Machine Learning-Based Prediction of Pathological Responses and Prognosis After Neoadjuvant Chemotherapy for Non-Small-Cell Lung Cancer: A Retrospective Study.

BACKGROUND: Neoadjuvant chemotherapy has variable efficacy in patients with non-small-cell lung cancer (NSCLC), yet reliable noninvasive predictive markers are lacking. This study aimed to develop a radiomics model predicting pathological complete response and postneoadjuvant chemotherapy survival in NSCLC. MATERIALS AND METHODS: Retrospective data collection involved 130 patients with NSCLC who underwent neoadjuvant chemotherapy and surgery. Patients were randomly divided into training and independent testing sets. Nine radiomics features from prechemotherapy computed tomography (CT) images were extracted from intratumoral and peritumoral regions. An auto-encoder model was constructed, and its performance was evaluated. X-tile software classified patients into high and low-risk groups based on their predicted probabilities. survival of patients in different risk groups and the role of postoperative adjuvant chemotherapy were examined. RESULTS: The model demonstrated area under the receiver operating characteristic (ROC) curve of 0.874 (training set) and 0.876 (testing set). The larger the area under curve (AUC), the better the model performance. Calibration curve and decision curve analysis indicated excellent model calibration (Hosmer-Lemeshow test, P = .763, the higher the P-value, the better the model fit) and potential clinical applicability. Survival analysis revealed significant differences in overall survival (P = .011) and disease-free survival (P = .017) between different risk groups. Adjuvant chemotherapy significantly improved survival in the low-risk group (P = .041) but not high-risk group (P = 0.56). CONCLUSION: This study represents the first successful prediction of pathological complete response achievement after neoadjuvant chemotherapy for NSCLC, as well as the patients' survival, utilizing intratumoral and peritumoral radiomics features.

Metabolic Heterogeneity and Potential Immunotherapeutic Responses Revealed by Single-Cell Transcriptomics of Breast Cancer.

BACKGROUND: Breast cancer (BC) exhibits remarkable heterogeneity. However, the transcriptomic heterogeneity of BC at the single-cell level has not been fully elucidated. METHODS: We acquired BC samples from 14 patients. Single-cell RNA sequencing (scRNA-seq), bioinformatic analyses, along with immunohistochemistry (IHC) and immunofluorescence (IF) assays were carried out. RESULTS: According to the scRNA-seq results, 10 different cell types were identified. We found that Cancer-Associated Fibroblasts (CAFs) exhibited distinct biological functions and may promote resistance to therapy. Metabolic analysis of tumor cells revealed heterogeneity in glycolysis, gluconeogenesis, and fatty acid synthetase reprogramming, which led to chemotherapy resistance. Furthermore, patients with multiple metastases and progression were predicted to benefit from immunotherapy based on a heterogeneity analysis of T cells and tumor cells. CONCLUSIONS: Our findings provide a comprehensive understanding of the heterogeneity of BC, provide comprehensive insight into the correlation between cancer metabolism and chemotherapy resistance, and enable the prediction of immunotherapy responses based on T-cell heterogeneity.

Trajectory patterns and cumulative burden of CEA during follow-up with non-small cell lung cancer outcomes: A retrospective longitudinal cohort study.

BACKGROUND: Previous studies of non-small cell lung cancer (NSCLC) focused on CEA measured at a single time point, ignoring serial CEA measurements. METHODS: This retrospective cohort included 2959 patients underwent surgery for stage I-III NSCLC. CEA trajectory patterns and long-term cumulative CEA burden were evaluated using the latent class growth mixture model. RESULTS: Four CEA trajectory groups were identified, named as low-stable, decreasing, early-rising and later-rising. Compared with the low-stable group, the adjusted hazard ratios associated with death were 1.27, 4.50, and 3.68 for the other groups. Cumulative CEA burden were positively associated with the risk of death in patients not belonging to the low-stable group. The 5-year overall survival (OS) rates decreased from 62.3% to 33.0% for the first and fourth quantile groups of cumulative CEA burden. Jointly, patients with decreasing CEA trajectory could be further divided into the decreasing & low and decreasing & high group, with 5-year OS rates to be 77.9% and 47.1%. Patients with rising CEA trajectory and high cumulative CEA were found to be more likely to develop bone metastasis. CONCLUSIONS: Longitudinal trajectory patterns and long-term cumulative burden of CEA were independent prognostic factors of NSCLC. We recommend CEA in postoperative surveillance of NSCLC.

