Learning Consistency and Specificity of Cells From Single-Cell Multi-Omic Data.

Advancements in single-cell technologies concomitantly develop the epigenomic and transcriptomic profiles at the cell levels, providing opportunities to explore the potential biological mechanisms. Even though significant efforts have been dedicated to them, it remains challenging for the integration analysis of multi-omic data of single-cell because of the heterogeneity, complicated coupling and interpretability of data. To handle these issues, we propose a novel self-representation Learning-based Multi-omics data Integrative Clustering algorithm (sLMIC) for the integration of single-cell epigenomic profiles (DNA methylation or scATAC-seq) and transcriptomic (scRNA-seq), which the consistent and specific features of cells are explicitly extracted facilitating the cell clustering. Specifically, sLMIC constructs a graph for each type of single-cell data, thereby transforming omics data into multi-layer networks, which effectively removes heterogeneity of omic data. Then, sLMIC employs the low-rank and exclusivity constraints to separate the self-representation of cells into two parts, i.e., the shared and specific features, which explicitly characterize the consistency and diversity of omic data, providing an effective strategy to model the structure of cell types. Feature extraction and cell clustering are jointly formulated as an overall objective function, where latent features of data are obtained under the guidance of cell clustering. The extensive experimental results on 13 multi-omics datasets of single-cell from diverse organisms and tissues indicate that sLMIC observably exceeds the advanced algorithms regarding various measurements.

Artificial intelligence for diagnosis and prognosis prediction of natural killer/T cell lymphoma using magnetic resonance imaging.

Accurate diagnosis and prognosis prediction are conducive to early intervention and improvement of medical care for natural killer/T cell lymphoma (NKTCL). Artificial intelligence (AI)-based systems are developed based on nasopharynx magnetic resonance imaging. The diagnostic systems achieve areas under the curve of 0.905-0.960 in detecting malignant nasopharyngeal lesions and distinguishing NKTCL from nasopharyngeal carcinoma in independent validation datasets. In comparison to human radiologists, the diagnostic systems show higher accuracies than resident radiologists and comparable ones to senior radiologists. The prognostic system shows promising performance in predicting survival outcomes of NKTCL and outperforms several clinical models. For patients with early-stage NKTCL, only the high-risk group benefits from early radiotherapy (hazard ratio = 0.414 vs. late radiotherapy; 95% confidence interval, 0.190-0.900, p = 0.022), while progression-free survival does not differ in the low-risk group. In conclusion, AI-based systems show potential in assisting accurate diagnosis and prognosis prediction and may contribute to therapeutic optimization for NKTCL.

FedDUS: Lung tumor segmentation on CT images through federated semi-supervised with dynamic update strategy.

BACKGROUND AND OBJECTIVE: Lung tumor annotation is a key upstream task for further diagnosis and prognosis. Although deep learning techniques have promoted automation of lung tumor segmentation, there remain challenges impeding its application in clinical practice, such as a lack of prior annotation for model training and data-sharing among centers. METHODS: In this paper, we use data from six centers to design a novel federated semi-supervised learning (FSSL) framework with dynamic model aggregation and improve segmentation performance for lung tumors. To be specific, we propose a dynamically updated algorithm to deal with model parameter aggregation in FSSL, which takes advantage of both the quality and quantity of client data. Moreover, to increase the accessibility of data in the federated learning (FL) network, we explore the FAIR data principle while the previous federated methods never involve. RESULT: The experimental results show that the segmentation performance of our model in six centers is 0.9348, 0.8436, 0.8328, 0.7776, 0.8870 and 0.8460 respectively, which is superior to traditional deep learning methods and recent federated semi-supervised learning methods. CONCLUSION: The experimental results demonstrate that our method is superior to the existing FSSL methods. In addition, our proposed dynamic update strategy effectively utilizes the quality and quantity information of client data and shows efficiency in lung tumor segmentation. The source code is released on (https://github.com/GDPHMediaLab/FedDUS).

Noninvasive Artificial Intelligence System for Early Predicting Residual Cancer Burden during Neoadjuvant Chemotherapy in Breast Cancer.

