Outcome prediction of SSTR-RADS-3A and SSTR-RADS-3B lesions in patients with neuroendocrine tumors based on 68Ga-DOTATATE PET/MR.

PURPOSE: Somatostatin receptor (SSTR)-targeted PET imaging has emerged as a common approach to evaluating those patients with well-differentiated neuroendocrine tumors (NETs). The SSTR reporting and data system (SSTR-RADS) version 1.0 provides a means of categorizing lesions from 1 to 5 according to the likelihood of NET involvement, with SSTR-RADS-3A (soft-tissue) and SSTR-RADS-3B (bone) lesions being those suggestive of but without definitive NET involvement. The goal of the present study was to assess the ability of 68Ga-DOTATATE PET/MR imaging data to predict outcomes for indeterminate SSTR-RADS-3A and 3B lesions. METHODS: NET patients with indeterminate SSTR-RADS-3A or SSTR-RADS-3B lesions who underwent 68Ga-DOTATATE PET/MR imaging from April 2020 through August 2023 were retrospectively evaluated. All patients underwent follow-up through December 2023 (median, 17 months; (3-31 months)), with imaging follow-up or biopsy findings ultimately being used to classify lesions as malignant or benign. Lesion maximum standardized uptake value (SUVmax) along with minimum and mean apparent diffusion coefficient (ADCmin and ADCmean) values were measured and assessed for correlations with outcomes on follow-up. RESULTS: In total, 33 indeterminate SSTR-RADS-3 lesions from 22 patients (19 SSTR-RADS-3A and 14 SSTR-RADS-3B) were identified based upon baseline 68Ga-DOTATATE PET/MR findings. Over the course of follow-up, 16 of these lesions (48.5%) were found to exhibit true NET positivity, including 9 SSTR-RADS-3A and 7 SSTR-RADS-3B lesions. For SSTR-RADS-3A lymph nodes, a diameter larger than 0.7 cm and an ADCmin of 779 × 10-6mm2/s or lower were identified as being more likely to be associated with metastatic lesions. Significant differences in ADCmin and ADCmean were identified when comparing metastatic and non-metastatic SSTR-RADS-3B bone lesions (P < 0.05), with these parameters offering a high predictive ability (AUC = 0.94, AUC = 0.86). CONCLUSION: Both diameter and ADCmin can aid in the accurate identification of the nature of lesions associated with SSTR-RADS-3A lymph nodes, whereas ADCmin and ADCmean values can inform the accurate interpretation of SSTR-RADS-3B bone lesions.

ACP-ESM2: The prediction of anticancer peptides based on pre-trained classifier.

Anticancer peptides (ACPs) are a type of protein molecule that has anti-cancer activity and can inhibit cancer cell growth and survival. Traditional classification approaches for ACPs are expensive and time-consuming. This paper proposes a pre-trained classifier model, ESM2-GRU, for ACP prediction to make it easier to predict ACPs, gain a better understanding of the structural and functional differences of anti-cancer peptides, and optimize the design for the development of more effective anti-cancer treatment strategies. The model is made up of the ESM2 pre-trained model, a bidirectional GRU recurrent neural network, and a fully connected layer. ACP sequences are first fed into the ESM2 model, which then expands the dimensions before feeding the findings back into the bidirectional GRU recurrent neural network. Finally, the fully connected layer generates the ultimate output. Experimental validation demonstrates that the ESM2-GRU model greatly improves classification performance on the benchmark dataset ACP606, with AUC, ACC, and MCC values of 0.975, 0.852, and 0.738, respectively. This exceptional prediction potential helps to identify specific types of anti-cancer peptides, improving their targeting and selectivity and, therefore, furthering the development of tailored medicine and treatments.

Combined measurement of serum zinc with PSA ameliorates prostate cancer screening efficiency via support vector machine algorithms.