No prognostic impact of staging bone scan in patients with stage IA non-small cell lung cancer.

OBJECTIVE: To investigate the survival benefit of preoperative bone scan in asymptomatic patients with early-stage non-small cell lung cancer (NSCLC). METHODS: This retrospective study included patients with radical resection for stage T1N0M0 NSCLC between March 2013 and December 2018. During postoperative follow-up, we monitored patient survival and the development of bone metastasis. We compared overall survival, bone metastasis-free survival, and recurrence-free survival in patients with or without preoperative bone scan. Propensity score matching and inverse probability of treatment weighting were used to minimize election bias. RESULTS: A total of 868 patients (58.19 ± 9.69 years; 415 men) were included in the study. Of 87.7% (761 of 868) underwent preoperative bone scan. In the multivariable analyses, bone scan did not improve overall survival (hazard ratio [HR] 1.49; 95% confidence intervals [CI] 0.91-2.42; p = 0.113), bone metastasis-free survival (HR 1.18; 95% CI 0.73-1.90; p = 0.551), and recurrence-free survival (HR 0.89; 95% CI 0.58-1.39; p = 0.618). Similar results were obtained after propensity score matching (overall survival [HR 1.28; 95% CI 0.74-2.23; p = 0.379], bone metastasis-free survival [HR 1.00; 95% CI 0.58-1.72; p = 0.997], and recurrence-free survival [HR 0.76; 95% CI 0.46-1.24; p = 0.270]) and inverse probability of treatment weighting. CONCLUSION: There were no significant differences in overall survival, bone metastasis-free survival, and recurrence-free survival between asymptomatic patients with clinical stage IA NSCLC with or without preoperative bone scan.

SC-Net: Symmetrical conical network for colorectal pathology image segmentation.

BACKGROUND AND OBJECTIVE: Image segmentation of histopathology of colorectal cancer is a core task of computer aided medical image diagnosis system. Existing convolutional neural networks generally extract multi-scale information in linear flow structures by inserting multi-branch modules, which is difficult to extract heterogeneous semantic information under multi-level and different receptive field and tough to establish context dependency among different receptive field features. METHODS: To address these issues, we propose a symmetric spiral progressive feature fusion encoder-decoder network called the Symmetric Conical Network (SC-Net). First, we design a Multi-scale Feature Extraction Block (MFEB) matching with the Symmetric Conical Network to obtain multi-branch heterogeneous semantic information under different receptive fields, so as to enrich the diversity of extracted feature information. The encoder is composed of MFEB through spiral and multi-branch arrangement to enhance context dependence between different information flow. Secondly, the information loss of contour, color and others in high-level semantic information through causally stacking MFEB, the Feature Mapping Layer (FML) is designed to map low-level features to high-level semantic features along the down-sampling branch and solve the problem of insufficient global feature extraction in deep levels. RESULTS: The SC-Net was evaluated on our self-constructed colorectal cancer dataset, a publicly available breast cancer dataset and a polyp dataset. The results revealed that the mDice of segmentation reached 0.8611, 0.7259 and 0.7144. We compare our model with the state-of-art semantic segmentation UNet++, PSPNet, Attention U-Net, R2U-Net and other advanced segmentation networks. The experimental results demonstrate that we achieve the most advanced performance. CONCLUSIONS: The results indicate that the proposed SC-Net excels in segmenting H&E stained pathology images, effectively preserving morphological features and spatial information even in scenarios with weak texture, poor contrast, and variations in appearance.

A nomogram for predicting liver metastasis in patients with gastric gastrointestinal stromal tumor.

