote the accumulation and phosphorylation of membrane EGFR. Furthermore, we verified that the SRPK1 spacer domain bound to GSK3ß and enhanced its autophosphorylation at Ser9 to activate the Wnt pathway, thereby promoting the expression of Wnt target genes such as Bcl-X. The correlation between SRPK1 and EGFR expression was confirmed in patients. In brief, our research suggested that the SRPK1/GSK3ß axis promotes gefitinib resistance by activating the Wnt pathway and may serve as a potential therapeutic target for overcoming gefitinib resistance in NSCLC.

Predicting muscle invasion in bladder cancer based on MRI: A comparison of radiomics, and single-task and multi-task deep learning.

BACKGROUND AND OBJECTIVES: Radiomics and deep learning are two popular technologies used to develop computer-aided detection and diagnosis schemes for analysing medical images. This study aimed to compare the effectiveness of radiomics, single-task deep learning (DL) and multi-task DL methods in predicting muscle-invasive bladder cancer (MIBC) status based on T2-weighted imaging (T2WI). METHODS: A total of 121 tumours (93 for training, from Centre 1; 28 for testing, from Centre 2) were included. MIBC was confirmed with pathological examination. A radiomics model, a single-task model, and a multi-task model based on T2WI were constructed in the training cohort with five-fold cross-validation, and validation was conducted in the external test cohort. Receiver operating characteristic (ROC) curve analysis was performed to evaluate the diagnostic performance of each model. DeLong's test and a permutation test were used to compare the performance of the models. RESULTS: The area under the ROC curve (AUC) values of the radiomics, single-task and multi-task models in the training cohort were: 0.920, 0.933 and 0.932, respectively; and were 0.844, 0.884 and 0.932, respectively, in the test cohort. The multi-task model achieved better performance in the test cohort than did the other models. No statistically significant differences in AUC values and Kappa coefficients were observed between pairwise models, in either the training or test cohorts. According to the Grad-CAM feature visualization results, the multi-task model focused more on the diseased tissue area in some samples of the test cohort compared with the single-task model. CONCLUSIONS: The T2WI-based radiomics, single-task, and multi-task models all exhibited good diagnostic performance in preoperatively predicting MIBC, in which the multi-task model had the best diagnostic performance. Compared with the radiomics method, our multi-task DL method had the advantage of saving time and effort. Compared with the single-task DL method, our multi-task DL method had the advantage of being more lesion-focused and more reliable for clinical reference.

CT-based radiomics signature of visceral adipose tissue for prediction of disease progression in patients with Crohn's disease: A multicentre cohort study.

Background: Visceral adipose tissue (VAT) is involved in the pathogenesis of Crohn's disease (CD). However, data describing its effects on CD progression remain scarce. We developed and validated a VAT-radiomics model (RM) using computed tomography (CT) images to predict disease progression in patients with CD and compared it with a subcutaneous adipose tissue (SAT)-RM. Methods: This retrospective study included 256 patients with CD (training, n = 156; test, n = 100) who underwent baseline CT examinations from June 19, 2015 to June 14, 2020 at three tertiary referral centres (The First Affiliated Hospital of Sun Yat-Sen University, The First Affiliated Hospital of Shantou University Medical College, and The First People's Hospital of Foshan City) in China. Disease progression referred to the development of penetrating or stricturing diseases or the requirement for CD-related surgeries during follow-up. A total of 1130 radiomics features were extracted from VAT on CT in the training cohort, and a machine-learning-based VAT-RM was developed to predict disease progression using selected reproducible features and validated in an external test cohort. Using the same modeling methodology, a SAT-RM was developed and compared with the VAT-RM. Findings: The VAT-RM exhibited satisfactory performance for predicting disease progression in total test cohort (the area under the ROC curve [AUC] = 0.850, 95% confidence Interval [CI] 0.764-0.913, P < 0.001) and in test cohorts 1 (AUC = 0.820, 95% CI 0.687-0.914, P < 0.001) and 2 (AUC = 0.871, 95% CI 0.744-0.949, P < 0.001). No significant differences in AUC were observed between test cohorts 1 and 2 (P = 0.673), suggesting considerable efficacy and robustness of the VAT-RM. In the total test cohort, the AUC of the VAT-RM for predicting disease progression was higher than that of SAT-RM (AUC = 0.786, 95% CI 0.692-0.861, P < 0.001). On multivariate Cox regression analysis, the VAT-RM (hazard ratio [HR] = 9.285, P = 0.005) was the most important independent predictor, followed by the SAT-RM (HR = 3.280, P = 0.060). Decision curve analysis further confirmed the better net benefit of the VAT-RM than the SAT-RM. Moreover, the SAT-RM failed to significantly improve predictive efficacy after it was added to the VAT-RM (integrated discrimination improvement = 0.031, P = 0.102). Interpretation: Our results suggest that VAT is an important determinant of disease progression in patients with CD. Our VAT-RM allows the accurate identification of high-risk patients prone to disease progression and offers notable advantages over SAT-RM. Funding: This study was supported by the National Natural Science Foundation of China, Guangdong Basic and Applied Basic Research Foundation, Shenzhen-Hong Kong Institute of Brain Science-Shenzhen Fundamental Research Institutions, Nature Science Foundation of Shenzhen, and Young S&T Talent Training Program of Guangdong Provincial Association for S&T. Translation: For the Chinese translation of the abstract see Supplementary Materials section.