OBJECTIVE: To develop an artificial intelligence (AI) system for the early prediction of residual cancer burden (RCB) scores during neoadjuvant chemotherapy (NAC) in breast cancer. SUMMARY BACKGROUND DATA: RCB III indicates drug resistance in breast cancer, and early detection methods are lacking. METHODS: This study enrolled 1048 patients with breast cancer from four institutions, who were all receiving NAC. Magnetic resonance images were collected at the pre- and mid-NAC stages, and radiomics and deep learning features were extracted. A multitask AI system was developed to classify patients into three groups (RCB 0-I, II, and III ) in the primary cohort (PC, n=335). Feature selection was conducted using the Mann-Whitney U- test, Spearman analysis, least absolute shrinkage and selection operator regression, and the Boruta algorithm. Single-modality models were developed followed by model integration. The AI system was validated in three external validation cohorts. (EVCs, n=713). RESULTS: Among the patients, 442 (42.18%) were RCB 0-I, 462 (44.08%) were RCB II and 144 (13.74%) were RCB III. Model-I achieved an area under the curve (AUC) of 0.975 in the PC and 0.923 in the EVCs for differentiating RCB III from RCB 0-II. Model-II distinguished RCB 0-I from RCB II-III, with an AUC of 0.976 in the PC and 0.910 in the EVCs. Subgroup analysis confirmed that the AI system was consistent across different clinical T stages and molecular subtypes. CONCLUSIONS: The multitask AI system offers a noninvasive tool for the early prediction of RCB scores in breast cancer, supporting clinical decision-making during NAC.

A Colorectal Coordinate-Driven Method for Colorectum and Colorectal Cancer Segmentation in Conventional CT Scans.

Automated colorectal cancer (CRC) segmentation in medical imaging is the key to achieving automation of CRC detection, staging, and treatment response monitoring. Compared with magnetic resonance imaging (MRI) and computed tomography colonography (CTC), conventional computed tomography (CT) has enormous potential because of its broad implementation, superiority for the hollow viscera (colon), and convenience without needing bowel preparation. However, the segmentation of CRC in conventional CT is more challenging due to the difficulties presenting with the unprepared bowel, such as distinguishing the colorectum from other structures with similar appearance and distinguishing the CRC from the contents of the colorectum. To tackle these challenges, we introduce DeepCRC-SL, the first automated segmentation algorithm for CRC and colorectum in conventional contrast-enhanced CT scans. We propose a topology-aware deep learning-based approach, which builds a novel 1-D colorectal coordinate system and encodes each voxel of the colorectum with a relative position along the coordinate system. We then induce an auxiliary regression task to predict the colorectal coordinate value of each voxel, aiming to integrate global topology into the segmentation network and thus improve the colorectum's continuity. Self-attention layers are utilized to capture global contexts for the coordinate regression task and enhance the ability to differentiate CRC and colorectum tissues. Moreover, a coordinate-driven self-learning (SL) strategy is introduced to leverage a large amount of unlabeled data to improve segmentation performance. We validate the proposed approach on a dataset including 227 labeled and 585 unlabeled CRC cases by fivefold cross-validation. Experimental results demonstrate that our method outperforms some recent related segmentation methods and achieves the segmentation accuracy in DSC for CRC of 0.669 and colorectum of 0.892, reaching to the performance (at 0.639 and 0.890, respectively) of a medical resident with two years of specialized CRC imaging fellowship.

Development and Validation of a Deep Learning Model to Reduce the Interference of Rectal Artifacts in MRI-based Prostate Cancer Diagnosis.