BACKGROUND: Liver metastasis (LIM) is an important factor in the diagnosis, treatment, follow-up, and prognosis of patients with gastric gastrointestinal stromal tumor (GIST). There is no simple tool to assess the risk of LIM in patients with gastric GIST. Our aim was to develop and validate a nomogram to identify patients with gastric GIST at high risk of LIM. METHODS: Patient data diagnosed as having gastric GIST between 2010 and 2019 were extracted from the Surveillance, Epidemiology, and End Results (SEER) database and randomly divided into training cohort and internal validation cohort in a 7:3 ratio. For external validation, retrospective data collection was performed on patients diagnosed as having gastric GIST at Yunnan Cancer Center (YNCC) between January 2015 and May 2023. Univariate and multivariate logistic regression analyses were used to identify independent risk factors associated with LIM in patients with gastric GIST. An individualized LIM nomogram specific for gastric GIST was formulated based on the multivariate logistic model; its discriminative performance, calibration, and clinical utility were evaluated. RESULTS: In the SEER database, a cohort of 2341 patients with gastric GIST was analyzed, of which 173 cases (7.39%) were found to have LIM; 239 patients with gastric GIST from the YNCC database were included, of which 25 (10.46%) had LIM. Multivariate analysis showed tumor size, tumor site, and sex were independent risk factors for LIM (P < .05). The nomogram based on the basic clinical characteristics of tumor size, tumor site, sex, and age demonstrated significant discrimination, with an area under the curve of 0.753 (95% CI, 0.692-0.814) and 0.836 (95% CI, 0.743-0.930) in the internal and external validation cohort, respectively. The Hosmer-Lemeshow test showed that the nomogram was well calibrated, whereas the decision curve analysis and the clinical impact plot demonstrated its clinical utility. CONCLUSION: Tumor size, tumor subsite, and sex were significantly correlated with the risk of LIM in gastric GIST. The nomogram for patients with GIST can effectively predict the individualized risk of LIM and contribute to the planning and decision making related to metastasis management in clinical practice.

ZNF326 as a potential prognostic and predictive biomarker in stage II colorectal cancer.

Background: Adjuvant chemotherapy is considered for stage II colorectal cancer (CRC) patients with poor prognostic risk factors. However, current stratification algorithms are still insufficient to identify high-risk patients. Methods: We conducted a screening strategy to define ZNF326 based on quantitative proteomics in 11 paired CRC patients selected by a nested case-control design, and tested the association between ZNF326 expression level with the prognosis of stage II CRC patients and the benefit from adjuvant chemotherapy in public datasets; further investigation was conducted through subgroup analyses. Results: We found that low ZNF326 expression was significantly associated with a lower 5-year overall survival (OS) rate among stage II patients in both the discovery [P=0.008; hazard ratio (HR): 3.13, 95% confidence interval (CI): 1.29-7.58] and validation (P=0.025; HR: 1.98, 95% CI: 1.08-3.65) cohorts. In the Cox multivariable analysis, low ZNF326 expression was both associated with shorter OS after adjustment for age, sex, and adjuvant chemotherapy in the discovery and validation data sets. Subgroup analyses yielded largely similar results. In a pooled database, the rate of 5-year OS was higher among stage II ZNF326-high tumors who were treated with adjuvant chemotherapy than it was among those who were not treated with adjuvant chemotherapy (P=0.011; HR: 0.28, 95% CI: 0.10-0.80). Conclusions: ZNF326 has the potential to be used in clinical practice for risk classification. ZNF326-low expression level identified a subgroup of patients with high-risk stage II CRC who appeared to less benefit from adjuvant chemotherapy.

Modality-Specific Information Disentanglement From Multi-Parametric MRI for Breast Tumor Segmentation and Computer-Aided Diagnosis.

Breast cancer is becoming a significant global health challenge, with millions of fatalities annually. Magnetic Resonance Imaging (MRI) can provide various sequences for characterizing tumor morphology and internal patterns, and becomes an effective tool for detection and diagnosis of breast tumors. However, previous deep-learning based tumor segmentation methods from multi-parametric MRI still have limitations in exploring inter-modality information and focusing task-informative modality/modalities. To address these shortcomings, we propose a Modality-Specific Information Disentanglement (MoSID) framework to extract both inter- and intra-modality attention maps as prior knowledge for guiding tumor segmentation. Specifically, by disentangling modality-specific information, the MoSID framework provides complementary clues for the segmentation task, by generating modality-specific attention maps to guide modality selection and inter-modality evaluation. Our experiments on two 3D breast datasets and one 2D prostate dataset demonstrate that the MoSID framework outperforms other state-of-the-art multi-modality segmentation methods, even in the cases of missing modalities. Based on the segmented lesions, we further train a classifier to predict the patients' response to radiotherapy. The prediction accuracy is comparable to the case of using manually-segmented tumors for treatment outcome prediction, indicating the robustness and effectiveness of the proposed segmentation method. The code is available at https://github.com/Qianqian-Chen/MoSID.

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.

Preoperative splenic area as a prognostic biomarker of early-stage non-small cell lung cancer.

