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.

An interpretable deep learning model for identifying the morphological characteristics of dMMR/MSI-H gastric cancer.

Accurate tumor diagnosis by pathologists relies on identifying specific morphological characteristics. However, summarizing these unique morphological features in tumor classifications can be challenging. Although deep learning models have been extensively studied for tumor classification, their indirect and subjective interpretation obstructs pathologists from comprehending the model and discerning the morphological features accountable for classifications. In this study, we introduce a new approach utilizing Style Generative Adversarial Networks, which enables a direct interpretation of deep learning models to detect significant morphological characteristics within datasets representing patients with deficient mismatch repair/microsatellite instability-high gastric cancer. Our approach effectively identifies distinct morphological features crucial for tumor classification, offering valuable insights for pathologists to enhance diagnostic accuracy and foster professional growth.

A deep learning-based semiautomated workflow for triaging follow-up MR scans in treated nasopharyngeal carcinoma.

It is imperative to optimally utilize virtues and obviate defects of fully automated analysis and expert knowledge in new paradigms of healthcare. We present a deep learning-based semiautomated workflow (RAINMAN) with 12,809 follow-up scans among 2,172 patients with treated nasopharyngeal carcinoma from three centers (ChiCTR.org.cn, Chi-CTR2200056595). A boost of diagnostic performance and reduced workload was observed in RAINMAN compared with the original manual interpretations (internal vs. external: sensitivity, 2.5% [p = 0.500] vs. 3.2% [p = 0.031]; specificity, 2.9% [p < 0.001] vs. 0.3% [p = 0.302]; workload reduction, 79.3% vs. 76.2%). The workflow also yielded a triaging performance of 83.6%, with increases of 1.5% in sensitivity (p = 1.000) and 0.6%-1.3% (all p < 0.05) in specificity compared to three radiologists in the reader study. The semiautomated workflow shows its unique superiority in reducing radiologist's workload by eliminating negative scans while retaining the diagnostic performance of radiologists.

Deep learning-based recurrence detector on magnetic resonance scans in nasopharyngeal carcinoma: A multicenter study.

OBJECTIVES: Accuracy in the detection of recurrent nasopharyngeal carcinoma (NPC) on follow-up magnetic resonance (MR) scans needs to be improved. MATERIAL AND METHODS: A total of 5 035 follow-up MR scans from 5 035 survivors with treated NPC between April 2007 and July 2020 were retrospectively collected from three cancer centers for developing and evaluating the deep learning (DL) model MODERN (MR-based Deep learning model for dEtecting Recurrent Nasopharyngeal carcinoma). In a reader study with 220 scans, the accuracy of two radiologists in detecting recurrence on scans with vs without MODERN was evaluated. The performance was measured using the area under the receiver operating characteristic curve (ROC-AUC) and accuracy with a 95% confidence interval (CI). RESULTS: MODERN exhibited sound performance in the validation cohort (internal: ROC-AUC, 0.88, 95% CI, 0.86-0.90; external 1: ROC-AUC, 0.88, 95% CI, 0.86-0.90; external 2: ROC-AUC, 0.85, 95% CI, 0.82-0.88). In a reader study, MODERN alone achieved reliable accuracy compared to that of radiologists (MODERN: 84.1%, 95% CI, 79.3%-88.9%; competent: 80.9%, 95% CI, 75.7%-86.1%, P < 0.001; expert: 85.9%, 95% CI, 81.3%-90.5%, P < 0.001). The accuracy of radiologists was boosted by the MODERN score (competent with MODERN score: 84.6%, 95% CI, 79.8%-89.3%, P < 0.001; expert with MODERN score: 87.7%, 95% CI, 83.4%-92.1%, P < 0.001). CONCLUSION: We developed a DL model for recurrence detection with reliable performance. Computer-human collaboration has the potential to refine the workflow in interpreting surveillant MR scans among patients with treated NPC.

The contrast-enhanced MRI can be substituted by unenhanced MRI in identifying and automatically segmenting primary nasopharyngeal carcinoma with the aid of deep learning models: An exploratory study in large-scale population of endemic area.