Background: Early screening of prostate cancer (PCa) is pivotal but challenging in the clinical scenario due to the phenomena of false positivity or false negativity of some serological evaluations, e.g. PSA testing. Decline of serum Zn2+ levels in PCa patients reportedly plays a crucial role in early screening of PCa. Accordingly, we combined 4 indices comprising the serum levels of total PSA (tPSA), free PSA (fPSA), Zn2+ and demographic information (especially age) in order to ameliorate the efficacies of PCa screening with support vector machine (SVM) algorithms. Methods: A total of 858 male patients with prostate disorders and 345 healthy male controls were enrolled. Patients' data included 4 variables and serum Zn2+ was quantified via a self-invented Zn2+ responsive AIE-based fluorescent probe as previously published. tPSA and fPSA were routinely determined by a chemiluminescent method. Mathematical simulations were conducted to establish a SVM model for the combined diagnostics with the four variables. Moreover, ROC and its characteristic AUC were also employed to evaluate the classification efficacy of the model. Sigmoid function was utilized to estimate corresponding probabilities of classifying the clinical subjects as per 5 grades, which were incorporated into our established prostate index (PI) stratification system. Results: In SVM model, the mean AUC of the ROC with the quartet of variables was approximately 84% for PCa diagnosis, whereas the mean AUC of the ROCs with tPSA, fPSA, [Zn2+] or age alone was 64%, 62%, 55% and 59%, respectively. We further established an integrated prostate index (PI) stratification system with 5 grades and a software package to support clinicians in predicting PCa, with the accuracy of our risk stratification system being 83.3%, 91.6% and 83.3% in predicting normal, benign and PCa cases in corresponding groups. Follow-up findings especially MRI results and PI-RADS scores supported the reliability of this stratification platform as well. Conclusion: Findings from our present study demonstrated that index combination via SVM algorithms may well facilitate clinicians in early differential screening of PCa. Meanwhile, our established PI stratification system based on SVM model and Sigmoid function provided substantial accuracy in preclinical risk prediction of developing prostate cancer.

LIPH contributes to glycolytic phenotype in pancreatic ductal adenocarcinoma by activating LPA/LPAR axis and maintaining ALDOA stability.

BACKGROUND: LIPH, a membrane-associated phosphatidic acid-selective phospholipase A1a, can produce LPA (Lysophosphatidic acid) from PA (Phosphatidic acid) on the outer leaflet of the plasma membrane. It is well known that LIPH dysfunction contributes to lipid metabolism disorder. Previous study shows that LIPH was found to be a potential gene related to poor prognosis with pancreatic ductal adenocarcinoma (PDAC). However, the biological functions of LIPH in PDAC remain unclear. METHODS: Cell viability assays were used to evaluate whether LIPH affected cell proliferation. RNA sequencing and immunoprecipitation showed that LIPH participates in tumor glycolysis by stimulating LPA/LPAR axis and maintaining aldolase A (ALDOA) stability in the cytosol. Subcutaneous, orthotopic xenograft models and patient-derived xenograft PDAC model were used to evaluate a newly developed Gemcitabine-based therapy. RESULTS: LIPH was significantly upregulated in PDAC and was related to later pathological stage and poor prognosis. LIPH downregulation in PDAC cells inhibited colony formation and proliferation. Mechanistically, LIPH triggered PI3K/AKT/HIF1A signaling via LPA/LPAR axis. LIPH also promoted glycolysis and de novo synthesis of glycerolipids by maintaining ALDOA stability in the cytosol. Xenograft models show that PDAC with high LIPH expression levels was sensitive to gemcitabine/ki16425/aldometanib therapy without causing discernible side effects. CONCLUSION: LIPH directly bridges PDAC cells and tumor microenvironment to facilitate aberrant aerobic glycolysis via activating LPA/LPAR axis and maintaining ALDOA stability, which provides an actionable gemcitabine-based combination therapy with limited side effects.

Increased Uptake in Microcystic Serous Cystadenoma Mimicking Pancreatic Neuroendocrine Tumor on 68 Ga-DOTATATE PET/MRI.

ABSTRACT: A 2.6-cm solid cystic lesion in the pancreatic head was found in a 51-year-old woman on CT. A pancreatic neuroendocrine tumor was suspected, and a 68 Ga-DOTATATE PET/MRI was performed, which showed increased tracer uptake in the lesion. However, postsurgical pathologic examination indicated a pancreatic serous cystadenoma. Here, we reported a case of microcystic pancreatic serous cystadenoma that could be misdiagnosed as a pancreatic neuroendocrine tumor on a 68 Ga-DOTATATE PET/MRI.

Multi-center evaluation of machine learning-based radiomic model in predicting disease free survival and adjuvant chemotherapy benefit in stage II colorectal cancer patients.

