Guiding induction chemotherapy of locoregionally advanced nasopharyngeal carcinoma with ternary classification of predicted individual treatment effect.

BACKGROUND AND PURPOSE: Induction chemotherapy (IC) before concurrent chemoradiotherapy does not universally improve long-term overall survival (OS) in locoregionally advanced nasopharyngeal carcinoma (LANPC). Conventional risk stratification often yields suboptimal IC decisions. Our study introduces a ternary classification of predicted individual treatment effect (PITE) to guide personalized IC decisions. MATERIALS AND METHODS: A two-center retrospective analysis of 1,213 patients with LANPC was conducted to develop and validate prognostic models integrating magnetic resonance imaging and clinical data to estimate individual 5-year OS probabilities for IC and non-IC treatments. Differences in these probabilities defined PITE, facilitating patient stratification into three IC recommendation categories. Model effectiveness was validated using Kaplan-Meier estimators, decision curve-like analysis, and evaluations of variable importance and distribution. RESULTS: The models exhibited strong predictive performance in both treatments across training and cross-validation sets, enabling accurate PITE calculations and patient classification. Compared with non-IC treatment, IC markedly improved OS in the IC-preferred group (HR = 0.62, p = 0.02), had no effect in the IC-neutral group (HR = 1.00, p = 0.70), and worsened OS in the IC-opposed group (HR = 2.00, p = 0.03). The ternary PITE classification effectively identified 41.7 % of high-risk patients not benefiting from IC, and yielded a 2.68 % higher mean 5-year OS probability over risk-based decisions. Significantly increasing distributions of key prognostic indicators, such as metastatic lymph node number and plasma Epstein-Barr virus DNA level from IC-opposed to IC-preferred groups, further validated the clinical relevance of PITE classification. CONCLUSION: The ternary PITE classification offers an accurate and clinically advantageous approach to guide personalized IC decision-making in patients with LANPC.

Motion and anatomy dual aware lung ventilation imaging by integrating Jacobian map and average CT image using dual path fusion network.

BACKGROUND: Deep learning-based computed tomography (CT) ventilation imaging (CTVI) is a promising technique for guiding functional lung avoidance radiotherapy (FLART). However, conventional approaches, which rely on anatomical CT data, may overlook important ventilation features due to the lack of motion data integration. PURPOSE: This study aims to develop a novel dual-aware CTVI method that integrates both anatomical information from CT images and motional information from Jacobian maps to generate more accurate ventilation images for FLART. METHODS: A dataset of 66 patients with four-dimensional CT (4DCT) images and reference ventilation images (RefVI) was utilized to develop the dual-path fusion network (DPFN) for synthesizing ventilation images (CTVIDual). The DPFN model was specifically designed to integrate motion data from 4DCT-generated Jacobian maps with anatomical data from average 4DCT images. The DPFN utilized two specialized feature extraction pathways, along with encoders and decoders, designed to handle both 3D average CT images and Jacobian map data. This dual-processing approach enabled the comprehensive extraction of lung ventilation-related features. The performance of DPFN was assessed by comparing CTVIDual to RefVI using various metrics, including Spearman's correlation coefficients (R), Dice similarity coefficients of high-functional region (DSCh), and low-functional region (DSCl). Additionally, CTVIDual was benchmarked against other CTVI methods, including a dual-phase CT-based deep learning method (CTVIDLCT), a radiomics-based method (CTVIFM), a super voxel-based method (CTVISVD), a Unet-based method (CTVIUnet), and two deformable registration-based methods (CTVIJac and CTVIHU). RESULTS: In the test group, the mean R between CTVIDual and RefVI was 0.70, significantly outperforming CTVIDLCT (0.68), CTVIFM (0.58), CTVISVD (0.62), and CTVIUnet (0.66), with p < 0.05. Furthermore, the DSCh and DSCl values of CTVIDual were 0.64 and 0.80, respectively, outperforming CTVISVD (0.63; 0.73) and CTVIUnet (0.62; 0.77). The performance of CTVIDual was also significantly better than that of CTVIJac and CTVIHU. CONCLUSIONS: A novel dual-aware CTVI model that integrates anatomical and motion information was developed to synthesize lung ventilation images. It was shown that the accuracy of lung ventilation estimation could be significantly enhanced by incorporating motional information, particularly in patients with tumor-induced blockages. This approach has the potential to improve the accuracy of CTVI, enabling more effective FLART.

Xerostomia prediction in patients with nasopharyngeal carcinoma during radiotherapy using segmental dose distribution in dosiomics and radiomics models.

