Interpretable machine learning based on CT-derived extracellular volume fraction to predict pathological grading of hepatocellular carcinoma.

PURPOSE: To develop a non-invasive auxiliary assessment method based on CT-derived extracellular volume (ECV) to predict the pathological grading (PG) of hepatocellular carcinoma (HCC). METHODS: The study retrospectively analyzed 238 patients who underwent HCC resection surgery between January 2013 and April 2023. Six machine learning algorithms were employed to construct predictive models for HCC PG: logistic regression, extreme gradient boosting, Light Gradient Boosting Machine (LightGBM), random forest, adaptive boosting, and Gaussian naive Bayes. Model performance was evaluated using receiver operating characteristic curve analysis, including area under the curve (AUC), sensitivity, specificity, accuracy, positive predictive value, negative predictive value, and F1 score. Calibration plots were used for visual evaluation of model calibration. Clinical decision curve analysis was performed to assess potential clinical utility by calculating net benefit. RESULTS: 166 patients from Hospital A were allocated to the training set, while 72 patients from Hospital B (constituting 30.25% of the total sample) were assigned to the test set. The model achieved an AUC of 1.000 (95%CI: 1.000-1.000) in the training set and 0.927 (95%CI: 0.837-0.999) in the validation set, respectively. Ultimately, the model achieved an AUC of 0.909 (95%CI: 0.837-0.980) in the test set, with an accuracy of 0.778, sensitivity of 0.906, specificity of 0.789, negative predictive value of 0.556, and F1 score of 0.908. CONCLUSION: This study successfully developed and validated a non-invasive auxiliary assessment method based on CT-derived ECV to predict the HCC PG, providing important supplementary information for clinical decision-making.

Development and validation of a prognostic model incorporating tumor thrombus grading for nonmetastatic clear cell renal cell carcinoma with tumor thrombus: A multicohort study.

There is significant variability with respect to the prognosis of nonmetastatic clear cell renal cell carcinoma (ccRCC) patients with venous tumor thrombus (VTT). By applying multiregion whole-exome sequencing on normal-tumor-thrombus-metastasis quadruples from 33 ccRCC patients, we showed that metastases were mainly seeded by VTT (81.8%) rather than primary tumors (PTs). A total of 706 nonmetastatic ccRCC patients with VTT from three independent cohorts were included in this study. C-index analysis revealed that pathological grading of VTT outperformed other indicators in risk assessment (OS: 0.663 versus 0.501-0.610, 0.667 versus 0.544-0.651, and 0.719 versus 0.511-0.700 for Training, China-Validation, and Poland-Validation cohorts, respectively). We constructed a risk predicting model, TT-GPS score, based on four independent variables: VTT height, VTT grading, perinephric fat invasion, and sarcomatoid differentiation in PT. The TT-GPS score displayed better discriminatory ability (OS, c-index: 0.706-0.840, AUC: 0.788-0.874; DFS, c-index: 0.691-0.717, AUC: 0.771-0.789) than previously reported models in risk assessment. In conclusion, we identified for the first-time pathological grading of VTT as an unheeded prognostic factor. By incorporating VTT grading, the TT-GPS score is a promising prognostic tool in predicting the survival of nonmetastatic ccRCC patients with VTT.

18F-FDG-PET/CT-based deep learning model for fully automated prediction of pathological grading for pancreatic ductal adenocarcinoma before surgery.

BACKGROUND: The determination of pathological grading has a guiding significance for the treatment of pancreatic ductal adenocarcinoma (PDAC) patients. However, there is a lack of an accurate and safe method to obtain pathological grading before surgery. The aim of this study is to develop a deep learning (DL) model based on 18F-fluorodeoxyglucose-positron emission tomography/computed tomography (18F-FDG-PET/CT) for a fully automatic prediction of preoperative pathological grading of pancreatic cancer. METHODS: A total of 370 PDAC patients from January 2016 to September 2021 were collected retrospectively. All patients underwent 18F-FDG-PET/CT examination before surgery and obtained pathological results after surgery. A DL model for pancreatic cancer lesion segmentation was first developed using 100 of these cases and applied to the remaining cases to obtain lesion regions. After that, all patients were divided into training set, validation set, and test set according to the ratio of 5:1:1. A predictive model of pancreatic cancer pathological grade was developed using the features computed from the lesion regions obtained by the lesion segmentation model and key clinical characteristics of the patients. Finally, the stability of the model was verified by sevenfold cross-validation. RESULTS: The Dice score of the developed PET/CT-based tumor segmentation model for PDAC was 0.89. The area under curve (AUC) of the PET/CT-based DL model developed on the basis of the segmentation model was 0.74, with an accuracy, sensitivity, and specificity of 0.72, 0.73, and 0.72, respectively. After integrating key clinical data, the AUC of the model improved to 0.77, with its accuracy, sensitivity, and specificity boosted to 0.75, 0.77, and 0.73, respectively. CONCLUSION: To the best of our knowledge, this is the first deep learning model to end-to-end predict the pathological grading of PDAC in a fully automatic manner, which is expected to improve clinical decision-making.