BACKGROUND: Our study aimed to explore the potential of radiomics features derived from CT images in predicting the prognosis and response to adjuvant chemotherapy (ACT) in patients with Stage II colorectal cancer (CRC). METHODS: A total of 478 patients with confirmed stage II CRC, with 313 from Shanghai (Training set) and 165 from Beijing (Validation set) were enrolled. Optimized features were selected using GridSearchCV and Iterative Feature Elimination (IFE) algorithm. Subsequently, we developed an ensemble random forest classifier to predict the probability of disease relapse.We evaluated the performance of the model using the concordance index (C-index), precision-recall curves, and area under the precision-recall curves (AUCPR). RESULTS: A radiomic model (namely the RF5 model) consisting of four radiomics features and T stage were developed. The RF5 model performed better than simple radiomics features or T stage alone, with higher C-index and AUCPR, as well as better sensitivity and specificity (C-indexRF5: 0.836; AUCPR = 0.711; Sensitivity = 0.610; Specificity = 0.935). We identified an optimal cutoff value of 0.1215 to split patients into high- or low-score subgroups, with those in the low-score group having better disease-free survival (DFS) (Training Set: P = 1.4e-11; Validation Set: P = 0.015). Furthermore, patients in the high-score group who received ACT had better DFS compared to those who did not receive ACT (P = 0.04). However, no statistical difference was found in low-score patients (P = 0.17). CONCLUSION: The radiomic model can serve as a reliable tool for assessing prognosis and identifying the optimal candidates for ACT in Stage II CRC patients. TRIAL REGISTRATION: Retrospectively registered.

The value of 18F-fluorodeoxyglucose positron emission tomography/magnetic resonance whole-body scans and local enhancement scans in the preoperative staging and resectability assessment of pancreatic adenocarcinoma.

Background: In addition to contrast-enhanced multiphase computed tomography (CT) and magnetic resonance imaging (MRI), integrated positron emission tomography (PET)/magnetic resonance (MR) is increasingly being used for the preoperative evaluation of pancreatic cancer. The purpose of this study was to explore the value of hybrid 18F-fluorodeoxyglucose (18F-FDG) PET/MR imaging in preoperative assessment and treatment decision-making. Methods: A retrospective data collection (of imaging, clinical, and pathological information) was conducted on patients who underwent 18F-FDG PET/MR with clinically diagnosed or suspected pancreatic cancer from March 2018 to March 2022 in Ruijin Hospital. The data of 76 patients were initially included, with 1 of the 76 patients eventually being excluded due to a misdiagnosis of inflammatory disease. Of the 75 patients, 38 underwent pancreatic tumor resection and 10 underwent laparoscopic exploration. The accuracy of 18F-FDG PET/MR for pancreatic cancer staging and the assessment of pancreatic resectability was evaluated based on pathological results, intraoperative findings, and documented final clinical stages of illness. The adjustments to patient treatment plans were also analyzed before and after 18F-FDG PET/MR examination. Results: The accuracy of clinical tumor node metastasis (TNM) staging of pancreatic cancer by 18F-FDG PET/MR was 73.3% (55/75). The area under the curve (AUC) of 18F-FDG PET/MR for diagnosing the advanced stage (III-IV) versus the nonadvanced stage (I-II) of disease was 0.922 [95% confidence interval (CI): 0.852-0.993]. The treatment regimen of 20.0% (15/75) of patients was impacted. The accuracy of the evaluation of the resectability of pancreatic cancer with 18F-FDG PET/MR was 91.9% (34/37). With the surgical and pathological results being used as a reference, the overall accuracy of preoperative 18F-FDG PET/MR for T staging was 62.2%, and the AUC for diagnosing T4 versus T1-3 was 0.872 (95% CI: 0.660-1.000). Conclusions: 18F-FDG PET/MR performs well in diagnosing advanced pancreatic cancer and thus may impact the treatment decisions for a considerable number of patients. 18F-FDG PET/MR has a high level of accuracy in evaluating the resectability of pancreatic cancer before surgery.

DRD2 expression based on 18F-fallypride PET/MR predicts the dopamine agonist resistance of prolactinomas: a pilot study.