OBJECTIVES: This study aimed to integrate radiomics and dosiomics features to develop a predictive model for xerostomia (XM) in nasopharyngeal carcinoma after radiotherapy. It explores the influence of distinct feature extraction methods and dose ranges on the performance. MATERIALS AND METHODS: Data from 363 patients with nasopharyngeal carcinoma were retrospectively analyzed. We pioneered a dose-segmentation strategy, where the overall dose distribution (OD) was divided into four segmental dose distributions (SDs) at intervals of 15 Gy. Features were extracted using manual definition and deep learning, applying OD or SD and integrating radiomics and dosiomics, yielding corresponding feature scores (manually defined radiomics, MDR; manually defined dosiomics, MDD; deep learning-based radiomics, DLR; deep learning-based dosiomics, DLD). Subsequently, 18 models were developed by combining features and model types (random forest and support vector machine). RESULTS AND CONCLUSION: Under OD, O(DLR_DLD) demonstrated exceptional performance, with an optimal area under the curve (AUC) of 0.81 and an average AUC of 0.71. Within SD, S(DLR_DLD) surpassed the other models, achieving an optimal AUC of 0.90 and an average AUC of 0.85. Therefore, the integration of dosiomics into radiomics can augment predictive efficacy. The dose-segmentation strategy can facilitate the extraction of more profound information. This indicates that ScoreDLR and ScoreMDR were negatively associated with XM, whereas ScoreDLD, derived from SD exceeding 15 Gy, displayed a positive association with XM. For feature extraction, deep learning was superior to manual definition.

A magnetic resonance imaging-based lymph node regression grading scheme for nasopharyngeal carcinoma after radiotherapy.

Background: Among patients with nasopharyngeal carcinoma (NPC), there is no established method to distinguish between patients with residual disease that may eventually progress and those who have achieved cured. We thus aimed to assess the prognostic value of magnetic resonance imaging (MRI)-based lymph node regression grade (LRG) in the risk stratification of patients with NPC following radiotherapy (RT). Methods: This study retrospectively enrolled 387 patients newly diagnosed with NPC between January 2010 and January 2013. A four-category MRI-LRG system based on the areal analysis of RT-induced fibrosis and residual tumor was established. Univariate analysis was performed using the Kaplan-Meier method, and comparisons were conducted via the log-rank test. Multivariate analyses were conducted using Cox regression models to calculate the hazard ratios (HRs) with 95% confidence intervals (CIs) and adjusted P values. Survival curves were calculated using the Kaplan-Meier method and compared using the log-rank test. Results: The sum of MRI-LRG scores (LRG-sum) was an independent prognostic factor for progression-free survival (PFS) (HR 2.50, 95% CI: 1.28-4.90; P<0.001). LRG-sum ≤9 and >9 showed a poorer 5-year PFS rate than did LRG-sum ≤2 (66.1%, 42.9%, and 77.6%, respectively; P<0.001). A survival clustering analysis-based decision tree model showed more complex interactions among LRG-sum and pretreatment and post-RT Epstein-Barr virus (EBV) DNA, yielding four patient clusters with differentiated disease progression risks (5-year PFS rates of 89.5%, 76.4%, 57.6%, and 27.8%, respectively), which showed better risk stratification than did post-RT EBV DNA alone (P<0.001). Conclusions: The MRI-LRG system adds prognostic information and is a potentially reliable, noninvasive means to stratify treatment modalities for patients with NPC.

Predicting distant metastasis in nasopharyngeal carcinoma using gradient boosting tree model based on detailed magnetic resonance imaging reports.

BACKGROUND: Development of distant metastasis (DM) is a major concern during treatment of nasopharyngeal carcinoma (NPC). However, studies have demonstrated improved distant control and survival in patients with advanced NPC with the addition of chemotherapy to concomitant chemoradiotherapy. Therefore, precise prediction of metastasis in patients with NPC is crucial. AIM: To develop a predictive model for metastasis in NPC using detailed magnetic resonance imaging (MRI) reports. METHODS: This retrospective study included 792 patients with non-distant metastatic NPC. A total of 469 imaging variables were obtained from detailed MRI reports. Data were stratified and randomly split into training (50%) and testing sets. Gradient boosting tree (GBT) models were built and used to select variables for predicting DM. A full model comprising all variables and a reduced model with the top-five variables were built. Model performance was assessed by area under the curve (AUC). RESULTS: Among the 792 patients, 94 developed DM during follow-up. The number of metastatic cervical nodes (30.9%), tumor invasion in the posterior half of the nasal cavity (9.7%), two sides of the pharyngeal recess (6.2%), tubal torus (3.3%), and single side of the parapharyngeal space (2.7%) were the top-five contributors for predicting DM, based on their relative importance in GBT models. The testing AUC of the full model was 0.75 (95% confidence interval [CI]: 0.69-0.82). The testing AUC of the reduced model was 0.75 (95%CI: 0.68-0.82). For the whole dataset, the full (AUC = 0.76, 95%CI: 0.72-0.82) and reduced models (AUC = 0.76, 95%CI: 0.71-0.81) outperformed the tumor node-staging system (AUC = 0.67, 95%CI: 0.61-0.73). CONCLUSION: The GBT model outperformed the tumor node-staging system in predicting metastasis in NPC. The number of metastatic cervical nodes was identified as the principal contributing variable.