Purpose To develop an MRI-based model for clinically significant prostate cancer (csPCa) diagnosis that can resist rectal artifact interference. Materials and Methods This retrospective study included 2203 male patients with prostate lesions who underwent biparametric MRI and biopsy between January 2019 and June 2023. Targeted adversarial training with proprietary adversarial samples (TPAS) strategy was proposed to enhance model resistance against rectal artifacts. The automated csPCa diagnostic models trained with and without TPAS were compared using multicenter validation datasets. The impact of rectal artifacts on the diagnostic performance of each model at the patient and lesion levels was compared using the area under the receiver operating characteristic curve (AUC) and the area under the precision-recall curve (AUPRC). The AUC between models was compared using the DeLong test, and the AUPRC was compared using the bootstrap method. Results The TPAS model exhibited diagnostic performance improvements of 6% at the patient level (AUC: 0.87 vs 0.81, P < .001) and 7% at the lesion level (AUPRC: 0.84 vs 0.77, P = .007) compared with the control model. The TPAS model demonstrated less performance decline in the presence of rectal artifact-pattern adversarial noise than the control model (ΔAUC: -17% vs -19%, ΔAUPRC: -18% vs -21%). The TPAS model performed better than the control model in patients with moderate (AUC: 0.79 vs 0.73, AUPRC: 0.68 vs 0.61) and severe (AUC: 0.75 vs 0.57, AUPRC: 0.69 vs 0.59) artifacts. Conclusion This study demonstrates that the TPAS model can reduce rectal artifact interference in MRI-based csPCa diagnosis, thereby improving its performance in clinical applications. Keywords: MR-Diffusion-weighted Imaging, Urinary, Prostate, Comparative Studies, Diagnosis, Transfer Learning Clinical trial registration no. ChiCTR23000069832 Supplemental material is available for this article. Published under a CC BY 4.0 license.

Development and validation of a computed tomography-based radiomics signature to predict "highest-risk" from patients with high-risk gastrointestinal stromal tumor.

Background: Some patients with high-risk gastrointestinal stromal tumor (GIST) experience disease progression after complete resection and adjuvant therapy. It is of great significance to distinguish these patients among those with high-risk GIST. Radiomics has been demonstrated as a promising tool to predict various tumors prognosis. Methods: From January 2006 to December 2018, a total of 100 high-risk GIST patients (training cohort: 60; validation cohort: 40) from Guangdong Provincial People's Hospital with preoperative enhanced computed tomography (CT) images were enrolled. The radiomics features were extracted and a risk score was built using least absolute shrinkage and selection operator-Cox model. The clinicopathological factors were analyzed and a nomogram was established with and without radiomics risk score. The concordance index (C-index), calibration plot, and decision curve analysis (DCA) were used to evaluate the performance of the radiomics nomograms. Results: We selected 11 radiomics features associated with recurrence or metastasis. The risk score was calculated and significantly associated with disease-free survival (DFS) in both the training and validation group. Cox regression analysis showed that Ki67 was an independent risk factor for DFS [P=0.004, hazard ratio 4.615, 95% confidence interval (CI): 1.624-13.114]. The combined radiomics nomogram, which integrated the radiomics risk score and significant clinicopathological factors, showed good performance in predicting DFS, with a C-index of 0.832 (95% CI: 0.761-0.903), which was better than the clinical nomogram (C-index 0.769, 95% CI: 0.679-0.859) in training cohort. The calibration curves and the DCA plot suggested satisfying accuracy and clinical utility of the model. Conclusions: The CT-based radiomics nomogram, combined with the clinicopathological factors and risk score, has good potential to assess the recurrence or metastasis of patients with high-risk GIST.

A computed tomography-based multitask deep learning model for predicting tumour stroma ratio and treatment outcomes in patients with colorectal cancer: a multicentre cohort study.

BACKGROUND: Tumour-stroma interactions, as indicated by tumour-stroma ratio (TSR), offer valuable prognostic stratification information. Current histological assessment of TSR is limited by tissue accessibility and spatial heterogeneity. The authors aimed to develop a multitask deep learning (MDL) model to noninvasively predict TSR and prognosis in colorectal cancer (CRC). MATERIALS AND METHODS: In this retrospective study including 2268 patients with resected CRC recruited from four centres, the authors developed an MDL model using preoperative computed tomography (CT) images for the simultaneous prediction of TSR and overall survival. Patients in the training cohort ( n =956) and internal validation cohort (IVC, n =240) were randomly selected from centre I. Patients in the external validation cohort 1 (EVC1, n =509), EVC2 ( n =203), and EVC3 ( n =360) were recruited from other three centres. Model performance was evaluated with respect to discrimination and calibration. Furthermore, the authors evaluated whether the model could predict the benefit from adjuvant chemotherapy. RESULTS: The MDL model demonstrated strong TSR discrimination, yielding areas under the receiver operating curves (AUCs) of 0.855 (95% CI, 0.800-0.910), 0.838 (95% CI, 0.802-0.874), and 0.857 (95% CI, 0.804-0.909) in the three validation cohorts, respectively. The MDL model was also able to predict overall survival and disease-free survival across all cohorts. In multivariable Cox analysis, the MDL score (MDLS) remained an independent prognostic factor after adjusting for clinicopathological variables (all P <0.05). For stage II and stage III disease, patients with a high MDLS benefited from adjuvant chemotherapy [hazard ratio (HR) 0.391 (95% CI, 0.230-0.666), P =0.0003; HR=0.467 (95% CI, 0.331-0.659), P <0.0001, respectively], whereas those with a low MDLS did not. CONCLUSION: The multitask DL model based on preoperative CT images effectively predicted TSR status and survival in CRC patients, offering valuable guidance for personalized treatment. Prospective studies are needed to confirm its potential to select patients who might benefit from chemotherapy.