PURPOSE: The dopamine agonists (DA) have been used widely to treat prolactinomas. However, it is difficult to predict whether the patient will be responsive to DA treatment. METHODS: We aimed to investigate whether the in vivo expression of DRD2 based on 18F-fallypride PET/MR could predict the therapeutic effect of DA on prolactinomas. Seven patients with prolactinomas completed 18F-fallypride PET/MR. Among them, three patients underwent surgery and further tumor immunohistochemistry. Imaging findings and immunohistochemical staining were compared with treatment outcomes. RESULTS: 18F-fallypride PET/MR was visually positive in 7 of 7 patients, and DRD2 target specificity could be confirmed by immunohistochemical staining. A significantly lower tracer standard uptake value (SUV) could be detected in the resistant patients (n = 3) than in the sensitive patients (n = 4; SUVmean, 4.67 ± 1.32 vs. 13.57 ± 2.42, p < 0.05). DRD2 expression determined by 18F-fallypride PET/MR corresponded with the DA treatment response. CONCLUSION: 18F-fallypride PET/MR may be a promising technique for predicting DA response in patients with prolactinoma.

[Grading of pancreatic neuroendocrine neoplasms with imaging: an update].

Imaging evaluation of pancreatic neuroendocrine neoplasms is developing with the updating of pathological grading. Tumor size, blood perfusion and apparent diffusion coefficient(ADC)value are the most commonly used imaging indicators for the evaluation of the malignancy of pancreatic neuroendocrine neoplasms, as well as other imaging characteristics representing the tumor invasiveness. Dual-probe imaging with SSR-PET and FDG-PET can further improve the accuracy of diagnosis and classification of pancreatic neuroendocrine neoplasms, and provide further information for the tumor treatment and prognosis. Quantitative analysis including texture analysis and radiomics is also a new topic in the research of pancreatic neuroendocrine neoplasms. The combination of morphology, functional imaging and quantitative analysis may contribute to the non-invasive evaluation of the malignancy of pancreatic neuroendocrine neoplasms.

Model-Driven Deep Learning Method for Pancreatic Cancer Segmentation Based on Spiral-Transformation.

Pancreatic cancer is a lethal malignant tumor with one of the worst prognoses. Accurate segmentation of pancreatic cancer is vital in clinical diagnosis and treatment. Due to the unclear boundary and small size of cancers, it is challenging to both manually annotate and automatically segment cancers. Considering 3D information utilization and small sample sizes, we propose a model-driven deep learning method for pancreatic cancer segmentation based on spiral transformation. Specifically, a spiral-transformation algorithm with uniform sampling was developed to map 3D images onto 2D planes while preserving the spatial relationship between textures, thus addressing the challenge in effectively applying 3D contextual information in a 2D model. This study is the first to introduce spiral transformation in a segmentation task to provide effective data augmentation, alleviating the issue of small sample size. Moreover, a transformation-weight-corrected module was embedded into the deep learning model to unify the entire framework. It can achieve 2D segmentation and corresponding 3D rebuilding constraint to overcome non-unique 3D rebuilding results due to the uniform and dense sampling. A smooth regularization based on rebuilding prior knowledge was also designed to optimize segmentation results. The extensive experiments showed that the proposed method achieved a promising segmentation performance on multi-parametric MRIs, where T2, T1, ADC, DWI images obtained the DSC of 65.6%, 64.0%, 64.5%, 65.3%, respectively. This method can provide a novel paradigm to efficiently apply 3D information and augment sample sizes in the development of artificial intelligence for cancer segmentation. Our source codes will be released at https://github.com/SJTUBME-QianLab/ Spiral-Segmentation.

Utilizing GCN and Meta-Learning Strategy in Unsupervised Domain Adaptation for Pancreatic Cancer Segmentation.