A novel fully automatic segmentation and counting system for metastatic lymph nodes on multimodal magnetic resonance imaging: Evaluation and prognostic implications in nasopharyngeal carcinoma.

BACKGROUND: The number of metastatic lymph nodes (MLNs) is crucial for the survival of nasopharyngeal carcinoma (NPC), but manual counting is laborious. This study aims to explore the feasibility and prognostic value of automatic MLNs segmentation and counting. METHODS: We retrospectively enrolled 980 newly diagnosed patients in the primary cohort and 224 patients from two external cohorts. We utilized the nnUnet model for automatic MLNs segmentation on multimodal magnetic resonance imaging. MLNs counting methods, including manual delineation-assisted counting (MDAC) and fully automatic lymph node counting system (AMLNC), were compared with manual evaluation (Gold standard). RESULTS: In the internal validation group, the MLNs segmentation results showed acceptable agreement with manual delineation, with a mean Dice coefficient of 0.771. The consistency among three counting methods was as follows 0.778 (Gold vs. AMLNC), 0.638 (Gold vs. MDAC), and 0.739 (AMLNC vs. MDAC). MLNs numbers were categorized into three-category variable (1-4, 5-9, > 9) and two-category variable (<4, ≥ 4) based on the gold standard and AMLNC. These categorical variables demonstrated acceptable discriminating abilities for 5-year overall survival (OS), progression-free, and distant metastasis-free survival. Compared with base prediction model, the model incorporating two-category AMLNC-counting numbers showed improved C-indexes for 5-year OS prediction (0.658 vs. 0.675, P = 0.045). All results have been successfully validated in the external cohort. CONCLUSIONS: The AMLNC system offers a time- and labor-saving approach for fully automatic MLNs segmentation and counting in NPC. MLNs counting using AMLNC demonstrated non-inferior performance in survival discrimination compared to manual detection.

Induction-concurrent chemoradiotherapy with or without sintilimab in patients with locoregionally advanced nasopharyngeal carcinoma in China (CONTINUUM): a multicentre, open-label, parallel-group, randomised, controlled, phase 3 trial.

BACKGROUND: Anti-PD-1 therapy and chemotherapy is a recommended first-line treatment for recurrent or metastatic nasopharyngeal carcinoma, but the role of PD-1 blockade remains unknown in patients with locoregionally advanced nasopharyngeal carcinoma. We assessed the addition of sintilimab, a PD-1 inhibitor, to standard chemoradiotherapy in this patient population. METHODS: This multicentre, open-label, parallel-group, randomised, controlled, phase 3 trial was conducted at nine hospitals in China. Adults aged 18-65 years with newly diagnosed high-risk non-metastatic stage III-IVa locoregionally advanced nasopharyngeal carcinoma (excluding T3-4N0 and T3N1) were eligible. Patients were randomly assigned (1:1) using blocks of four to receive gemcitabine and cisplatin induction chemotherapy followed by concurrent cisplatin radiotherapy (standard therapy group) or standard therapy with 200 mg sintilimab intravenously once every 3 weeks for 12 cycles (comprising three induction, three concurrent, and six adjuvant cycles to radiotherapy; sintilimab group). The primary endpoint was event-free survival from randomisation to disease recurrence (locoregional or distant) or death from any cause in the intention-to-treat population. Secondary endpoints included adverse events. This trial is registered with ClinicalTrials.gov (NCT03700476) and is now completed; follow-up is ongoing. FINDINGS: Between Dec 21, 2018, and March 31, 2020, 425 patients were enrolled and randomly assigned to the sintilimab (n=210) or standard therapy groups (n=215). At median follow-up of 41·9 months (IQR 38·0-44·8; 389 alive at primary data cutoff [Feb 28, 2023] and 366 [94%] had at least 36 months of follow-up), event-free survival was higher in the sintilimab group compared with the standard therapy group (36-month rates 86% [95% CI 81-90] vs 76% [70-81]; stratified hazard ratio 0·59 [0·38-0·92]; p=0·019). Grade 3-4 adverse events occurred in 155 (74%) in the sintilimab group versus 140 (65%) in the standard therapy group, with the most common being stomatitis (68 [33%] vs 64 [30%]), leukopenia (54 [26%] vs 48 [22%]), and neutropenia (50 [24%] vs 46 [21%]). Two (1%) patients died in the sintilimab group (both considered to be immune-related) and one (<1%) in the standard therapy group. Grade 3-4 immune-related adverse events occurred in 20 (10%) patients in the sintilimab group. INTERPRETATION: Addition of sintilimab to chemoradiotherapy improved event-free survival, albeit with higher but manageable adverse events. Longer follow-up is necessary to determine whether this regimen can be considered as the standard of care for patients with high-risk locoregionally advanced nasopharyngeal carcinoma. FUNDING: National Natural Science Foundation of China, Key-Area Research and Development Program of Guangdong Province, Natural Science Foundation of Guangdong Province, Overseas Expertise Introduction Project for Discipline Innovation, Guangzhou Municipal Health Commission, and Cancer Innovative Research Program of Sun Yat-sen University Cancer Center. TRANSLATION: For the Chinese translation of the abstract see Supplementary Materials section.

