Immunotherapy for recurrent or metastatic nasopharyngeal carcinoma.

Immunotherapy, particularly immune checkpoint inhibitors (ICIs), such as anti-programmed death 1/programmed death-ligand 1 (PD-1/PD-L1) therapy, has emerged as a pivotal treatment modality for solid tumors, including recurrent or metastatic nasopharyngeal carcinoma (R/M-NPC). Despite the advancements in the utilization of ICIs, there is still room for further improving patient outcomes. Another promising approach to immunotherapy for R/M-NPC involves adoptive cell therapy (ACT), which aims to stimulate systemic anti-tumor immunity. However, individual agent therapies targeting dendritic cells (DCs) appear to still be in the clinical trial phase. This current review underscores the potential of immunotherapy as a valuable adjunct to the treatment paradigm for R/M-NPC patients. Further research is warranted to enhance the efficacy of immunotherapy through the implementation of strategies such as combination therapies and overcoming immune suppression. Additionally, the development of a biomarker-based scoring system is essential for identifying suitable candidates for precision immunotherapy.

Automated and reusable deep learning (AutoRDL) framework for predicting response to neoadjuvant chemotherapy and axillary lymph node metastasis in breast cancer using ultrasound images: a retrospective, multicentre study.

Background: Previous deep learning models have been proposed to predict the pathological complete response (pCR) and axillary lymph node metastasis (ALNM) in breast cancer. Yet, the models often leveraged multiple frameworks, required manual annotation, and discarded low-quality images. We aimed to develop an automated and reusable deep learning (AutoRDL) framework for tumor detection and prediction of pCR and ALNM using ultrasound images with diverse qualities. Methods: The AutoRDL framework includes a You Only Look Once version 5 (YOLOv5) network for tumor detection and a progressive multi-granularity (PMG) network for pCR and ALNM prediction. The training cohort and the internal validation cohort were recruited from Guangdong Provincial People's Hospital (GPPH) between November 2012 and May 2021. The two external validation cohorts were recruited from the First Affiliated Hospital of Kunming Medical University (KMUH), between January 2016 and December 2019, and Shunde Hospital of Southern Medical University (SHSMU) between January 2014 and July 2015. Prior to model training, super-resolution via iterative refinement (SR3) was employed to improve the spatial resolution of low-quality images from the KMUH. We developed three models for predicting pCR and ALNM: a clinical model using multivariable logistic regression analysis, an image model utilizing the PMG network, and a combined model that integrates both clinical and image data using the PMG network. Findings: The YOLOv5 network demonstrated excellent accuracy in tumor detection, achieving average precisions of 0.880-0.921 during validation. In terms of pCR prediction, the combined modelpost-SR3 outperformed the combined modelpre-SR3, image modelpost-SR3, image modelpre-SR3, and clinical model (AUC: 0.833 vs 0.822 vs 0.806 vs 0.790 vs 0.712, all p < 0.05) in the external validation cohort (KMUH). Consistently, the combined modelpost-SR3 exhibited the highest accuracy in ALNM prediction, surpassing the combined modelpre-SR3, image modelpost-SR3, image modelpre-SR3, and clinical model (AUC: 0.825 vs 0.806 vs 0.802 vs 0.787 vs 0.703, all p < 0.05) in the external validation cohort 1 (KMUH). In the external validation cohort 2 (SHSMU), the combined model also showed superiority over the clinical and image models (0.819 vs 0.712 vs 0.806, both p < 0.05). Interpretation: Our proposed AutoRDL framework is feasible in automatically predicting pCR and ALNM in real-world settings, which has the potential to assist clinicians in optimizing individualized treatment options for patients. Funding: National Key Research and Development Program of China (2023YFF1204600); National Natural Science Foundation of China (82227802, 82302306); Clinical Frontier Technology Program of the First Affiliated Hospital of Jinan University, China (JNU1AF-CFTP-2022-a01201); Science and Technology Projects in Guangzhou (202201020022, 2023A03J1036, 2023A03J1038); Science and Technology Youth Talent Nurturing Program of Jinan University (21623209); and Postdoctoral Science Foundation of China (2022M721349).