Automated pancreatic cancer segmentation is highly crucial for computer-assisted diagnosis. The general practice is to label images from selected modalities since it is expensive to label all modalities. This practice brought about a significant interest in learning the knowledge transfer from the labeled modalities to unlabeled ones. However, the imaging parameter inconsistency between modalities leads to a domain shift, limiting the transfer learning performance. Therefore, we propose an unsupervised domain adaptation segmentation framework for pancreatic cancer based on GCN and meta-learning strategy. Our model first transforms the source image into a target-like visual appearance through the synergistic collaboration between image and feature adaptation. Specifically, we employ encoders incorporating adversarial learning to separate domain-invariant features from domain-specific ones to achieve visual appearance translation. Then, the meta-learning strategy with good generalization capabilities is exploited to strike a reasonable balance in the training of the source and transformed images. Thus, the model acquires more correlated features and improve the adaptability to the target images. Moreover, a GCN is introduced to supervise the high-dimensional abstract features directly related to the segmentation outcomes, and hence ensure the integrity of key structural features. Extensive experiments on four multi-parameter pancreatic-cancer magnetic resonance imaging datasets demonstrate improved performance in all adaptation directions, confirming our model's effectiveness for unlabeled pancreatic cancer images. The results are promising for reducing the burden of annotation and improving the performance of computer-aided diagnosis of pancreatic cancer. Our source codes will be released at https://github.com/SJTUBME-QianLab/UDAseg, once this manuscript is accepted for publication.

Differentiation of Hepatocellular Carcinoma from Hepatic Hemangioma and Focal Nodular Hyperplasia using Computed Tomographic Spectral Imaging.

BACKGROUND AND AIMS: Hepatocellular carcinoma (HCC) is the most common primary hepatic malignancy. This study was designed to investigate the value of computed tomography (CT) spectral imaging in differentiating HCC from hepatic hemangioma (HH) and focal nodular hyperplasia (FNH). METHODS: This was a retrospective study of 51 patients who underwent spectral multiple-phase CT at 40-140 keV during the arterial phase (AP) and portal venous phase (PP). Slopes of the spectral curves, iodine density, water density derived from iodine- and water-based material decomposition images, iodine uptake ratio (IUR), normalized iodine concentration, and the ratio of iodine concentration in liver lesions between AP and PP were measured or calculated. RESULTS: As energy level decreased, the CT values of HCC (n=31), HH (n=17), and FNH (n=7) increased in both AP and PP. There were significant differences in IUR in the AP, IUR in the PP, normalized iodine concentration in the AP, slope in the AP, and slope in the PP among HCC, HH, and FNH. The CT values in AP, IUR in the AP and PP, normalized iodine concentration in the AP, slope in the AP and PP had high sensitivity and specificity in differentiating HH and HCC from FNH. Quantitative CT spectral data had higher sensitivity and specificity than conventional qualitative CT image analysis during the combined phases. CONCLUSIONS: Mean CT values at low energy (40-90 keV) and quantitative analysis of CT spectral data (IUR in the AP) could be helpful in the differentiation of HCC, HH, and FNH.

Pathologic T1 and T2 encapsulated invasive carcinomas arising from mucinous cystic neoplasms of the pancreas have favorable prognosis and might be treated conservatively.

Carcinoma arising from a mucinous cystic neoplasm (MCN) of the pancreas is termed MCN with associated invasive carcinoma (MCN-AIC) in the fifth WHO classification of digestive tumors (2019). The prognosis of this malignancy varies depending on the relationship of the invasive carcinoma to the cyst capsule, but limited data are available. This study identified 165 surgically resected MCNs including 15 MCN-AICs from a single center between 2008 and 2018 and analyzed their clinicopathologic features. The results confirmed that non-invasive MCNs were completely cured by surgery. All MCN-AICs showing an encapsulated invasion pattern (defined as invasive carcinoma limited to the ovarian-type stroma, cystic septa, and capsule) had an excellent prognosis with a 5-year survival rate of 100%, even when the size of the invasive component was up to stage T2. By contrast, MCN-AICs with extracapsular involvement had unfavorable clinical outcomes. Our study demonstrates that the pattern of invasion of MCN-AIC can predict patient prognosis. Pathologic stage T1 and T2 encapsulated MCN-AICs may be completely cured with surgical resection alone or when combined with postoperative chemotherapy.

Differentiating TP53 Mutation Status in Pancreatic Ductal Adenocarcinoma Using Multiparametric MRI-Derived Radiomics.