Radiomics-based lymph nodes prognostic models from three MRI regions in nasopharyngeal carcinoma.

Accurate prediction of the prognosis of nasopharyngeal carcinoma (NPC) is important for treatment. Lymph nodes metastasis is an important predictor for distant failure and regional recurrence in patients with NPC. Traditionally, subjective radiological evaluation increases concerns regarding the accuracy and consistency of predictions. Radiomics is an objective and quantitative evaluation algorithm for medical images. This retrospective analysis was conducted based on the data of 729 patients newly diagnosed with NPC without distant metastases to evaluate the performance of radiomics pretreatment using magnetic resonance imaging (MRI)-determined metastatic lymph nodes models to predict NPC prognosis with three delineation methods. Radiomics features were extracted from all lymph nodes (ALN), largest lymph node (LLN), and largest slice of the largest lymph node (LSLN) to generate three radiomics signatures. The radiomics signatures, clinical model, and radiomics-clinic merged models were developed in training cohort for predicting overall survival (OS). The results showed that LSLN signature with clinical factors predicted OS with high accuracy and robustness using pretreatment MR-determined metastatic lymph nodes (C-index [95 % confidence interval]: 0.762[0.760-0.763]), providing a new tool for treatment planning in NPC.

Radiomics-based nomogram guides adaptive de-intensification in locoregionally advanced nasopharyngeal carcinoma following induction chemotherapy.

OBJECTIVES: This study aimed to construct a radiomics-based model for prognosis and benefit prediction of concurrent chemoradiotherapy (CCRT) versus intensity-modulated radiotherapy (IMRT) in locoregionally advanced nasopharyngeal carcinoma (LANPC) following induction chemotherapy (IC). MATERIALS AND METHODS: A cohort of 718 LANPC patients treated with IC + IMRT or IC + CCRT were retrospectively enrolled and assigned to a training set (n = 503) and a validation set (n = 215). Radiomic features were extracted from pre-IC and post-IC MRI. After feature selection, a delta-radiomics signature was built with LASSO-Cox regression. A nomogram incorporating independent clinical indicators and the delta-radiomics signature was then developed and evaluated for calibration and discrimination. Risk stratification by the nomogram was evaluated with Kaplan-Meier methods. RESULTS: The delta-radiomics signature, which comprised 19 selected features, was independently associated with prognosis. The nomogram, composed of the delta-radiomics signature, age, T category, N category, treatment, and pre-treatment EBV DNA, showed great calibration and discrimination with an area under the receiver operator characteristic curve of 0.80 (95% CI 0.75-0.85) and 0.75 (95% CI 0.64-0.85) in the training and validation sets. Risk stratification by the nomogram, excluding the treatment factor, resulted in two groups with distinct overall survival. Significantly better outcomes were observed in the high-risk patients with IC + CCRT compared to those with IC + IMRT, while comparable outcomes between IC + IMRT and IC + CCRT were shown for low-risk patients. CONCLUSION: The radiomics-based nomogram can predict prognosis and survival benefits from concurrent chemotherapy for LANPC following IC. Low-risk patients determined by the nomogram may be potential candidates for omitting concurrent chemotherapy during IMRT. CLINICAL RELEVANCE STATEMENT: The radiomics-based nomogram was constructed for risk stratification and patient selection. It can help guide clinical decision-making for patients with locoregionally advanced nasopharyngeal carcinoma following induction chemotherapy, and avoid unnecessary toxicity caused by overtreatment. KEY POINTS: • The benefits from concurrent chemotherapy remained controversial for locoregionally advanced nasopharyngeal carcinoma following induction chemotherapy. • Radiomics-based nomogram achieved prognosis and benefits prediction of concurrent chemotherapy. • Low-risk patients defined by the nomogram were candidates for de-intensification.