Image-based artificial intelligence for the prediction of pathological complete response to neoadjuvant chemoradiotherapy in patients with rectal cancer: a systematic review and meta-analysis.

OBJECTIVE: Artificial intelligence (AI) holds enormous potential for noninvasively identifying patients with rectal cancer who could achieve pathological complete response (pCR) following neoadjuvant chemoradiotherapy (nCRT). We aimed to conduct a meta-analysis to summarize the diagnostic performance of image-based AI models for predicting pCR to nCRT in patients with rectal cancer. METHODS: This study followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. A literature search of PubMed, Embase, Cochrane Library, and Web of Science was performed from inception to July 29, 2023. Studies that developed or utilized AI models for predicting pCR to nCRT in rectal cancer from medical images were included. The Quality Assessment of Diagnostic Accuracy Studies-AI was used to appraise the methodological quality of the studies. The bivariate random-effects model was used to summarize the individual sensitivities, specificities, and areas-under-the-curve (AUCs). Subgroup and meta-regression analyses were conducted to identify potential sources of heterogeneity. Protocol for this study was registered with PROSPERO (CRD42022382374). RESULTS: Thirty-four studies (9933 patients) were identified. Pooled estimates of sensitivity, specificity, and AUC of AI models for pCR prediction were 82% (95% CI: 76-87%), 84% (95% CI: 79-88%), and 90% (95% CI: 87-92%), respectively. Higher specificity was seen for the Asian population, low risk of bias, and deep-learning, compared with the non-Asian population, high risk of bias, and radiomics (all P < 0.05). Single-center had a higher sensitivity than multi-center (P = 0.001). The retrospective design had lower sensitivity (P = 0.012) but higher specificity (P < 0.001) than the prospective design. MRI showed higher sensitivity (P = 0.001) but lower specificity (P = 0.044) than non-MRI. The sensitivity and specificity of internal validation were higher than those of external validation (both P = 0.005). CONCLUSIONS: Image-based AI models exhibited favorable performance for predicting pCR to nCRT in rectal cancer. However, further clinical trials are warranted to verify the findings.

An interpretable machine learning model based on contrast-enhanced CT parameters for predicting treatment response to conventional transarterial chemoembolization in patients with hepatocellular carcinoma.

OBJECTIVE: To explore the potential of pre-therapy computed tomography (CT) parameters in predicting the treatment response to initial conventional TACE (cTACE) in intermediate-stage hepatocellular carcinoma (HCC) and develop an interpretable machine learning model. METHODS: This retrospective study included 367 patients with intermediate-stage HCC who received cTACE as first-line therapy from three centers. We measured the mean attenuation values of target lesions on multi-phase contrast-enhanced CT and further calculated three CT parameters, including arterial (AER), portal venous (PER), and arterial portal venous (APR) enhancement ratios. We used logistic regression analysis to select discriminative features and trained three machine learning models via 5-fold cross-validation. The performance in predicting treatment response was evaluated in terms of discrimination, calibration, and clinical utility. Afterward, a Shapley additive explanation (SHAP) algorithm was leveraged to interpret the outputs of the best-performing model. RESULTS: The mean diameter, ECOG performance status, and cirrhosis were the important clinical predictors of cTACE treatment response, by multiple logistic regression. Adding the CT parameters to clinical variables showed significant improvement in performance (net reclassification index, 0.318, P < 0.001). The Random Forest model (hereafter, RF-combined model) integrating CT parameters and clinical variables demonstrated the highest performance on external validation dataset (AUC of 0.800). The decision curve analysis illustrated the optimal clinical benefits of RF-combined model. This model could successfully stratify patients into responders and non-responders with distinct survival (P = 0.001). CONCLUSION: The RF-combined model can serve as a robust and interpretable tool to identify the appropriate crowd for cTACE sessions, sparing patients from receiving ineffective and unnecessary treatments.