OBJECTIVES: This study assessed the preoperative prediction of TP53 status based on multiparametric magnetic resonance imaging (mpMRI) radiomics extracted from two-dimensional (2D) and 3D images. METHODS: 57 patients with pancreatic cancer who underwent preoperative MRI were included. The diagnosis and TP53 gene test were based on resections. Of the 57 patients included 37 mutated TP53 genes and the remaining 20 had wild-type TP53 genes. Two radiologists performed manual tumour segmentation on seven different MRI image acquisition sequences per patient, including multi-phase [pre-contrast, late arterial phase (ap), portal venous phase, and delayed phase] dynamic contrast enhanced (DCE) T1-weighted imaging, T2-weighted imaging (T2WI), Diffusion-weighted imaging (DWI), and apparent diffusion coefficient (ADC). PyRadiomics-package was used to generate 558 two-dimensional (2D) and 994 three-dimensional (3D) image features. Models were constructed by support vector machine (SVM) for differentiating TP53 status and DX score method were used for feature selection. The evaluation of the model performance included area under the curve (AUC), accuracy, calibration curves, and decision curve analysis. RESULTS: The 3D ADC-ap-DWI-T2WI model with 11 selected features yielded the best performance for differentiating TP53 status, with accuracy = 0.91 and AUC = 0.96. The model showed the good calibration. The decision curve analysis indicated that the radiomics model had clinical utility. CONCLUSIONS: A non-invasive and quantitative mpMRI-based radiomics model can accurately predict TP53 mutation status in pancreatic cancer patients and contribute to the precision treatment.

Diffuse large B-cell lymphoma segmentation in PET-CT images via hybrid learning for feature fusion.

PURPOSE: Diffuse large B-cell lymphoma (DLBCL) is an aggressive type of lymphoma with high mortality and poor prognosis that especially has a high incidence in Asia. Accurate segmentation of DLBCL lesions is crucial for clinical radiation therapy. However, manual delineation of DLBCL lesions is tedious and time-consuming. Automatic segmentation provides an alternative solution but is difficult for diffuse lesions without the sufficient utilization of multimodality information. Our work is the first study focusing on positron emission tomography and computed tomography (PET-CT) feature fusion for the DLBCL segmentation issue. We aim to improve the fusion performance of complementary information contained in PET-CT imaging with a hybrid learning module in the supervised convolutional neural network. METHODS: First, two encoder branches extract single-modality features, respectively. Next, the hybrid learning component utilizes them to generate spatial fusion maps which can quantify the contribution of complementary information. Such feature fusion maps are then concatenated with specific-modality (i.e., PET and CT) feature maps to obtain a representation of the final fused feature maps in different scales. Finally, the reconstruction part of our network creates a prediction map of DLBCL lesions by integrating and up-sampling the final fused feature maps from encoder blocks in different scales. RESULTS: The ability of our method was evaluated to detect foreground and segment lesions in three independent body regions (nasopharynx, chest, and abdomen) of a set of 45 PET-CT scans. Extensive ablation experiments compared our method to four baseline techniques for multimodality fusion (input-level (IL) fusion, multichannel (MC) strategy, multibranch (MB) strategy, and quantitative weighting (QW) fusion). The results showed that our method achieved a high detection accuracy (99.63% in the nasopharynx, 99.51% in the chest, and 99.21% in the abdomen) and had the superiority in segmentation performance with the mean dice similarity coefficient (DSC) of 73.03% and the modified Hausdorff distance (MHD) of 4.39 mm, when compared with the baselines (DSC: IL: 53.08%, MC: 63.59%, MB: 69.98%, and QW: 72.19%; MHD: IL: 12.16 mm, MC: 6.46 mm, MB: 4.83 mm, and QW: 4.89 mm). CONCLUSIONS: A promising segmentation method has been proposed for the challenging DLBCL lesions in PET-CT images, which improves the understanding of complementary information by feature fusion and may guide clinical radiotherapy. The statistically significant analysis based on P-value calculation has indicated a degree of significant difference between our proposed method and other baselines (almost metrics: P < 0.05). This is a preliminary research using a small sample size, and we will collect data continually to achieve the larger verification study.

Weakly supervised deep learning for determining the prognostic value of 18F-FDG PET/CT in extranodal natural killer/T cell lymphoma, nasal type.