BACKGROUND: The correlation between the preoperative splenic area measured on CT scans and the overall survival (OS) of early-stage non-small cell lung cancer (NSCLC) patients remains unclear. METHODS: A retrospective discovery cohort and validation cohort consisting of consecutive NSCLC patients who underwent resection and preoperative CT scans were created. The patients were divided into two groups based on the measurement of their preoperative splenic area: normal and abnormal. The Cox proportional hazard model was used to analyse the correlation between splenic area and OS. RESULTS: The discovery and validation cohorts included 2532 patients (1374 (54.27%) males; median (IQR) age 59 (52-66) years) and 608 patients (403 (66.28%) males; age 69 (62-76) years), respectively. Patients with a normal splenic area had a 6% higher 5-year OS (n = 727 (80%)) than patients with an abnormal splenic area (n = 1805 (74%)) (p = 0.007) in the discovery cohort. A similar result was obtained in the validation cohort. In the univariable analysis, the OS hazard ratios (HRs) for the patients with abnormal splenic areas were 1.32 (95% confidence interval (CI): 1.08, 1.61) in the discovery cohort and 1.59 (95% CI: 1.01, 2.50) in the validation cohort. Multivariable analysis demonstrated that abnormal splenic area was independent of shorter OS in the discovery (HR: 1.32, 95% CI: 1.08, 1.63) and validation cohorts (HR: 1.84, 95% CI: 1.12, 3.02). CONCLUSION: Preoperative CT measurements of the splenic area serve as a prognostic indicator for early-stage NSCLC patients, offering a novel metric with potential implications for personalized therapeutic strategies in top-tier oncology research.

Effect of initial recurrence site on the prognosis of different tissue types of non-small cell lung cancer: a retrospective cohort study.

PURPOSE: To explore the correlation between the initial recurrence site and survival after recurrence (PRS) in non-small cell lung cancer (NSCLC). METHODS: We collected 588 stages I-III NSCLC patients with recurrence after radical resection in Yunnan Cancer Hospital from January 2013 to December 2018. We used Kaplan-Meier survival curves to compare PRS in patients with different site recurrences. The univariate and multivariate Cox proportional hazard models were used to analyze the impact of the initial recurrence site on PRS. RESULTS: The recurrence site included the lung (n = 109), brain (n = 113), bone (n = 79), abdomen (n = 28), pleura (n = 24), lymph node (n = 81), and multisite (n = 154). In the total population, patients with multisite recurrence had substantially worse PRS (24.8 months, 95% confidence interval [CI]: 17.46-32.20) than that of patients without multiple sites recurrence (42.2 months, 95% CI 32.24-52.10) (P = 0.026). However, patients with lung recurrence had better RFS (63.1 months, 95% CI 51.13-74.00) than those who did not (31.0 months, 95% CI 25.10-36.96) (P < 0.001). In adenocarcinoma, patients with pleural recurrence had substantially worse PRS (21.3 months, 95% CI 15.07-27.46) than that of patients without pleural recurrence (46.9 months, 95% CI 35.07-58.80) (P = 0.031). Multivariate Cox proportional hazards regression analysis revealed that lung recurrence (HR 0.58, 95% CI 0.40-0.82; P = 0.003) was independent protective prognostic factor for PRS in the total population, while pleural recurrence (HR 2.18, 95% CI 1.14-4.17; P = 0.018) was independent adverse prognostic factors for PRS in adenocarcinoma patients. CONCLUSION: The initial recurrence site was associated with PRS in NSCLC patients. Identification of recurrence sites could guide the subsequent treatment.

Effects of parasternal intercostal block on surgical site wound infection and pain in patients undergoing cardiac surgery: A meta-analysis.

This study aimed to assess the effect of parasternal intercostal block on postoperative wound infection, pain, and length of hospital stay in patients undergoing cardiac surgery. PubMed, Embase, Cochrane Library, China National Knowledge Infrastructure, VIP, and Wanfang databases were extensively queried using a computer, and randomised controlled studies (RCTs) from the inception of each database to July 2023 were sought using keywords in English and Chinese language. Literature quality was assessed using Cochrane-recommended tools, and the included data were collated and analysed using Stata 17.0 software for meta-analysis. Ultimately, eight RCTs were included. Meta-analysis revealed that utilising parasternal intercostal block during cardiac surgery significantly reduced postoperative wound pain (standardised mean difference [SMD] = -1.01, 95% confidence intervals [CI]: -1.70 to -0.31, p = 0.005) and significantly shortened hospital stay (SMD = -0.40, 95% CI: -0.77 to -0.04, p = 0.029), though it may increase the risk of wound infection (OR = 5.03, 95% CI:0.58-44.02, p = 0.144); however, the difference was not statistically significant. The application of parasternal intercostal block during cardiac surgery can significantly reduce postoperative pain and shorten hospital stay. This approach is worth considering for clinical implementation. Decisions regarding its adoption should be made in conjunction with the relevant clinical indices and surgeon's experience.