BACKGROUND AND OBJECTIVES: Administration of contrast is not desirable for all cases in clinical setting, and no consensus in sequence selection for deep learning model development has been achieved, thus we aim to explore whether contrast-enhanced magnetic resonance imaging (ceMRI) can be substituted in the identification and segmentation of nasopharyngeal carcinoma (NPC) with the aid of deep learning models in a large-scale cohort. METHODS: A total of 4478 eligible individuals were randomly split into training, validation and test sets, and self-constrained 3D DenseNet and V-Net models were developed using axial T1-weighted imaging (T1WI), T2WI or enhanced T1WI (T1WIC) images separately. The differential diagnostic performance between NPC and benign hyperplasia were compared among models using chi-square test. Segmentation evaluation metrics, including dice similarity coefficient (DSC) and average surface distance (ASD), were compared using paired student's t-test between T1WIC and T1WI or T2WI models or M_T1/T2, a merged output of malignant region derived from T1WI and T2WI models. RESULTS: All models exhibited similar satisfactory diagnostic performance in discriminating NPC from benign hyperplasia, all attaining overall accuracy over 99.00% in all T stages of NPC. And T1WIC model exhibited similar average DSC and ASD with those of M_T1/T2 (DSC, 0.768±0.070 vs 0.764±0.070; ASD, 1.573±10.954 mm vs 1.626±10.975 mm 1.626±0.975 mm vs 1.573±0.954 mm, all p > 0.0167) in primary NPC using DenseNet, but yielded a significantly higher DSC and lower ASD than either T1WI model or T2WI model (DSC, 0.759±0.065 or 0.755±0.071; ASD, 1.661±0.898 mm or 1.722±1.133 mm, respectively, all p < 0.01) in the entire test set of NPC cohort. Moreover, the average DSCs and ASDs were not statistically significant between T1WIC model and M_T1/T2 in both.

Prognostic and Treatment Guiding Significance of MRI-Based Tumor Burden Features and Nodal Necrosis in Nasopharyngeal Carcinoma.

We aimed to develop a nomogram integrating MRI-based tumor burden features (MTBF), nodal necrosis, and some clinical factors to forecast the distant metastasis-free survival (DMFS) of patients suffering from non-metastatic nasopharyngeal carcinoma (NPC). A total of 1640 patients treated at Sun Yat-sen University Cancer Center (Guangzhou, China) from 2011 to 2016 were enrolled, among which 1148 and 492 patients were randomized to a training cohort and an internal validation cohort, respectively. Additionally, 200 and 257 patients were enrolled in the Foshan and Dongguan validation cohorts, respectively, which served as independent external validation cohorts. The MTBF were developed from the stepwise regression of six multidimensional tumor burden variables, based on which we developed a nomogram also integrating nodal necrosis and clinical features. This model divided the patients into high- and low-risk groups by an optimal cutoff. Compared with those of patients in the low-risk group, the DMFS [hazard ratio (HR): 4.76, 95% confidence interval (CI): 3.39-6.69; p < 0.0001], and progression-free survival (PFS; HR: 4.11, 95% CI: 3.13-5.39; p < 0.0001) of patients in the high-risk group were relatively poor. Furthermore, in the training cohort, the 3-year DMFS of high-risk patients who received induction chemotherapy (ICT) combined with concurrent chemoradiotherapy (CCRT) was better than that of those who were treated with CCRT alone (p = 0.0340), whereas low-risk patients who received ICT + CCRT had a similar DMFS to those who only received CCRT. The outcomes we obtained were all verified in the three validation cohorts. The survival model can be used as a reliable prognostic tool for NPC patients and is helpful to determine patients who will benefit from ICT.

Deep learning for risk prediction in patients with nasopharyngeal carcinoma using multi-parametric MRIs.

BACKGROUND: Magnetic resonance images (MRI) is the main diagnostic tool for risk stratification and treatment decision in nasopharyngeal carcinoma (NPC). However, the holistic feature information of multi-parametric MRIs has not been fully exploited by clinicians to accurately evaluate patients. OBJECTIVE: To help clinicians fully utilize the missed information to regroup patients, we built an end-to-end deep learning model to extract feature information from multi-parametric MRIs for predicting and stratifying the risk scores of NPC patients. METHODS: In this paper, we proposed an end-to-end multi-modality deep survival network (MDSN) to precisely predict the risk of disease progression of NPC patients. Extending from 3D dense net, this proposed MDSN extracted deep representation from multi-parametric MRIs (T1w, T2w, and T1c). Moreover, deep features and clinical stages were integrated through MDSN to more accurately predict the overall risk score (ORS) of individual NPC patient. RESULT: A total of 1,417 individuals treated between January 2012 and December 2014 were included for training and validating the end-to-end MDSN. Results were then tested in a retrospective cohort of 429 patients included in the same institution. The C-index of the proposed method with or without clinical stages was 0.672 and 0.651 on the test set, respectively, which was higher than the that of the stage grouping (0.610). CONCLUSIONS: The C-index of the model which integrated clinical stages with deep features is 0.062 higher than that of stage grouping alone (0.672 vs 0.610). We conclude that features extracted from multi-parametric MRIs based on MDSN can well assist the clinical stages in regrouping patients.