Development and Validation of a Feature-Based Broad-Learning System for Opportunistic Osteoporosis Screening Using Lumbar Spine Radiographs.

RATIONALE AND OBJECTIVES: Osteoporosis is primarily diagnosed using dual-energy X-ray absorptiometry (DXA); yet, DXA is significantly underutilized, causing osteoporosis, an underdiagnosed condition. We aimed to provide an opportunistic approach to screen for osteoporosis using artificial intelligence based on lumbar spine X-ray radiographs. MATERIALS AND METHODS: In this institutional review board-approved retrospective study, female patients aged ≥50 years who received both X-ray scans and DXA of the lumbar vertebrae, in three centers, were included. A total of 1180 cases were used for training and 145 cases were used for testing. We proposed a novel broad-learning system (BLS) and then compared the performance of BLS models using radiomic features and deep features as a source of input. The deep features were extracted using ResNet18 and VGG11, respectively. The diagnostic performances of these BLS models were evaluated with the area under the curve (AUC), sensitivity, and specificity. RESULTS: The incidence rate of osteoporosis in the training and test sets was 35.9% and 37.9%, respectively. The radiomic feature-based BLS model achieved higher testing AUC (0.802 vs. 0.654 vs. 0.632, both P = .002), sensitivity (78.2% vs. 56.4% vs. 50.9%), and specificity (82.2% vs. 74,4% vs. 75.6%) than the two deep feature-based BLS models. CONCLUSION: Our proposed radiomic feature-based BLS model has the potential to expand osteoporosis screening to a broader population by identifying osteoporosis on lumbar spine X-ray radiographs.

Prognostic significance of MRI-defined sarcopenia in patients with nasopharyngeal carcinoma: A propensity score matched analysis of real-world data.

BACKGROUND AND PURPOSE: Image-defined sarcopenia is linked to increased mortality among patients with cancer. Nevertheless, its effect on patients with nasopharyngeal carcinoma (NPC) is incompletely established. This study's aim was to investigate the prognostic significance of MRI-defined sarcopenia on the survival of patients undergoing concurrent chemoradiotherapy (CCRT) ± inducing chemotherapy (IC) for NPC treatment. METHODS: 1,307 patients with stage II-IVa NPC were included in this retrospective study. Sarcopenia was defined using skeletal muscle index (SMI) determined through baseline MRI at the C3 level. The association of sarcopenia with overall survival (OS) and progression-free survival (PFS) was assessed by Cox regression models using 1:1 propensity score matching (PSM) analysis. We also conducted a stratification analysis using BMI and treatment strategies. RESULTS: Sarcopenia was an independent risk factor for both OS and PFS (all P < 0.05). However, BMI was not substantially linked to OS and PFS (all P > 0.05). Sarcopenic patients showed lower rates of OS (HR = 2.00, 95% CI: 1.54-2.60, P < 0.001) and PFS (HR = 1.67, 95% CI: 1.35-2.07, P < 0.001) in contrast with nonsarcopenic patients. According to stratification analysis, being overweight was linked to a protective effect in nonsarcopenic patients only. Sarcopenic patients showed similar OS and PFS regardless of the treatment modality. CONCLUSIONS: Sarcopenia is underrecognized in NPC patients. Measurement of sarcopenia using routine MRI scans in NPC patients provided significant prognostic information, outperforming BMI. Patients with sarcopenia failed to benefit from an additional IC regimen.

Polyphyllin D punctures hypertrophic lysosomes to reverse drug resistance of hepatocellular carcinoma by targeting acid sphingomyelinase.