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.

miR-26b-5p promotes osteogenesis of bone mesenchymal stem cells via suppressing FGF21.

BACKGROUND: miR-26b-5p actively participates in the osteogenic differentiation of bone mesenchymal stem cells (BMSCs). The database showed that fibroblast growth factor (FGF)-21 is a potential binding site of miR-26b-5p. This study aimed to investigate the molecular osteogenic mechanisms of miR-26b-5p targeting FGF21. METHODS: Bone marrow was aspirated from the anterior superior iliac spine during bone marrow puncture. BMSCs were used to establish an in vitro cell model, and BMSCs markers were analyzed by flow cytometry. miR-26b-5p were overexpressed for 48 hours, and then placed in an osteogenic induction medium for osteogenic induction culture, the expression of RNA was detected using RT-qPCR. On day 7 of induction, RT-qPCR was used to measure Runx2, Osterix (Osx), and target gene FGF21 expression levels in each group. RT-qPCR, the dual-luciferase reporter gene system and western blot were used to verify that FGF21 was a direct target of miR-26b-5p. RESULTS: BMSCs were identified according to the antigenic characteristics. miR-26b-5p expression was significantly upregulated after the expression of miR-26b-5p mimics, and FGF21 expression was downregulated; in miR-26b-5p inhibitor, the opposite results were revealed. After overexpression of miR-26b-5p, the alkaline phosphatase activity and nodules of Alizarin red S in the culture medium was increased; the opposite results were revealed in miR-26b-5p inhibitor. The expressions of Runx2 and Osx in the miR-26b-5p group were also significantly higher; in the miR-26b-5p inhibitor group, the opposite results were revealed. Luciferase reporter assays demonstrated that FGF21 was a direct target of miR-26b-5p. The western blotting analysis showed that FGF21 expression was significantly downregulated in the miR-26b-5p overexpressed group. Finally, the expressions of the characteristic osteogenic factors in the miR-26b-5p control + FGF21 group was significantly lower, but then increased significantly in the miR-26b-5p mimics + FGF21 group; the expressions of the characteristic osteogenic factors in the miR-26b-5p control + si-FGF21 group was significantly higher. CONCLUSIONS: miR-26b-5p can regulate the osteogenic differentiation of BMSCs and participate in PMOP pathogenesis via suppressing FGF21.

A robust and efficient AI assistant for breast tumor segmentation from DCE-MRI via a spatial-temporal framework.

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) allows screening, follow up, and diagnosis for breast tumor with high sensitivity. Accurate tumor segmentation from DCE-MRI can provide crucial information of tumor location and shape, which significantly influences the downstream clinical decisions. In this paper, we aim to develop an artificial intelligence (AI) assistant to automatically segment breast tumors by capturing dynamic changes in multi-phase DCE-MRI with a spatial-temporal framework. The main advantages of our AI assistant include (1) robustness, i.e., our model can handle MR data with different phase numbers and imaging intervals, as demonstrated on a large-scale dataset from seven medical centers, and (2) efficiency, i.e., our AI assistant significantly reduces the time required for manual annotation by a factor of 20, while maintaining accuracy comparable to that of physicians. More importantly, as the fundamental step to build an AI-assisted breast cancer diagnosis system, our AI assistant will promote the application of AI in more clinical diagnostic practices regarding breast cancer.

A Hierarchical Graph V-Net With Semi-Supervised Pre-Training for Histological Image Based Breast Cancer Classification.