PURPOSE: To develop a weakly supervised deep learning (WSDL) method that could utilize incomplete/missing survival data to predict the prognosis of extranodal natural killer/T cell lymphoma, nasal type (ENKTL) based on pretreatment 18F-FDG PET/CT results. METHODS: One hundred and sixty-seven patients with ENKTL who underwent pretreatment 18F-FDG PET/CT were retrospectively collected. Eighty-four patients were followed up for at least 2 years (training set = 64, test set = 20). A WSDL method was developed to enable the integration of the remaining 83 patients with incomplete/missing follow-up information in the training set. To test generalization, these data were derived from three types of scanners. Prediction similarity index (PSI) was derived from deep learning features of images. Its discriminative ability was calculated and compared with that of a conventional deep learning (CDL) method. Univariate and multivariate analyses helped explore the significance of PSI and clinical features. RESULTS: PSI achieved area under the curve scores of 0.9858 and 0.9946 (training set) and 0.8750 and 0.7344 (test set) in the prediction of progression-free survival (PFS) with the WSDL and CDL methods, respectively. PSI threshold of 1.0 could significantly differentiate the prognosis. In the test set, WSDL and CDL achieved prediction sensitivity, specificity, and accuracy of 87.50% and 62.50%, 83.33% and 83.33%, and 85.00% and 75.00%, respectively. Multivariate analysis confirmed PSI to be an independent significant predictor of PFS in both the methods. CONCLUSION: The WSDL-based framework was more effective for extracting 18F-FDG PET/CT features and predicting the prognosis of ENKTL than the CDL method.

Pancreatic Neuroendocrine Neoplasms: CT Spectral Imaging in Grading.

RATIONALE AND OBJECTIVES: The purpose of this study was to define the CT spectral imaging characteristics of pancreatic neuroendocrine neoplasms (PNENs) and evaluate their potential for differential diagnosis of nonlow grade (non-LG) PNENs from low grade (LG) PNENs. MATERIALS AND METHODS: CT spectral imaging data of 54 pathologically proven PNENs were retrospectively reviewed. Patients were divided into two groups: 40 cases with grade 1 in LG PNENs group and 14 cases with grade 2 and grade 3 in non-LG PNENs group. RESULTS: Gender, calcification, inhomogeneity, invasiveness, PD dilatation, lymph node enlargement, size, normalized iodine (water) concentration in arterial phase (AP) (Iodine (ap)), normalized effective-Z (Zap), slope of normalized CT spectral curves in both AP, and portal venous phase were found to be significant variables for differentiating non-LG PNENs from LG PNENs (p < 0.05). Non-LG PNENs had larger size and lower Zap and Iodine (ap) than LG PNENs. The tumor size, Zap and Iodine (ap) had fair to good diagnostic performance with the area under receiver-operating-characteristic curve (AUC) 0.843, 0.733, and 0.728, respectively. Multivariate analysis with logistic regression had higher AUC (p<0.05) than all the single parameters except for size. CONCLUSION: There were significant differences in CT spectral imaging parameters between non-LG and LG PNENs. Tumor size was the most promising independent parameter and the combination of quantitative parameters with qualitative parameters is the best predictor in differentiating of non-LG PNENs from LG PNENs. CT spectral imaging can help determine the malignancy of PNENs, which can better assist in surgical planning.

Combined Spiral Transformation and Model-Driven Multi-Modal Deep Learning Scheme for Automatic Prediction of TP53 Mutation in Pancreatic Cancer.

Pancreatic cancer is a malignant form of cancer with one of the worst prognoses. The poor prognosis and resistance to therapeutic modalities have been linked to TP53 mutation. Pathological examinations, such as biopsies, cannot be frequently performed in clinical practice; therefore, noninvasive and reproducible methods are desired. However, automatic prediction methods based on imaging have drawbacks such as poor 3D information utilization, small sample size, and ineffectiveness multi-modal fusion. In this study, we proposed a model-driven multi-modal deep learning scheme to overcome these challenges. A spiral transformation algorithm was developed to obtain 2D images from 3D data, with the transformed image inheriting and retaining the spatial correlation of the original texture and edge information. The spiral transformation could be used to effectively apply the 3D information with less computational resources and conveniently augment the data size with high quality. Moreover, model-driven items were designed to introduce prior knowledge in the deep learning framework for multi-modal fusion. The model-driven strategy and spiral transformation-based data augmentation can improve the performance of the small sample size. A bilinear pooling module was introduced to improve the performance of fine-grained prediction. The experimental results show that the proposed model gives the desired performance in predicting TP53 mutation in pancreatic cancer, providing a new approach for noninvasive gene prediction. The proposed methodologies of spiral transformation and model-driven deep learning can also be used for the artificial intelligence community dealing with oncological applications. Our source codes with a demon will be released at https://github.com/SJTUBME-QianLab/SpiralTransform.