Optimal cut-off value for identifying objective response in patients with nasopharyngeal carcinoma after induction chemotherapy.

BACKGROUND: We aimed to establish the most suitable threshold for objective response (OR) in the Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 in patients with nasopharyngeal carcinoma (NPC). METHODS: According to RECIST 1.1, we retrospectively evaluated MR images of NPC lesions in patients before and after induction chemotherapy (IC). Restricted cubic spline and maximally selected rank statistics were used to determine the cut-off value. Survival rates and differences between groups were compared with Kaplan-Meier curves and log-rank tests. RESULTS: Of 1126 patients, 365 cases who received IC treatment were suitable for RECIST 1.1 evaluation. The 20% cut-off value maximized between-group differences according to maximally selected rank statistics. No difference in distant metastasis-free survival between OR and non-response groups was shown using the primary threshold of OR (30%), while it differed when 20% was employed. CONCLUSIONS: With an optimal cut-off value of 20%, RECIST may assist clinicians to accurately evaluate disease response in NPC patients.

Value of radiological depth of invasion in non-pT4 Oral tongue squamous cell carcinoma: implication for preoperative MR T-staging.

OBJECTIVE: The prognostic stratification for oral tongue squamous cell carcinoma (OTSCC) is heavily based on postoperative pathological depth of invasion (pDOI). This study aims to propose a preoperative MR T-staging system based on tumor size for non-pT4 OTSCC. METHODS: Retrospectively, 280 patients with biopsy-confirmed, non-metastatic, pT1-3 OTSCC, treated between January 2010 and December 2017, were evaluated. Multiple MR sequences, including axial T2-weighted imaging (WI), unenhanced T1WI, and axial, fat-suppressed coronal, and sagittal contrast-enhanced (CE) T1WI, were utilized to measure radiological depth of invasion (rDOI), tumor thickness, and largest diameter. Intra-class correlation (ICC) and univariate and multivariate analyses were used to evaluate measurement reproducibility, and factors' significance, respectively. Cutoff values were established using an exhaustive method. RESULTS: Intra-observer (ICC = 0.81-0.94) and inter-observer (ICC = 0.79-0.90) reliability were excellent for rDOI measurements, and all measurements were significantly associated with overall survival (OS) (all p < .001). Measuring the rDOI on axial CE-T1WI with cutoffs of 8 mm and 12 mm yielded an optimal MR T-staging system for rT1-3 disease (5-year OS of rT1 vs rT2 vs rT3: 94.0% vs 72.8% vs 57.5%). Using multivariate analyses, the proposed T-staging exhibited increasingly worse OS (hazard ratio of rT2 and rT3 versus rT1, 3.56 [1.35-9.6], p = .011; 4.33 [1.59-11.74], p = .004; respectively), which outperformed pathological T-staging based on nonoverlapping Kaplan-Meier curves and improved C-index (0.682 vs. 0.639, p < .001). CONCLUSIONS: rDOI is a critical predictor of OTSCC mortality and facilitates preoperative prognostic stratification, which should be considered in future oral subsite MR T-staging. CLINICAL RELEVANCE STATEMENT: Utilizing axial CE-T1WI, an MR T-staging system for non-pT4 OTSCC was developed by employing rDOI measurement with optimal thresholds of 8 mm and 12 mm, which is comparable with pathological staging and merits consideration in future preoperative oral subsite planning. KEY POINTS: • Tumor morphology, measuring sequences, and observers could impact MR-derived measurements and compromise the consistency with histology. • MR-derived measurements, including radiological depth of invasion (rDOI), tumor thickness, and largest diameter, have a prognostic impact on OS (all p < .001). • rDOI with cutoffs of 8 mm and 12 mm on axial CE-T1WI is an optimal predictor of OS and could facilitate risk stratification in non-pT4 OTSCC disease.

TdDS-UNet: top-down deeply supervised U-Net for the delineation of 3D colorectal cancer.