A novel stratification scheme combined with internal arteries in CT imaging for guiding postoperative adjuvant transarterial chemoembolization in hepatocellular carcinoma: a retrospective cohort study.

BACKGROUND: Although postoperative adjuvant transarterial chemoembolization (PA-TACE) improves survival outcomes in a subset of patients with resected hepatocellular carcinoma (HCC), the lack of reliable biomarkers for patient selection remains a significant challenge. The present study aimed to evaluate whether computed tomography imaging can provide more value for predicting benefits from PA-TACE and to establish a new scheme for guiding PA-TACE benefits. METHODS: In this retrospective study, patients with HCC who had undergone preoperative contrast-enhanced computed tomography and curative hepatectomy were evaluated. Inverse probability of treatment weight was performed to balance the difference of baseline characteristics. Cox models were used to test the interaction among PA-TACE, imaging features, and pathological indicators. An HCC imaging and pathological classification (HIPC) scheme incorporating these imaging and pathological indicators was established. RESULTS: This study included 1488 patients [median age, 52 years (IQR, 45-61 years); 1309 male]. Microvascular invasion (MVI) positive, and diameter >5 cm tumors achieved a higher recurrence-free survival (RFS), and overall survival (OS) benefit, respectively, from PA-TACE than MVI negative, and diameter ≤5 cm tumors. Patients with internal arteries (IA) positive benefited more than those with IA-negative in terms of RFS ( P =0.016) and OS ( P =0.018). PA-TACE achieved significant RFS and OS improvements in HIPC3 (IA present and diameter >5 cm, or two or three tumors) patients but not in HIPC1 (diameter ≤5 cm, MVI negative) and HIPC2 (other single tumor) patients. Our scheme may decrease the number of patients receiving PA-TACE by ~36.5% compared to the previous suggestion. CONCLUSIONS: IA can provide more value for predicting the benefit of PA-TACE treatment. The proposed HIPC scheme can be used to stratify patients with and without survival benefits from PA-TACE.

Artificial intelligence-based analysis of tumor-infiltrating lymphocyte spatial distribution for colorectal cancer prognosis.

BACKGROUND: Artificial intelligence (AI) technology represented by deep learning has made remarkable achievements in digital pathology, enhancing the accuracy and reliability of diagnosis and prognosis evaluation. The spatial distribution of CD3 + and CD8 + T cells within the tumor microenvironment has been demonstrated to have a significant impact on the prognosis of colorectal cancer (CRC). This study aimed to investigate CD3 CT (CD3 + T cells density in the core of the tumor [CT]) prognostic ability in patients with CRC by using AI technology. METHODS: The study involved the enrollment of 492 patients from two distinct medical centers, with 358 patients assigned to the training cohort and an additional 134 patients allocated to the validation cohort. To facilitate tissue segmentation and T-cells quantification in whole-slide images (WSIs), a fully automated workflow based on deep learning was devised. Upon the completion of tissue segmentation and subsequent cell segmentation, a comprehensive analysis was conducted. RESULTS: The evaluation of various positive T cell densities revealed comparable discriminatory ability between CD3 CT and CD3-CD8 (the combination of CD3 + and CD8 + T cells density within the CT and invasive margin) in predicting mortality (C-index in training cohort: 0.65 vs. 0.64; validation cohort: 0.69 vs. 0.69). The CD3 CT was confirmed as an independent prognostic factor, with high CD3 CT density associated with increased overall survival (OS) in the training cohort (hazard ratio [HR] = 0.22, 95% confidence interval [CI]: 0.12-0.38, P <0.001) and validation cohort (HR = 0.21, 95% CI: 0.05-0.92, P = 0.037). CONCLUSIONS: We quantify the spatial distribution of CD3 + and CD8 + T cells within tissue regions in WSIs using AI technology. The CD3 CT confirmed as a stage-independent predictor for OS in CRC patients. Moreover, CD3 CT shows promise in simplifying the CD3-CD8 system and facilitating its practical application in clinical settings.