Hypertrophic lysosomes are critical for tumor progression and drug resistance; however, effective and specific lysosome-targeting compounds for cancer therapy are lacking. Here we conducted a lysosomotropic pharmacophore-based in silico screen in a natural product library (2,212 compounds), and identified polyphyllin D (PD) as a novel lysosome-targeted compound. PD treatment was found to cause lysosomal damage, as evidenced by the blockade of autophagic flux, loss of lysophagy, and the release of lysosomal contents, thus exhibiting anticancer effects on hepatocellular carcinoma (HCC) cell both in vitro and in vivo. Closer mechanistic examination revealed that PD suppressed the activity of acid sphingomyelinase (SMPD1), a lysosomal phosphodieserase that catalyzes the hydrolysis of sphingomyelin to produce ceramide and phosphocholine, by directly occupying its surface groove, with Trp148 in SMPD1 acting as a major binding residue; this suppression of SMPD1 activity irreversibly triggers lysosomal injury and initiates lysosome-dependent cell death. Furthermore, PD-enhanced lysosomal membrane permeabilization to release sorafenib, augmenting the anticancer effect of sorafenib both in vivo and in vitro. Overall, our study suggests that PD can potentially be further developed as a novel autophagy inhibitor, and a combination of PD with classical chemotherapeutic anticancer drugs could represent a novel therapeutic strategy for HCC intervention.

Comparative Study of C-TIRADS, ACR-TIRADS, and EU-TIRADS for Diagnosis and Management of Thyroid Nodules.

RATIONALE AND OBJECTIVES: Chinese Thyroid Imaging Reporting and Data Systems (C-TIRADS) was developed to provide a more simplified tool for stratifying thyroid nodules. Here we aimed to validate the efficacy of C-TIRADS in distinguishing benign from malignant and in guiding fine-needle aspiration biopsies in comparison with the American College of Radiology TIRADS (ACR-TIRADS) and European TIRADS (EU-TIRADS). MATERIALS AND METHODS: This study retrospectively included 3438 thyroid nodules (≥10 mm) in 3013 patients (mean age, 47.1 years ± 12.9) diagnosed between January 2013 and November 2019. Ultrasound features of the nodules were evaluated and categorized according to the lexicons of the three TIRADS. We compared these TIRADS by using the area under the receiver operating characteristic curve (AUROC), the area under the precision-recall curve (AUPRC), sensitivity, specificity, net reclassification improvement (NRI), and unnecessary fine-needle aspiration biopsy (FNAB) rate. RESULTS: Of the 3438 thyroid nodules, 707 (20.6%) were malignant. C-TIRADS showed higher discrimination performance (AUROC, 0.857; AUPRC, 0.605) than ACR-TIRADS (AUROC, 0.844; AUPRC, 0.567) and EU-TIRADS (AUROC, 0.802; AUPRC, 0.455). The sensitivity of C-TIRADS (85.3%) was lower than that of ACR-TIRADS (89.1%) but higher than that of EU-TIRADS (78.4%). The specificity of C-TIRADS (76.9%) was similar to that of EU-TIRADS (78.9%) and higher than that of ACR-TIRADS (69.5%). The unnecessary FNAB rate was lowest with C-TIRADS (21.2%), followed by ACR-TIRADS (41.7%) and EU-TIRADS (58.3%). C-TIRADS obtained significant NRI for recommending FNAB over ACR-TIRADS (19.0%, P < 0.001) and EU-TIRADS (25.5%, P < 0.001). CONCLUSION: C-TIRADS may be a clinically applicable tool to manage thyroid nodules, which warrants thorough tests in other geographic settings.

Diagnostic Performance of Magnetic Resonance Angiography for Artery Stenosis After Kidney Transplant: A Systematic Review and Meta-Analysis.