Automatically delineating colorectal cancers with fuzzy boundaries from 3D images is a challenging task, but the problem of fuzzy boundary delineation in existing deep learning-based methods have not been investigated in depth. Here, an encoder-decoder-based U-shaped network (U-Net) based on top-down deep supervision (TdDS) was designed to accurately and automatically delineate the fuzzy boundaries of colorectal cancer. TdDS refines the semantic targets of the upper and lower stages by mapping ground truths that are more consistent with the stage properties than upsampling deep supervision. This stage-specific approach can guide the model to learn a coarse-to-fine delineation process and improve the delineation accuracy of fuzzy boundaries by gradually shrinking the boundaries. Experimental results showed that TdDS is more customizable and plays a role similar to the attentional mechanism, and it can further improve the capability of the model to delineate colorectal cancer contours. A total of 103, 12, and 29 3D pelvic magnetic resonance imaging volumes were used for training, validation, and testing, respectively. The comparative results indicate that the proposed method exhibits the best comprehensive performance, with a dice similarity coefficient (DSC) of 0.805 ± 0.053 and a hausdorff distance (HD) of 9.28 ± 5.14 voxels. In the delineation performance analysis section also showed that 44.49% of the delineation results are satisfactory and do not require revisions. This study can provide new technical support for the delineation of 3D colorectal cancer. Our method is open source, and the code is available athttps://github.com/odindis/TdDS/tree/main.

LD-UNet: A long-distance perceptual model for segmentation of blurred boundaries in medical images.

The blurriness of boundaries in medical image target regions hinders further improvement in automatic segmentation accuracy and is a challenging problem. To address this issue, we propose a model called long-distance perceptual UNet (LD-UNet), which has a powerful long-|distance perception ability and can effectively perceive the semantic context of an entire image. Specifically, LD-UNet utilizes global and local long-distance induction modules, which endow the model with contextual semantic induction capabilities for long-distance feature dependencies. The modules perform long-distance semantic perception at the high and low stages of LD-UNet, respectively, effectively improving the accuracy of local blurred information assessment. We also propose a top-down deep supervision method to enhance the ability of the model to fit data. Then, extensive experiments on four types of tumor data with blurred boundaries are conducted. The dataset includes nasopharyngeal carcinoma, esophageal carcinoma, pancreatic carcinoma, and colorectal carcinoma. The dice similarity coefficient scores obtained by LD-UNet on the four datasets are 73.35%, 85.93%, 70.04%, and 82.71%. Experimental results demonstrate that LD-UNet is more effective in improving the segmentation accuracy of blurred boundary regions than other methods with long-distance perception, such as transformers. Among all models, LD-UNet achieves the best performance. By visualizing the feature dependency field of the models, we further explore the advantages of LD-UNet in segmenting blurred boundaries.

Semi-automatic fine delineation scheme for pancreatic cancer.

BACKGROUND: Pancreatic cancer fine delineation in medical images by physicians is a major challenge due to the vast volume of medical images and the variability of patients. PURPOSE: A semi-automatic fine delineation scheme was designed to assist doctors in accurately and quickly delineating the cancer target region to improve the delineation accuracy of pancreatic cancer in computed tomography (CT) images and effectively reduce the workload of doctors. METHODS: A target delineation scheme in image blocks was also designed to provide more information for the deep learning delineation model. The start and end slices of the image block were manually delineated by physicians, and the cancer in the middle slices were accurately segmented using a three-dimensional Res U-Net model. Specifically, the input of the network is the CT image of the image block and the delineation of the cancer in the start and end slices, while the output of the network is the cancer area in the middle slices of the image block. Meanwhile, the model performance of pancreatic cancer delineation and the workload of doctors in different image block sizes were studied. RESULTS: We used 37 3D CT volumes for training, 11 volumes for validating and 11 volumes for testing. The influence of different image block sizes on doctors' workload was compared quantitatively. Experimental results showed that the physician's workload was minimal when the image block size was 5, and all cancer could be accurately delineated. The Dice similarity coefficient was 0.894 ± 0.029, the 95% Hausdorff distance was 3.465 ± 0.710 mm, the normalized surface Dice was 0.969 ± 0.019. By completing the accurate delineation of all the CT images, the speed of the new method is 2.16 times faster than that of manual sketching. CONCLUSION: Our proposed 3D semi-automatic delineative method based on the idea of block prediction could accurately delineate CT images of pancreatic cancer and effectively deal with the challenges of class imbalance, background distractions, and non-rigid geometrical features. This study had a significant advantage in reducing doctors' workload, and was expected to help doctors improve their work efficiency in clinical application.

MRI-based radiomic models to predict surgical margin status and infer tumor immune microenvironment in breast cancer patients with breast-conserving surgery: a multicenter validation study.