Development of a self-constrained 3D DenseNet model in automatic detection and segmentation of nasopharyngeal carcinoma using magnetic resonance images.

OBJECTIVES: We aimed to develop a dual-task model to detect and segment nasopharyngeal carcinoma (NPC) automatically in magnetic resource images (MRI) based on deep learning method, since the differential diagnosis of NPC and atypical benign hyperplasia was difficult and the radiotherapy target contouring of NPC was labor-intensive. MATERIALS AND METHODS: A self-constrained 3D DenseNet (SC-DenseNet) architecture was improved using separated training and validation sets. A total of 4100 individuals were finally enrolled and split into the training, validation and test sets at a proximate ratio of 8:1:1 using simple randomization. The diagnostic metrics of the established model against experienced radiologists was compared in the test set. The dice similarity coefficient (DSC) of manual and model-defined tumor region was used to evaluate the efficacy of segmentation. RESULTS: Totally, 3142 nasopharyngeal carcinoma (NPC) and 958 benign hyperplasia were included. The SC-DenseNet model showed encouraging performance in detecting NPC, attained a higher overall accuracy, sensitivity and specificity than those of the experienced radiologists (97.77% vs 95.87%, 99.68% vs 99.24% and 91.67% vs 85.21%, respectively). Moreover, the model also exhibited promising performance in automatic segmentation of tumor region in NPC, with an average DSC at 0.77 ± 0.07 in the test set. CONCLUSIONS: The SC-DenseNet model showed competence in automatic detection and segmentation of NPC in MRI, indicating the promising application value as an assistant tool in clinical practice, especially in screening project.

Real-time artificial intelligence for detection of upper gastrointestinal cancer by endoscopy: a multicentre, case-control, diagnostic study.

BACKGROUND: Upper gastrointestinal cancers (including oesophageal cancer and gastric cancer) are the most common cancers worldwide. Artificial intelligence platforms using deep learning algorithms have made remarkable progress in medical imaging but their application in upper gastrointestinal cancers has been limited. We aimed to develop and validate the Gastrointestinal Artificial Intelligence Diagnostic System (GRAIDS) for the diagnosis of upper gastrointestinal cancers through analysis of imaging data from clinical endoscopies. METHODS: This multicentre, case-control, diagnostic study was done in six hospitals of different tiers (ie, municipal, provincial, and national) in China. The images of consecutive participants, aged 18 years or older, who had not had a previous endoscopy were retrieved from all participating hospitals. All patients with upper gastrointestinal cancer lesions (including oesophageal cancer and gastric cancer) that were histologically proven malignancies were eligible for this study. Only images with standard white light were deemed eligible. The images from Sun Yat-sen University Cancer Center were randomly assigned (8:1:1) to the training and intrinsic verification datasets for developing GRAIDS, and the internal validation dataset for evaluating the performance of GRAIDS. Its diagnostic performance was evaluated using an internal and prospective validation set from Sun Yat-sen University Cancer Center (a national hospital) and additional external validation sets from five primary care hospitals. The performance of GRAIDS was also compared with endoscopists with three degrees of expertise: expert, competent, and trainee. The diagnostic accuracy, sensitivity, specificity, positive predictive value, and negative predictive value of GRAIDS and endoscopists for the identification of cancerous lesions were evaluated by calculating the 95% CIs using the Clopper-Pearson method. FINDINGS: 1 036 496 endoscopy images from 84 424 individuals were used to develop and test GRAIDS. The diagnostic accuracy in identifying upper gastrointestinal cancers was 0·955 (95% CI 0·952-0·957) in the internal validation set, 0·927 (0·925-0·929) in the prospective set, and ranged from 0·915 (0·913-0·917) to 0·977 (0·977-0·978) in the five external validation sets. GRAIDS achieved diagnostic sensitivity similar to that of the expert endoscopist (0·942 [95% CI 0·924-0·957] vs 0·945 [0·927-0·959]; p=0·692) and superior sensitivity compared with competent (0·858 [0·832-0·880], p<0·0001) and trainee (0·722 [0·691-0·752], p<0·0001) endoscopists. The positive predictive value was 0·814 (95% CI 0·788-0·838) for GRAIDS, 0·932 (0·913-0·948) for the expert endoscopist, 0·974 (0·960-0·984) for the competent endoscopist, and 0·824 (0·795-0·850) for the trainee endoscopist. The negative predictive value was 0·978 (95% CI 0·971-0·984) for GRAIDS, 0·980 (0·974-0·985) for the expert endoscopist, 0·951 (0·942-0·959) for the competent endoscopist, and 0·904 (0·893-0·916) for the trainee endoscopist. INTERPRETATION: GRAIDS achieved high diagnostic accuracy in detecting upper gastrointestinal cancers, with sensitivity similar to that of expert endoscopists and was superior to that of non-expert endoscopists. This system could assist community-based hospitals in improving their effectiveness in upper gastrointestinal cancer diagnoses. FUNDING: The National Key R&D Program of China, the Natural Science Foundation of Guangdong Province, the Science and Technology Program of Guangdong, the Science and Technology Program of Guangzhou, and the Fundamental Research Funds for the Central Universities.