RATIONALE AND OBJECTIVES: Magnetic resonance angiography (MRA) is used to diagnose artery stenosis after kidney transplant. However, there is a lack of applicable consensus guidelines, and the diagnostic value of this technique is unclear. Therefore, the aim of the present study was to evaluate the diagnostic performance of MRA for the detection of artery stenosis after kidney transplant. MATERIALS AND METHODS: We searched PubMed, Web of Science, Cochrane Library, and Embase from database inception to September 1, 2022. Two independent reviewers assessed the methodological quality of eligible studies using the quality assessment of diagnostic accuracy studies-2 tool. The diagnostic odds ratio, pooled sensitivity, and specificity values, positive likelihood ratios, and negative likelihood ratios were calculated to synthesize data with a bivariate random-effects model. Meta-regression analysis was performed in cases of high among-study heterogeneity. RESULTS: Eleven studies were included in the meta-analysis. The area under the summary receiver operating characteristic curve was 0.96 (95% confidence interval [CI]: 0.94-0.98). The pooled sensitivity and specificity values for MRA in diagnosing artery stenosis after kidney transplant were 0.96 (95% CI: 0.76-0.99) and 0.93 (95% CI: 0.86-0.96), respectively. CONCLUSION: MRA demonstrated high sensitivity and specificity for diagnosing artery stenosis after kidney transplant, suggesting that it may be used reliably in clinical practice. However, further large-scale studies are required to validate the present findings.

Integrative Scoring System for Survival Prediction in Patients With Locally Advanced Nasopharyngeal Carcinoma: A Retrospective Multicenter Study.

PURPOSE: Tumor stage is crucial for prognostic evaluation and therapeutic decisions in locally advanced nasopharyngeal carcinoma (NPC) but is imprecise. We aimed to propose a new prognostic system by integrating quantitative imaging features and clinical factors. MATERIALS AND METHODS: This retrospective study included 1,319 patients with stage III-IVa NPC between April 1, 2010, and July 31, 2019, who underwent pretherapy magnetic resonance imaging (MRI) and received concurrent chemoradiotherapy with or without induction chemotherapy. The hand-crafted and deep-learned features were extracted from MRI for each patient. After feature selection, the clinical score, radiomic score, deep score, and integrative scores were constructed via Cox regression analysis. The scores were validated in two external cohorts. The predictive accuracy and discrimination were measured by the area under the curve (AUC) and risk group stratification. The end points were progression-free survival (PFS), overall survival (OS), and distant metastasis-free survival (DMFS). RESULTS: Both radiomics and deep learning were complementary to clinical variables (age, T stage, and N stage; all P < .05). The clinical-deep score was superior or equivalent to clinical-radiomic score, whereas it was noninferior to clinical-radiomic-deep score (all P > .05). These findings were also verified in the evaluation of OS and DMFS. The clinical-deep score yielded an AUC of 0.713 (95% CI, 0.697 to 0.729) and 0.712 (95% CI, 0.693 to 0.731) in the two external validation cohorts for predicting PFS with good calibration. This scoring system could stratify patients into high- and low-risk groups with distinct survivals (all P < .05). CONCLUSION: We established and validated a prognostic system integrating clinical data and deep learning to provide an individual prediction of survival for patients with locally advanced NPC, which might inform clinicians in treatment decision making.

Efficacy and safety of adjuvant chemotherapy for locally advanced cervical cancer: A systematic review and meta-analysis.

BACKGROUND: Locally advanced cervical cancer (LACC) is generally treated using concurrent chemo-radiotherapy (CCRT); yet, the effectiveness of adjuvant chemotherapy (ACT) following CCRT remains controversial. METHODS: The databases Embase, Web of Science, and PubMed were analyzed for relevant research. Primary endpoints included overall survival (OS) and progression-free survival (PFS). RESULTS: Fifteen trials with 4041 patients were included. Pooled HRs for PFS and OS were 0.81 (95 % CI: 0.67-0.96) and 0.69 (95 % CI: 0.51-0.93), respectively. However, subgroup analyses indicated that in randomized trials and trials with larger sample sizes (n > 100) as well as ACT cycles ≤ 3, ACT was not linked with improved PFS and OS. Moreover, ACT induced a greater rate of hematologic toxicities (P < 0.05). CONCLUSION: Higher quality of evidence suggests that ACT could not yield additional survival benefits for LACC; however, identifying high-risk patients who may benefit from ACT is required to design further clinical trials and better inform treatment decisions.