Development of a machine learning-based radiomics signature for estimating breast cancer TME phenotypes and predicting anti-PD-1/PD-L1 immunotherapy response.

BACKGROUNDS: Since breast cancer patients respond diversely to immunotherapy, there is an urgent need to explore novel biomarkers to precisely predict clinical responses and enhance therapeutic efficacy. The purpose of our present research was to construct and independently validate a biomarker of tumor microenvironment (TME) phenotypes via a machine learning-based radiomics way. The interrelationship between the biomarker, TME phenotypes and recipients' clinical response was also revealed. METHODS: In this retrospective multi-cohort investigation, five separate cohorts of breast cancer patients were recruited to measure breast cancer TME phenotypes via a radiomics signature, which was constructed and validated by integrating RNA-seq data with DCE-MRI images for predicting immunotherapy response. Initially, we constructed TME phenotypes using RNA-seq of 1089 breast cancer patients in the TCGA database. Then, parallel DCE-MRI images and RNA-seq of 94 breast cancer patients obtained from TCIA were applied to develop a radiomics-based TME phenotypes signature using random forest in machine learning. The repeatability of the radiomics signature was then validated in an internal validation set. Two additional independent external validation sets were analyzed to reassess this signature. The Immune phenotype cohort (n = 158) was divided based on CD8 cell infiltration into immune-inflamed and immune-desert phenotypes; these data were utilized to examine the relationship between the immune phenotypes and this signature. Finally, we utilized an Immunotherapy-treated cohort with 77 cases who received anti-PD-1/PD-L1 treatment to evaluate the predictive efficiency of this signature in terms of clinical outcomes. RESULTS: The TME phenotypes of breast cancer were separated into two heterogeneous clusters: Cluster A, an "immune-inflamed" cluster, containing substantial innate and adaptive immune cell infiltration, and Cluster B, an "immune-desert" cluster, with modest TME cell infiltration. We constructed a radiomics signature for the TME phenotypes ([AUC] = 0.855; 95% CI 0.777-0.932; p < 0.05) and verified it in an internal validation set (0.844; 0.606-1; p < 0.05). In the known immune phenotypes cohort, the signature can identify either immune-inflamed or immune-desert tumor (0.814; 0.717-0.911; p < 0.05). In the Immunotherapy-treated cohort, patients with objective response had higher baseline radiomics scores than those with stable or progressing disease (p < 0.05); moreover, the radiomics signature achieved an AUC of 0.784 (0.643-0.926; p < 0.05) for predicting immunotherapy response. CONCLUSIONS: Our imaging biomarker, a practicable radiomics signature, is beneficial for predicting the TME phenotypes and clinical response in anti-PD-1/PD-L1-treated breast cancer patients. It is particularly effective in identifying the "immune-desert" phenotype and may aid in its transformation into an "immune-inflamed" phenotype.

SwinHR: Hemodynamic-powered hierarchical vision transformer for breast tumor segmentation.

Accurate and automated segmentation of breast tumors in dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) plays a critical role in computer-aided diagnosis and treatment of breast cancer. However, this task is challenging, due to random variation in tumor sizes, shapes, appearances, and blurred boundaries of tumors caused by inherent heterogeneity of breast cancer. Moreover, the presence of ill-posed artifacts in DCE-MRI further complicate the process of tumor region annotation. To address the challenges above, we propose a scheme (named SwinHR) integrating prior DCE-MRI knowledge and temporal-spatial information of breast tumors. The prior DCE-MRI knowledge refers to hemodynamic information extracted from multiple DCE-MRI phases, which can provide pharmacokinetics information to describe metabolic changes of the tumor cells over the scanning time. The Swin Transformer with hierarchical re-parameterization large kernel architecture (H-RLK) can capture long-range dependencies within DCE-MRI while maintaining computational efficiency by a shifted window-based self-attention mechanism. The use of H-RLK can extract high-level features with a wider receptive field, which can make the model capture contextual information at different levels of abstraction. Extensive experiments are conducted in large-scale datasets to validate the effectiveness of our proposed SwinHR scheme, demonstrating its superiority over recent state-of-the-art segmentation methods. Also, a subgroup analysis split by MRI scanners, field strength, and tumor size is conducted to verify its generalization. The source code is released on (https://github.com/GDPHMediaLab/SwinHR).