OBJECTIVES: Accurate preoperative estimation of the risk of breast-conserving surgery (BCS) resection margin positivity would be beneficial to surgical planning. In this multicenter validation study, we developed an MRI-based radiomic model to predict the surgical margin status. METHODS: We retrospectively collected preoperative breast MRI of patients undergoing BCS from three hospitals (SYMH, n = 296; SYSUCC, n = 131; TSPH, n = 143). Radiomic-based model for risk prediction of the margin positivity was trained on the SYMH patients (7:3 ratio split for the training and testing cohorts), and externally validated in the SYSUCC and TSPH cohorts. The model was able to stratify patients into different subgroups with varied risk of margin positivity. Moreover, we used the immune-radiomic models and epithelial-mesenchymal transition (EMT) signature to infer the distribution patterns of immune cells and tumor cell EMT status under different marginal status. RESULTS: The AUCs of the radiomic-based model were 0.78 (0.66-0.90), 0.88 (0.79-0.96), and 0.76 (0.68-0.84) in the testing cohort and two external validation cohorts, respectively. The actual margin positivity rates ranged between 0-10% and 27.3-87.2% in low-risk and high-risk subgroups, respectively. Positive surgical margin was associated with higher levels of EMT and B cell infiltration in the tumor area, as well as the enrichment of B cells, immature dendritic cells, and neutrophil infiltration in the peritumoral area. CONCLUSIONS: This MRI-based predictive model can be used as a reliable tool to predict the risk of margin positivity of BCS. Tumor immune-microenvironment alteration was associated with surgical margin status. CLINICAL RELEVANCE STATEMENT: This study can assist the pre-operative planning of BCS. Further research on the tumor immune microenvironment of different resection margin states is expected to develop new margin evaluation indicators and decipher the internal mechanism. KEY POINTS: • The MRI-based radiomic prediction model (CSS model) incorporating features extracted from multiple sequences and segments could estimate the margin positivity risk of breast-conserving surgery. • The radiomic score of the CSS model allows risk stratification of patients undergoing breast-conserving surgery, which could assist in surgical planning. • With the help of MRI-based radiomics to estimate the components of the immune microenvironment, for the first time, it is found that the margin status of breast-conserving surgery is associated with the infiltration of immune cells in the microenvironment and the EMT status of breast tumor cells.

Optimal Size Threshold for MRI-Detected Retropharyngeal Lymph Nodes to Predict Outcomes in Nasopharyngeal Carcinoma: A Two-Center Study.

BACKGROUND. Retropharyngeal lymph node (RLN) metastases have profound prognostic implications in patients with nasopharyngeal carcinoma (NPC). However, the AJCC staging system does not specify a size threshold for determining RLN involvement, resulting in inconsistent thresholds in practice. OBJECTIVE. The purpose of this article was to determine the optimal size threshold for determining the presence of metastatic RLNs on MRI in patients with NPC, in terms of outcome predictions. METHODS. This retrospective study included 1752 patients (median age, 46 years; 1297 men, 455 women) with NPC treated by intensity-modulated radiotherapy (RT) from January 2010 to March 2014 from two hospitals; 438 patients underwent MRI 3-4 months after treatment. Two radiologists measured the minimal axial diameter (MAD) of the largest RLN for each patient using a consensus process. A third radiologist measured MAD in 260 randomly selected patients to assess interobserver agreement. Initial ROC and restricted cubic spline (RCS) analyses were used to derive an optimal MAD threshold for predicting progression-free survival (PFS). The threshold's predictive utility was assessed in multivariable Cox regression analyses, controlling for standard clinical predictors. The threshold's utility for predicting PFS and overall survival (OS) was compared with a 5-mm threshold using Kaplan-Meier curves and log-rank tests. RESULTS. The intraclass correlation coefficient for MAD was 0.943. ROC and RCS analyses yielded an optimal threshold of 6 mm. In multivariable analyses, MAD of 6 mm and greater independently predicted PFS in all patients (HR = 1.35, p = .02), patients with N0 or N1 disease (HR = 1.80, p = .008), and patients who underwent posttreatment MRI (HR = 1.68, p = .04). In patients with N1 disease without cervical lymph node involvement, 5-year PFS was worse for MAD greater than or equal to 6 mm than for MAD that was greater than or equal to 5 mm but less than 6 mm (77.2% vs 89.7%, p = .03). OS was significantly different in patients with stage I and stage II disease defined using a 6-mm threshold (p = .04), but not using a 5-mm threshold (p = .09). The 5-year PFS rate was associated with a post-RT MAD of 6 mm and greater (HR = 1.68, p = .04) but not a post-RT MAD greater than or equal to 5 mm (HR = 1.09, p = .71). CONCLUSION. The findings support a threshold MAD of 6 mm for determining RLN involvement in patients with NPC. CLINICAL IMPACT. Future AJCC staging updates should consider incorporation of the 6-mm threshold for N-category and tumor-stage determinations.