MRI texture-based machine learning models for the evaluation of renal function on different segmentations: a proof-of-concept study.

BACKGROUND: To develop and validate an MRI texture-based machine learning model for the noninvasive assessment of renal function. METHODS: A retrospective study of 174 diabetic patients (training cohort, n = 123; validation cohort, n = 51) who underwent renal MRI scans was included. They were assigned to normal function (n = 71), mild or moderate impairment (n = 69), and severe impairment groups (n = 34) according to renal function. Four methods of kidney segmentation on T2-weighted images (T2WI) were compared, including regions of interest covering all coronal slices (All-K), the largest coronal slices (LC-K), and subregions of the largest coronal slices (TLCO-K and PIZZA-K). The speeded-up robust features (SURF) and support vector machine (SVM) algorithms were used for texture feature extraction and model construction, respectively. Receiver operating characteristic (ROC) curve analysis was used to evaluate the diagnostic performance of models. RESULTS: The models based on LC-K and All-K achieved the nonsignificantly highest accuracy in the classification of renal function (all p values > 0.05). The optimal model yielded high performance in classifying the normal function, mild or moderate impairment, and severe impairment, with an area under the curve of 0.938 (95% confidence interval [CI] 0.935-0.940), 0.919 (95%CI 0.916-0.922), and 0.959 (95%CI 0.956-0.962) in the training cohorts, respectively, as well as 0.802 (95%CI 0.800-0.807), 0.852 (95%CI 0.846-0.857), and 0.863 (95%CI 0.857-0.887) in the validation cohorts, respectively. CONCLUSION: We developed and internally validated an MRI-based machine-learning model that can accurately evaluate renal function. Once externally validated, this model has the potential to facilitate the monitoring of patients with impaired renal function.

Turning gray selenium and sublimed sulfur into a nanocomposite to accelerate tissue regeneration by isothermal recrystallization.

BACKGROUND: Globally, millions of patients suffer from regenerative deficiencies, such as refractory wound healing, which is characterized by excessive inflammation and abnormal angiogenesis. Growth factors and stem cells are currently employed to accelerate tissue repair and regeneration; however, they are complex and costly. Thus, the exploration of new regeneration accelerators is of considerable medical interest. This study developed a plain nanoparticle that accelerates tissue regeneration with the involvement of angiogenesis and inflammatory regulation. METHODS: Grey selenium and sublimed sulphur were thermalized in PEG-200 and isothermally recrystallised to composite nanoparticles (Nano-Se@S). The tissue regeneration accelerating activities of Nano-Se@S were evaluated in mice, zebrafish, chick embryos, and human cells. Transcriptomic analysis was performed to investigate the potential mechanisms involved during tissue regeneration. RESULTS: Through the cooperation of sulphur, which is inert to tissue regeneration, Nano-Se@S demonstrated improved tissue regeneration acceleration activity compared to Nano-Se. Transcriptome analysis revealed that Nano-Se@S improved biosynthesis and ROS scavenging but suppressed inflammation. The ROS scavenging and angiogenesis-promoting activities of Nano-Se@S were further confirmed in transgenic zebrafish and chick embryos. Interestingly, we found that Nano-Se@S recruits leukocytes to the wound surface at the early stage of regeneration, which contributes to sterilization during regeneration. CONCLUSION: Our study highlights Nano-Se@S as a tissue regeneration accelerator, and Nano-Se@S may provide new inspiration for therapeutics for regenerative-deficient diseases.

Proteome-Wide Analysis Reveals TFEB Targets for Establishment of a Prognostic Signature to Predict Clinical Outcomes of Colorectal Cancer.