Noninvasive Assessment of Diabetic Kidney Disease With MRI: Hype or Hope?

Owing to the increasing prevalence of diabetic mellitus, diabetic kidney disease (DKD) is presently the leading cause of chronic kidney disease and end-stage renal disease worldwide. Early identification and disease interception is of paramount clinical importance for DKD management. However, current diagnostic, disease monitoring and prognostic tools are not satisfactory, due to their low sensitivity, low specificity, or invasiveness. Magnetic resonance imaging (MRI) is noninvasive and offers a host of contrast mechanisms that are sensitive to pathophysiological changes and risk factors associated with DKD. MRI tissue characterization involves structural and functional information including renal morphology (kidney volume (TKV) and parenchyma thickness using T1- or T2-weighted MRI), renal microstructure (diffusion weighted imaging, DWI), renal tissue oxygenation (blood oxygenation level dependent MRI, BOLD), renal hemodynamics (arterial spin labeling and phase contrast MRI), fibrosis (DWI) and abdominal or perirenal fat fraction (Dixon MRI). Recent (pre)clinical studies demonstrated the feasibility and potential value of DKD evaluation with MRI. Recognizing this opportunity, this review outlines key concepts and current trends in renal MRI technology for furthering our understanding of the mechanisms underlying DKD and for supplementing clinical decision-making in DKD. Progress in preclinical MRI of DKD is surveyed, and challenges for clinical translation of renal MRI are discussed. Future directions of DKD assessment and renal tissue characterization with (multi)parametric MRI are explored. Opportunities for discovery and clinical break-through are discussed including biological validation of the MRI findings, large-scale population studies, standardization of DKD protocols, the synergistic connection with data science to advance comprehensive texture analysis, and the development of smart and automatic data analysis and data visualization tools to further the concepts of virtual biopsy and personalized DKD precision medicine. We hope that this review will convey this vision and inspire the reader to become pioneers in noninvasive assessment and management of DKD with MRI. LEVEL OF EVIDENCE: 1 TECHNICAL EFFICACY: Stage 2.

Vision Transformers for Computational Histopathology.

Computational histopathology is focused on the automatic analysis of rich phenotypic information contained in gigabyte whole slide images, aiming at providing cancer patients with more accurate diagnosis, prognosis, and treatment recommendations. Nowadays deep learning is the mainstream methodological choice in computational histopathology. Transformer, as the latest technological advance in deep learning, learns feature representations and global dependencies based on self-attention mechanisms, which is increasingly gaining prevalence in this field. This article presents a comprehensive review of state-of-the-art vision transformers that have been explored in histopathological image analysis for classification, segmentation, and survival risk regression applications. We first overview preliminary concepts and components built into vision transformers. Various recent applications including whole slide image classification, histological tissue component segmentation, and survival outcome prediction with tailored transformer architectures are then discussed. We finally discuss key challenges revolving around the use of vision transformers and envisioned future perspectives. We hope that this review could provide an elaborate guideline for readers to explore vision transformers in computational histopathology, such that more advanced techniques assisting in the precise diagnosis and treatment of cancer patients could be developed.

Decoding tumor stage by peritumoral and intratumoral radiomics in resectable esophageal squamous cell carcinoma.