A deep survival analysis method based on ranking.

Survival analyses of populations and the establishment of prognoses for individual patients are important activities in the practice of medicine. Standard survival models, such as the Cox proportional hazards model, require extensive feature engineering or prior knowledge to model at an individual level. Some survival analysis models can avoid these problems by using machine learning extended the CPH model, and higher performance has been reported. In this paper, we propose an innovative loss function that is defined as the sum of an extended mean squared error loss and a pairwise ranking loss based on ranking information on survival data. We apply this loss function to optimize a deep feed-forward neural network (RankDeepSurv), which can be used to model survival data. We demonstrate that the performance of our model, RankDeepSurv, is superior to that of other state-of-the-art survival models based on an analysis of 4 public medical clinical datasets. When modelling the prognosis of nasopharyngeal carcinoma (NPC), RankDeepSurv achieved better prognostic accuracy than the CPH established by clinical experts. The difference between high and low risk groups in the RankDeepSurv model is greater than the difference in the CPH. The results show that our method has considerable potential to model survival data in medical settings.

Correction to: Radiomics on multi-modalities MR sequences can subtype patients with non-metastatic nasopharyngeal carcinoma (NPC) into distinct survival subgroups.

The original version of this article, published on 14 March 2019, unfortunately contained a mistake.

Radiomics on multi-modalities MR sequences can subtype patients with non-metastatic nasopharyngeal carcinoma (NPC) into distinct survival subgroups.

OBJECTIVES: To explore and evaluate the feasibility of radiomics in stratifying nasopharyngeal carcinoma (NPC) into distinct survival subgroups through multi-modalities MRI. METHODS: A total of 658 patients (training cohort: 424; validation cohort: 234) with non-metastatic NPC were enrolled in the retrospective analysis. Each slice was considered as a sample and 4863 radiomics features on the tumor region were extracted from T1-weighted, T2-weighted, and contrast-enhanced T1-weighted MRI. Consensus clustering and manual aggregation were performed on the training cohort to generate a baseline model and classification reference used to train a support vector machine classifier. The risk of each patient was defined as the maximum risk among the slices. Each patient in the validation cohort was assigned to the risk model using the trained classifier. Harrell's concordance index (C-index) was used to measure the prognosis performance, and differences between subgroups were compared using the log-rank test. RESULTS: The training cohort was clustered into four groups with distinct survival patterns. Each patient was assigned to one of the four groups according to the estimated risk. Our method gave a performance (C-index = 0.827, p < .004 and C-index = 0.814, p < .002) better than the T-stage (C-index = 0.815, p = .002 and C-index = 0.803, p = .024), competitive to and more stable than the TNM staging system (C-index = 0.842, p = .003 and C-index = 0.765, p = .050) in the training cohort and the validation cohort. CONCLUSIONS: Through investigating a large one-institutional cohort, the quantitative multi-modalities MRI image phenotypes reveal distinct survival subtypes. KEY POINTS: • Radiomics phenotype of MRI revealed the subtype of nasopharyngeal carcinoma (NPC) patients with distinct survival patterns. • The slice-wise analysis method on MRI helps to stratify patients and provides superior prognostic performance over the TNM staging method. • Risk estimation using the highest risk among slices performed better than using the majority risk in prognosis.