A Rulefit-based prognostic analysis using structured MRI report to select potential beneficiaries from induction chemotherapy in advanced nasopharyngeal carcinoma: A dual-centre study.

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

MSU-Net: Multi-scale Sensitive U-Net based on pixel-edge-region level collaborative loss for nasopharyngeal MRI segmentation.

Radiotherapy is the traditional treatment of early nasopharyngeal carcinoma (NPC). Automatic accurate segmentation of risky lesions in the nasopharynx is crucial in radiotherapy. U-Net has been proved its effective medical image segmentation ability. However, the great difference in the structure and size of nasopharynx among different patients requires a network that pays more attention to multi-scale information. In this paper, we propose a multi-scale sensitive U-Net (MSU-Net) based on pixel-edge-region level collaborative loss (LCo-PER) for NPC segmentation task. A series of novel feature fusion modules based on spatial continuity and multi-scale semantic are proposed for extracting multi-level features while efficiently searching for all size lesions. A spatial continuity information extraction module (SCIEM) is proposed for effectively using the spatial continuity information of context slices to search small lesions. And a multi-scale semantic feature extraction module (MSFEM) is proposed for extracting features of different receptive fields. LCo-PER is proposed for the network training which makes network model could take into account the size of different lesions. The global Dice, Precision, Recall and IOU of the testing set are 84.50%, 97.48%, 84.33% and 82.41%, respectively. The results show that our method is better than the other state-of-the-art methods for NPC segmentation which obtain higher accuracy and effective segmentation performance.

Evolutionary route of nasopharyngeal carcinoma metastasis and its clinical significance.

It is critical to understand factors associated with nasopharyngeal carcinoma (NPC) metastasis. To track the evolutionary route of metastasis, here we perform an integrative genomic analysis of 163 matched blood and primary, regional lymph node metastasis and distant metastasis tumour samples, combined with single-cell RNA-seq on 11 samples from two patients. The mutation burden, gene mutation frequency, mutation signature, and copy number frequency are similar between metastatic tumours and primary and regional lymph node tumours. There are two distinct evolutionary routes of metastasis, including metastases evolved from regional lymph nodes (lymphatic route, 61.5%, 8/13) and from primary tumours (hematogenous route, 38.5%, 5/13). The hematogenous route is characterised by higher IFN-γ response gene expression and a higher fraction of exhausted CD8+ T cells. Based on a radiomics model, we find that the hematogenous group has significantly better progression-free survival and PD-1 immunotherapy response, while the lymphatic group has a better response to locoregional radiotherapy.

Prognostic potential of a voxelwise invasion risk map of nasopharyngeal carcinoma based on a coordinate system of the nasopharynx.

Background: Tumor invasion risk (TIR) is an important prognostic factor in nasopharyngeal carcinoma (NPC). We propose a novel prognostic analytic method for NPC based on a voxelwise analysis of TIR in a coordinate system of the nasopharynx. Methods: A stable nasopharynx coordinate system was constructed based on anatomical landmarks to obtain an accurate TIR profile for NPC. The coordinate system was validated by image registration of the lateral pterygoid muscle (LPM). The tumors were registered to the coordinate system through shift, scale, and rotation transformations. The voxelwise TIR map for NPC was obtained by superposition of all registered and mirrored tumor regions of interest. The minimum risk (MinR) point of the tumor region was used as an independent prognostic factor for NPC. The cutoff value was calculated with density plot and validated with restricted cubic splines (RCSs), and then the patients were divided into 2 groups for overall survival (OS) analysis. Results: The first voxelwise TIR map of NPC was obtained based on 778 patients. The OS of patients with a low TIR was 76.8% and was 92.6% for patients with a high TIR [P<0.001; hazard ratio (HR) =1/0.45; 95% CI: 0.27-0.77; adjusted P=0.004]. Thus, patients with a low TIR had a poor prognosis, whereas patients with a high TIR had a good prognosis. The MinR may be better at grading the prognosis of patients compared to the American Joint Committee on Cancer (AJCC) staging or tumor/node (T/N) classification systems. Conclusions: The voxelwise TIR map provides a new method for the prognostic analysis of NPC. Potential clinical applications of voxelwise TIR mapping are clinical target volume (CTV) delineation and dose-painting for NPC.