PURPOSE: To evaluate the potential application of radiomics in predicting Tumor-Node-Metastasis (TNM) stage in patients with resectable esophageal squamous cell carcinoma (ESCC). METHODS: This retrospective study included 122 consecutive patients (mean age, 57 years; 27 women). Corresponding tumor of interest was identified on axial arterial-phase CT images with manual annotation. Radiomics features were extracted from intra- and peritumoral regions. Features were pruned to train LASSO regression model with 93 patients to construct a radiomics signature, whose performance was validated in a test set of 29 patients. Prognostic value of radiomics-predicted TNM stage was estimated by survival analysis in the entire cohort. RESULTS: The radiomics signature incorporating one intratumoral and four peritumoral features was significantly associated with TNM stage. This signature discriminated tumor stage with an area under curve (AUC) of 0.823 in the training set, with similar performance in the test set (AUC 0.813). Recurrence-free survival (RFS) was significantly different between different radiomics-predicted TNM stage groups (Low-risk vs high-risk, log-rank P = 0.004). Univariate and multivariate Cox regression analyses revealed that radiomics-predicted TNM stage was an independent preoperative factor for RFS. CONCLUSIONS: The proposed radiomics signature combing intratumoral and peritumoral features was predictive of TNM stage and associated with prognostication in ESCC.

Imaging-proteomic analysis for prediction of neoadjuvant chemotherapy responses in patients with breast cancer.

BACKGROUND: Optimizing patient selection for neoadjuvant chemotherapy in patients with breast cancer remains an unmet clinical need. Quantitative features from medical imaging were reported to be predictive of treatment responses. However, the biologic meaning of these latent features is poorly understood, preventing the clinical use of such noninvasive imaging markers. The study aimed to develop a deep learning signature (DLS) from pretreatment magnetic resonance imaging (MRI) for predicting responses to neoadjuvant chemotherapy in patients with breast cancer and to further investigate the biologic meaning of the DLS by identifying its underlying pathways using paired MRI and proteomic sequencing data. METHODS: MRI-based DLS was constructed (radiogenomic training dataset, n = 105) and validated (radiogenomic validation dataset, n = 26) for the prediction of pathologic complete response (pCR) to neoadjuvant chemotherapy. Proteomic sequencing revealed biological functions facilitating pCR (n = 139). Their associations with DLS were uncovered by radiogenomic analysis. RESULTS: The DLS achieved a prediction accuracy of 0.923 with an AUC of 0.958, higher than the performance of the model trained by transfer learning. Cellular membrane formation, endocytosis, insulin-like growth factor binding, protein localization to membranes, and cytoskeleton-dependent trafficking were differentially regulated in patients showing pCR. Oncogenic signaling pathways, features correlated with human phenotypes, and features correlated with general biological processes were significantly correlated with DLS in both training and validation dataset (p.adj < 0.05). CONCLUSIONS: Our study offers a biologically interpretable DLS for the prediction of pCR to neoadjuvant chemotherapy in patients with breast cancer, which may guide personalized medication.

Large-scale pancreatic cancer detection via non-contrast CT and deep learning.

Pancreatic ductal adenocarcinoma (PDAC), the most deadly solid malignancy, is typically detected late and at an inoperable stage. Early or incidental detection is associated with prolonged survival, but screening asymptomatic individuals for PDAC using a single test remains unfeasible due to the low prevalence and potential harms of false positives. Non-contrast computed tomography (CT), routinely performed for clinical indications, offers the potential for large-scale screening, however, identification of PDAC using non-contrast CT has long been considered impossible. Here, we develop a deep learning approach, pancreatic cancer detection with artificial intelligence (PANDA), that can detect and classify pancreatic lesions with high accuracy via non-contrast CT. PANDA is trained on a dataset of 3,208 patients from a single center. PANDA achieves an area under the receiver operating characteristic curve (AUC) of 0.986-0.996 for lesion detection in a multicenter validation involving 6,239 patients across 10 centers, outperforms the mean radiologist performance by 34.1% in sensitivity and 6.3% in specificity for PDAC identification, and achieves a sensitivity of 92.9% and specificity of 99.9% for lesion detection in a real-world multi-scenario validation consisting of 20,530 consecutive patients. Notably, PANDA utilized with non-contrast CT shows non-inferiority to radiology reports (using contrast-enhanced CT) in the differentiation of common pancreatic lesion subtypes. PANDA could potentially serve as a new tool for large-scale pancreatic cancer screening.

Using the non-distortion IVIM to reduce the need for contrast agents in nasopharyngeal MRI.