Numerous patch-based methods have recently been proposed for histological image based breast cancer classification. However, their performance could be highly affected by ignoring spatial contextual information in the whole slide image (WSI). To address this issue, we propose a novel hierarchical Graph V-Net by integrating 1) patch-level pre-training and 2) context-based fine-tuning, with a hierarchical graph network. Specifically, a semi-supervised framework based on knowledge distillation is first developed to pre-train a patch encoder for extracting disease-relevant features. Then, a hierarchical Graph V-Net is designed to construct a hierarchical graph representation from neighboring/similar individual patches for coarse-to-fine classification, where each graph node (corresponding to one patch) is attached with extracted disease-relevant features and its target label during training is the average label of all pixels in the corresponding patch. To evaluate the performance of our proposed hierarchical Graph V-Net, we collect a large WSI dataset of 560 WSIs, with 30 labeled WSIs from the BACH dataset (through our further refinement), 30 labeled WSIs and 500 unlabeled WSIs from Yunnan Cancer Hospital. Those 500 unlabeled WSIs are employed for patch-level pre-training to improve feature representation, while 60 labeled WSIs are used to train and test our proposed hierarchical Graph V-Net. Both comparative assessment and ablation studies demonstrate the superiority of our proposed hierarchical Graph V-Net over state-of-the-art methods in classifying breast cancer from WSIs. The source code and our annotations for the BACH dataset have been released at https://github.com/lyhkevin/Graph-V-Net.

A nomogram-based optimized Radscore for preoperative prediction of lymph node metastasis in patients with cervical cancer after neoadjuvant chemotherapy.

Purpose: To construct a superior single-sequence radiomics signature to assess lymphatic metastasis in patients with cervical cancer after neoadjuvant chemotherapy (NACT). Methods: The first half of the study was retrospectively conducted in our hospital between October 2012 and December 2021. Based on the history of NACT before surgery, all pathologies were divided into the NACT and surgery groups. The incidence rate of lymphatic metastasis in the two groups was determined based on the results of pathological examination following lymphadenectomy. Patients from the primary and secondary centers who received NACT were enrolled for radiomics analysis in the second half of the study. The patient cohorts from the primary center were randomly divided into training and test cohorts at a ratio of 7:3. All patients underwent magnetic resonance imaging after NACT. Segmentation was performed on T1-weighted imaging (T1WI), T2-weighted imaging, contrast-enhanced T1WI (CET1WI), and diffusion-weighted imaging. Results: The rate of lymphatic metastasis in the NACT group (33.2%) was significantly lower than that in the surgery group (58.7%, P=0.007). The area under the receiver operating characteristic curve values of Radscore_CET1WI for predicting lymph node metastasis and non-lymphatic metastasis were 0.800 and 0.797 in the training and test cohorts, respectively, exhibiting superior diagnostic performance. After combining the clinical variables, the tumor diameter on magnetic resonance imaging was incorporated into the Rad_clin model constructed using Radscore_CET1WI. The Hosmer-Lemeshow test of the Rad_clin model revealed no significant differences in the goodness of fit in the training (P=0.594) or test cohort (P=0.748). Conclusions: The Radscore provided by CET1WI may achieve a higher diagnostic performance in predicting lymph node metastasis. Superior performance was observed with the Rad_clin model.

Improving prognosis and assessing adjuvant chemotherapy benefit in locally advanced rectal cancer with deep learning for MRI: A retrospective, multi-cohort study.

PURPOSE: Adjuvant therapy is recommended to minimize the risk of distant metastasis (DM) and local recurrence (LR) in patients with locally advanced rectal cancer (LARC). However, its role is controversial. We aimed to develop a pretreatment MRI-based deep learning model to predict LR, DM, and overall survival (OS) over 5 years after surgery and to identify patients benefitting from adjuvant chemotherapy (AC). MATERIALS AND METHODS: The multi-survival tasks network (MuST) model was developed in a primary cohort (n = 308) and validated using two external cohorts (n = 247, 245). An AC decision tree integrating the MuST-DM score, perineural invasion (PNI), and preoperative carbohydrate antigen 19-9 (CA19-9) was constructed to assess chemotherapy benefits and aid personalized treatment of patients. We also quantified the prognostic improvement of the decision tree. RESULTS: The MuST network demonstrated high prognostic accuracy in the primary and two external cohorts for the prediction of three different survival tasks. Within the stratified analysis and decision tree, patients with CA19-9 levels > 37 U/mL and high MuST-DM scores exhibited favorable chemotherapy efficacy. Similar results were observed in PNI-positive patients with low MuST-DM scores. PNI-negative patients with low MuST-DM scores exhibited poor chemotherapy efficacy. Based on the decision tree, 14 additional patients benefiting from AC and 391 patients who received over-treatment were identified in this retrospective study. CONCLUSION: The MuST model accurately and non-invasively predicted OS, DM, and LR. A specific and direct tool linking chemotherapy decisions and benefit quantification has also been provided.