Dephosphorylation of transcription factor EB (TFEB) at Ser142 and Ser138 determines its nuclear localization and transcriptional activity. The link between TFEB-associated genes and colorectal cancer (CRC) progression and prognosis remains unclear. To systematically identify the targets of TFEB, we performed data-independent acquisition (DIA)-based quantitative proteomics to compare global protein changes in wild-type (WT) DLD1 cells and TFEBWT- or TFEBS142A/S138A (activated status)-expressing DLD1 cells. A total of 6048 proteins were identified and quantified in three independent experiments. The differentially expressed proteins in TFEBS142A/S138A versus TFEBWT and TFEBWT versus control groups were compared, and 60 proteins were identified as products of TFEB transcriptional regulation. These proteins were significantly associated with vesicular endocytic trafficking, the HIF-1 signaling pathway, and metabolic processes. Furthermore, we generated a TFEB-associated gene signature using a univariate and LASSO Cox regression model to screen robust prognostic markers. An eight-gene signature (PLSCR3, SERPINA1, ATP6V1C2, TIMP1, SORT1, MAP2, KDM4B, and DDAH2) was identified. According to the signature, patients were assigned to high-risk and low-risk groups. Higher risk scores meant worse overall survival and higher epithelial-mesenchymal transition (EMT) scores. Additionally, as per the clinicopathological parameters and gene signature, a nomogram was constructed that was utilized to enhance the quantification capacity in risk assessment for individual patients. This research shows that TFEB directly mediates network effects in CRC, and the identified TFEB gene signature-based model may provide important information for the clinical judgment of prognosis.

Deep learning nomogram based on Gd-EOB-DTPA MRI for predicting early recurrence in hepatocellular carcinoma after hepatectomy.

OBJECTIVES: The accurate prediction of post-hepatectomy early recurrence in patients with hepatocellular carcinoma (HCC) is crucial for decision-making regarding postoperative adjuvant treatment and monitoring. We aimed to explore the feasibility of deep learning (DL) features derived from gadoxetate disodium (Gd-EOB-DTPA) MRI, qualitative features, and clinical variables for predicting early recurrence. METHODS: In this bicentric study, 285 patients with HCC who underwent Gd-EOB-DTPA MRI before resection were divided into training (n = 195) and validation (n = 90) sets. DL features were extracted from contrast-enhanced MRI images using VGGNet-19. Three feature selection methods and five classification methods were combined for DL signature construction. Subsequently, an mp-MR DL signature fused with multiphase DL signatures of contrast-enhanced images was constructed. Univariate and multivariate logistic regression analyses were used to identify early recurrence risk factors including mp-MR DL signature, microvascular invasion (MVI), and tumor number. A DL nomogram was built by incorporating deep features and significant clinical variables to achieve early recurrence prediction. RESULTS: MVI (p = 0.039), tumor number (p = 0.001), and mp-MR DL signature (p < 0.001) were independent risk factors for early recurrence. The DL nomogram outperformed the clinical nomogram in the training set (AUC: 0.949 vs. 0.751; p < 0.001) and validation set (AUC: 0.909 vs. 0.715; p = 0.002). Excellent DL nomogram calibration was achieved in both training and validation sets. Decision curve analysis confirmed the clinical usefulness of DL nomogram. CONCLUSION: The proposed DL nomogram was superior to the clinical nomogram in predicting early recurrence for HCC patients after hepatectomy. KEY POINTS: • Deep learning signature based on Gd-EOB-DTPA MRI was the predominant independent predictor of early recurrence for hepatocellular carcinoma (HCC) after hepatectomy. • Deep learning nomogram based on clinical factors and Gd-EOB-DTPA MRI features is promising for predicting early recurrence of HCC. • Deep learning nomogram outperformed the conventional clinical nomogram in predicting early recurrence.

Machine-learning methods based on the texture and non-texture features of MRI for the preoperative prediction of sentinel lymph node metastasis in breast cancer.