Comparison of Nasopharyngeal MR, 18 F-FDG PET/CT, and 18 F-FDG PET/MR for Local Detection of Natural Killer/T-Cell Lymphoma, Nasal Type.

OBJECTIVES: The present study aims to compare the diagnostic efficacy of MR, 18F-FDG PET/CT, and 18F-FDG PET/MR for the local detection of early-stage extranodal natural killer/T-cell lymphoma, nasal type (ENKTL). PATIENTS AND METHODS: Thirty-six patients with histologically proven early-stage ENKTL were enrolled from a phase 2 study (Cohort A). Eight nasopharyngeal anatomical regions from each patient were imaged using 18F-FDG PET/CT and MR. A further nine patients were prospectively enrolled from a multicenter, phase 3 study; these patients underwent 18F-FDG PET/CT and PET/MR after a single 18F-FDG injection (Cohort B). Region-based sensitivity and specificity were calculated. The standardized uptake values (SUV) obtained from PET/CT and PET/MR were compared, and the relationship between the SUV and apparent diffusion coefficients (ADC) of PET/MR were analyzed. RESULTS: In Cohort A, of the 288 anatomic regions, 86 demonstrated lymphoma involvement. All lesions were detected by 18F-FDG PET/CT, while only 70 were detected by MR. 18F-FDG PET/CT exhibited a higher sensitivity than MR (100% vs. 81.4%, χ2 = 17.641, P < 0.001) for local detection of malignancies. The specificity of 18F-FDG PET/CT and MR were 98.5 and 97.5%, respectively (χ2 = 0.510, P = 0.475). The accuracy of 18F-FDG PET/CT was 99.0% and the accuracy of MR was 92.7% (χ2 = 14.087, P < 0.001). In Cohort B, 72 anatomical regions were analyzed. PET/CT and PET/MR have a sensitivity of 100% and a specificity of 92.5%. The two methods were consistent (κ = 0.833, P < 0.001). There was a significant correlation between PET/MR SUVmax and PET/CT SUVmax (r = 0.711, P < 0.001), and SUVmean (r = 0.685, P < 0.001). No correlation was observed between the SUV and the ADC. CONCLUSION: In early-stage ENKTL, nasopharyngeal MR showed a lower sensitivity and a similar specificity when compared with 18F-FDG PET/CT. PET/MR showed similar performance compared with PET/CT.

Performance of 18F-FDG PET/CT Radiomics for Predicting EGFR Mutation Status in Patients With Non-Small Cell Lung Cancer.

OBJECTIVE: To assess the performance of pretreatment 18F-fluorodeoxyglucose positron emission tomography/computed tomography (18F-FDG PET/CT) radiomics features for predicting EGFR mutation status in patients with non-small cell lung cancer (NSCLC). PATIENTS AND METHODS: We enrolled total 173 patients with histologically proven NSCLC who underwent preoperative 18F-FDG PET/CT. Tumor tissues of all patients were tested for EGFR mutation status. A PET/CT radiomics prediction model was established through multi-step feature selection. The predictive performances of radiomics model, clinical features and conventional PET-derived semi-quantitative parameters were compared using receiver operating curves (ROCs) analysis. RESULTS: Four CT and two PET radiomics features were finally selected to build the PET/CT radiomics model. Compared with area under the ROC curve (AUC) equal to 0.664, 0.683 and 0.662 for clinical features, maximum standardized uptake values (SUVmax) and total lesion glycolysis (TLG), the PET/CT radiomics model showed better performance to discriminate between EGFR positive and negative mutations with the AUC of 0.769 and the accuracy of 67.06% after 10-fold cross-validation. The combined model, based on the PET/CT radiomics and clinical feature (gender) further improved the AUC to 0.827 and the accuracy to 75.29%. Only one PET radiomics feature demonstrated significant but low predictive ability (AUC = 0.661) for differentiating 19 Del from 21 L858R mutation subtypes. CONCLUSIONS: EGFR mutations status in patients with NSCLC could be well predicted by the combined model based on 18F-FDG PET/CT radiomics and clinical feature, providing an alternative useful method for the selection of targeted therapy.