BACKGROUND: Patients with nasopharyngeal carcinoma (NPC) who undergo longitudinal follow-up contrast-enhanced MRI are at risk of developing gadolinium deposition in their neural tissue, which may potentially harm them. Therefore, for these patients, a non-contrast-enhanced method is potentially beneficial as an alternative approach to predict enhancement in T1-weighted imaging (CE-T1WI). The traditional intravoxel incoherent motion (IVIM) is one of the non-contrast-enhanced methods; however, the severe distortion and signal loss limit its application in patients with NPC. The present study aimed to investigate whether non-distortion IVIM could reduce the need of CE-T1WI in the follow-up of patients with NPC. METHODS: The patients with NPC underwent Turbo Spin-echo MVXD diffusion-weighted imaging-based IVIM (non-distortion IVIM) from November 2021 to May 2022. Firstly, thirty patients with NPC were underwent both non-distortion IVIM and traditional IVIM. The distortion rate (DR) of the non-distortion IVIM was compared with the traditional IVIM. Then, twenty-one NPC patients with tumors (areas >50mm2) were included and correlation coefficient analysis was used to assess the relationship between their non-distortion IVIM and CE-T1WI. Linear regression analysis was performed to determine whether non-distortion IVIM predictors could predict CE-T1WI. RESULTS: The correlation was observed between the parameter f of the non-distortion IVIM and the enhancement ratio of CE-T1WI (r = 0.543, P = 0.011). Moreover, the linear regression analysis revealed that f was an independent IVIM predictor of CE-T1WI in patients with NPC (P = 0.011). The DR of the non-distortion IVIM was significantly smaller than that of the traditional IVIM (0.12 ± 0.05 vs 0.48 ± 0.16, P < 0.001). CONCLUSIONS: In patients with NPC, non-distortion IVIM showed potential clinical benefits to reduce the need for contrast agents, and it can independently predict the enhancement ratio.

Computerized tertiary lymphoid structures density on H&E-images is a prognostic biomarker in resectable lung adenocarcinoma.

The increased amount of tertiary lymphoid structures (TLSs) is associated with a favorable prognosis in patients with lung adenocarcinoma (LUAD). However, evaluating TLSs manually is an experience-dependent and time-consuming process, which limits its clinical application. In this multi-center study, we developed an automated computational workflow for quantifying the TLS density in the tumor region of routine hematoxylin and eosin (H&E)-stained whole-slide images (WSIs). The association between the computerized TLS density and disease-free survival (DFS) was further explored in 802 patients with resectable LUAD of three cohorts. Additionally, a Cox proportional hazard regression model, incorporating clinicopathological variables and the TLS density, was established to assess its prognostic ability. The computerized TLS density was an independent prognostic biomarker in patients with resectable LUAD. The integration of the TLS density with clinicopathological variables could support individualized clinical decision-making by improving prognostic stratification.

Breast Fibroglandular Tissue Segmentation for Automated BPE Quantification With Iterative Cycle-Consistent Semi-Supervised Learning.

Background Parenchymal Enhancement (BPE) quantification in Dynamic Contrast-Enhanced Magnetic Resonance Imaging (DCE-MRI) plays a pivotal role in clinical breast cancer diagnosis and prognosis. However, the emerging deep learning-based breast fibroglandular tissue segmentation, a crucial step in automated BPE quantification, often suffers from limited training samples with accurate annotations. To address this challenge, we propose a novel iterative cycle-consistent semi-supervised framework to leverage segmentation performance by using a large amount of paired pre-/post-contrast images without annotations. Specifically, we design the reconstruction network, cascaded with the segmentation network, to learn a mapping from the pre-contrast images and segmentation predictions to the post-contrast images. Thus, we can implicitly use the reconstruction task to explore the inter-relationship between these two-phase images, which in return guides the segmentation task. Moreover, the reconstructed post-contrast images across multiple auto-context modeling-based iterations can be viewed as new augmentations, facilitating cycle-consistent constraints across each segmentation output. Extensive experiments on two datasets with various data distributions show great segmentation and BPE quantification accuracy compared with other state-of-the-art semi-supervised methods. Importantly, our method achieves 11.80 times of quantification accuracy improvement along with 10 times faster, compared with clinical physicians, demonstrating its potential for automated BPE quantification. The code is available at https://github.com/ZhangJD-ong/Iterative-Cycle-consistent-Semi-supervised-Learning-for-fibroglandular-tissue-segmentation.