Background: The establishment of an accurate, stable, and non-invasive prediction model of sentinel lymph node (SLN) metastasis in breast cancer is difficult nowadays. The aim of this work is to identify the optimal machine learning model based on the three-dimensional (3D) image features of magnetic resonance imaging (MRI) for the preoperative prediction of SLN metastasis in breast cancer patients. Methods: A total of 172 patients with histologically proven breast cancer were enrolled retrospectively, including 74 SLN metastasis patients and 98 non-SLN metastasis patients. All of them underwent diffusion-weighted imaging (DWI) magnetic resonance imaging (MRI) scan. Firstly, a total of 10,320 texture and four non-texture features were extracted from the region of interests (ROIs) of image. Twenty-four feature selection methods and 11 classification methods were then evaluated by using 10-fold cross-validation to identify the optimal machine learning model in terms of the mean area under the curve (AUC), accuracy (ACC), and stability. Results: The result showed that the model based on the combination of minimum redundancy maximum relevance (MRMR) + random forest (RF) exhibited the optimal predictive performance (AUC: 0.97±0.03; ACC: 0.89±0.05; stability: 2.94). Moreover, we independently investigated the performance of feature selection methods and classification methods, and observed that L1-support vector machine (L1-SVM) (AUC: 0.80±0.08; ACC: 0.76±0.07) and sequential forward floating selection (SFFS) (stability: 3.04) presented the best average predictive performance and stability among all feature selection methods, respectively. RF (AUC: 0.85±0.11; ACC: 0.80±0.09) and SVM (stability: 8.43) showed the best average predictive performance and stability among all classification methods, respectively. Conclusions: The identified model based on the 3D image features of MRI provides a non-invasive way for the preoperative prediction of SLN metastasis in breast cancer patients.

Traumatic bone marrow lesions in dual-energy computed tomography.

Traumatic bone marrow lesions (TBMLs) are considered to represent a range of concealed bone injuries, including haemorrhage, infarction, and localised oedema caused by trabecular microfracture occurring in the cancellous bone. If TBMLs are not managed timeously, they potentially cause a series of complications that can lead to irreversible morbidity and prolonged recovery time. This article reviews interesting image findings of bone marrow lesions in dual-energy computed tomography (DECT). In addition to combining the benefits of traditional CT imaging, DECT also reveals and identifies various structures using diverse attenuation characteristics of different radiographic spectra. Therefore, DECT has the capacity to detect TBMLs, which have traditionally been diagnosed using MRI. Through evaluating DECT virtual non-calcium maps, the detection of TBMLs is rendered easier and more efficient in some acute accidents.

Thy-Wise: An interpretable machine learning model for the evaluation of thyroid nodules.

Current risk stratification systems for thyroid nodules suffer from low specificity and high biopsy rates. Recently, machine learning (ML) is introduced to assist thyroid nodule diagnosis but lacks interpretability. Here, we developed and validated ML models on 3965 thyroid nodules, as compared to the American College of Radiology Thyroid Imaging, Reporting and Data System (ACR TI-RADS). Subsequently, a SHapley Additive exPlanation (SHAP) algorithm was leveraged to interpret the results of the best-performing ML model. Clinical characteristics including thyroid-function tests were collected from medical records. Five ACR TI-RADS ultrasonography (US) categories plus nodule size were assessed by experienced radiologists. Random forest (RF), support vector machine (SVM) and extreme gradient boosting (XGBoost) were used to build US-only and US-clinical ML models. The ML models and ACR TI-RADS were compared in terms of diagnostic performance and unnecessary biopsy rate. Among the ML models, the US-only RF model (hereafter, Thy-Wise) achieved the optimal performance. Compared to ACR TI-RADS, Thy-Wise showed higher accuracy (82.4% vs 74.8% for the internal validation; 82.1% vs 73.4% for external validation) and specificity (78.7% vs 68.3% for internal validation; 78.5% vs 66.9% for external validation) while maintaining sensitivity (91.7% vs 91.2% for internal validation; 91.9% vs 91.1% for external validation), as well as reduced unnecessary biopsies (15.3% vs 32.3% for internal validation; 15.7% vs 47.3% for external validation). The SHAP-based interpretation of Thy-Wise enables clinicians to better understand the reasoning behind the diagnosis, which may facilitate the clinical translation of this model.