The value of detailed MR imaging report of primary tumor and lymph nodes on prognostic nomograms for nasopharyngeal carcinoma after intensity-modulated radiotherapy.

PURPOSE: To establish effective prognostic nomograms using clinical features and detailed magnetic resonance imaging (MRI) findings for primary tumor and regional lymph nodes after intensity-modulated radiotherapy (IMRT) in patients with nasopharyngeal carcinoma. METHOD: The nomogram for overall survival (OS) was based on a retrospective study of 595 patients who underwent IMRT at Sun Yat-sen University Cancer Center from 2010 to 2012. The predictive accuracy and discriminative ability of our nomogram models were determined by concordance index and calibration curve, and were compared with the nomogram models combining clinical features with tumor-node-metastasis (TNM) classification. The results were validated using bootstrap resampling and a cohort study of 241 patients. The same data cohort was used to predict the progress-free survival (PFS) of nasopharyngeal carcinoma with 3:1 training cohort (N = 558) and validation cohort (N = 278). RESULTS: The following factors were assembled into our prognostic survival nomogram models: Age, Epstein-Barr virus DNA copy number before treatment (EBV_DNA_CN), tensor veli palatini (TVP) involvement, musculus pterygoideus lateralis (MPL) involvement, prestyloid space (PS) involvement, prevertebral space (PVS) involvement, base of sphenoid bone (BOSB) involvement, paranasal sinus (PNS) involvement, the laterality of Ⅱ (Ⅱ_laterality), the laterality of retropharyngeal lymph node (RPLN_laterality), nodal grouping (NG), extranodal neoplastic spread (ENS), contrast-enhancing rim (CER) and Nodal_number. The calibration curves showed good agreement between nomogram-predicted and actual survival. Our nomogram models for OS and PFS, provided better results than the nomogram models combining clinical features with TNM classification. Results were further confirmed in the validation set. CONCLUSION: Detailed MRI findings of primary tumor and regional lymph nodes can improve the performance of prognostic nomograms for patients with nasopharyngeal carcinoma.

Development and validation of an endoscopic images-based deep learning model for detection with nasopharyngeal malignancies.

BACKGROUND: Due to the occult anatomic location of the nasopharynx and frequent presence of adenoid hyperplasia, the positive rate for malignancy identification during biopsy is low, thus leading to delayed or missed diagnosis for nasopharyngeal malignancies upon initial attempt. Here, we aimed to develop an artificial intelligence tool to detect nasopharyngeal malignancies under endoscopic examination based on deep learning. METHODS: An endoscopic images-based nasopharyngeal malignancy detection model (eNPM-DM) consisting of a fully convolutional network based on the inception architecture was developed and fine-tuned using separate training and validation sets for both classification and segmentation. Briefly, a total of 28,966 qualified images were collected. Among these images, 27,536 biopsy-proven images from 7951 individuals obtained from January 1st, 2008, to December 31st, 2016, were split into the training, validation and test sets at a ratio of 7:1:2 using simple randomization. Additionally, 1430 images obtained from January 1st, 2017, to March 31st, 2017, were used as a prospective test set to compare the performance of the established model against oncologist evaluation. The dice similarity coefficient (DSC) was used to evaluate the efficiency of eNPM-DM in automatic segmentation of malignant area from the background of nasopharyngeal endoscopic images, by comparing automatic segmentation with manual segmentation performed by the experts. RESULTS: All images were histopathologically confirmed, and included 5713 (19.7%) normal control, 19,107 (66.0%) nasopharyngeal carcinoma (NPC), 335 (1.2%) NPC and 3811 (13.2%) benign diseases. The eNPM-DM attained an overall accuracy of 88.7% (95% confidence interval (CI) 87.8%-89.5%) in detecting malignancies in the test set. In the prospective comparison phase, eNPM-DM outperformed the experts: the overall accuracy was 88.0% (95% CI 86.1%-89.6%) vs. 80.5% (95% CI 77.0%-84.0%). The eNPM-DM required less time (40 s vs. 110.0 ± 5.8 min) and exhibited encouraging performance in automatic segmentation of nasopharyngeal malignant area from the background, with an average DSC of 0.78 ± 0.24 and 0.75 ± 0.26 in the test and prospective test sets, respectively. CONCLUSIONS: The eNPM-DM outperformed oncologist evaluation in diagnostic classification of nasopharyngeal mass into benign versus malignant, and realized automatic segmentation of malignant area from the background of nasopharyngeal endoscopic images.