Pre-operative MRI features predict early post-operative recurrence of hepatocellular carcinoma with different degrees of pathological differentiation.

PURPOSE: To investigate the value of pre-operative gadoxetate disodium (Gd-EOB-DTPA) enhanced MRI predicting early post-operative recurrence (< 2 years) of hepatocellular carcinoma (HCC) with different degrees of pathological differentiation. METHODS: Retrospective analysis of pre-operative MR imaging features of 177 patients diagnosed as suffering from HCC and that underwent radical resection. Multivariate logistic regression assessment was adopted to assess predictors for HCC recurrence with different degrees of pathological differentiation. The area under the curve (AUC) of receiver operating characteristics (ROC) was utilized to assess the diagnostic efficacy of the predictors. RESULTS: Among the 177 patients, 155 (87.5%) were males, 22 (12.5%) were females; the mean age was 49.97 ± 10.71 years. Among the predictors of early post-operative recurrence of highly-differentiated HCC were an unsmooth tumor margin and an incomplete/without tumor capsule (p = 0.037 and 0.033, respectively) whereas those of early post-operative recurrence of moderately-differentiated HCC were incomplete/without tumor capsule, peritumoral enhancement along with peritumoral hypointensity (p = 0.006, 0.046 and 0.004, respectively). The predictors of early post-operative recurrence of poorly-differentiated HCC were peritumoral enhancement, peritumoral hypointensity, and tumor thrombosis (p = 0.033, 0.006 and 0.021, respectively). The AUCs of the multi-predictor diagnosis of early post-operative recurrence of highly-, moderately-, and poorly-differentiated HCC were 0.841, 0.873, and 0.875, respectively. The AUCs of the multi-predictor diagnosis were each higher than for those predicted separately. CONCLUSIONS: The imaging parameters for predicting early post-operative recurrence of HCC with different degrees of pathological differentiation were different and combining these predictors can improve the diagnostic efficacy of early post-operative HCC recurrence.

Application of intravoxel incoherent motion diffusion-weighted imaging in hepatocellular carcinoma.

The morbidity and mortality of hepatocellular carcinoma (HCC) rank 6th and 4th, respectively, among malignant tumors worldwide. Traditional diffusion-weighted imaging (DWI) uses the apparent diffusion coefficient (ADC) obtained by applying the monoexponential model to reflect water molecule diffusion in active tissue; however, the value of ADC is affected by microcirculation perfusion. Using a biexponential model, intravoxel incoherent motion (IVIM)-DWI quantitatively measures information related to pure water molecule diffusion and microcirculation perfusion, thus compensating for the shortcomings of DWI. The number of studies examining the application of IVIM-DWI in patients with HCC has gradually increased over the last few years, and many results show that IVIM-DWI has vital value for HCC differentiation, pathological grading, and predicting and evaluating the treatment response. The present study principally reviews the principle of IVIM-DWI and its research progress in HCC differentiation, pathological grading, predicting and evaluating the treatment response, predicting postoperative recurrence and predicting gene expression prediction.

Analysis of DWI in the classification of glioma pathology and its therapeutic application in clinical surgery: a case-control study.

Background: Glioma is a common primary craniocerebral malignant tumor, due to the lack of specificity of imaging examination and clinical manifestations, its diagnostic accuracy is relatively low, which may result in misdiagnosis and missed diagnosis. The apparent diffusion coefficient (ADC) in magnetic resonance diffusion weighted imaging (DWI) can reflect the histological characteristics of gliomas, which can be widely applied to classify gliomas and evaluate the extent of metastasis of glioma. The present study aimed to assess the clinical value of magnetic resonance DWI in the pathological grading of glioma and its therapeutic application in clinical surgery. Methods: This article retrospectively analyzed the clinical data of 41 patients with glioma confirmed by surgical pathology results from January 1, 2019 to March 31, 2020 in the People's Hospital of Gaozhou. Among them, 16 patients had low-grade gliomas [World Health Organization (WHO) grade I-II] and 25 patients had high-grade gliomas (WHO grade III-IV). They were subjected to conventional T1WI and T2WI plain scans, along with DWI and enhanced scans before surgery. The ADC values of the glioma parenchyma, the peritumoral edema area, the surrounding white matter, and the contralateral normal white matter were measured. We selected some tumor tissues for pathological analysis as well, and conducted pathological grading according to WHO grading standards. Results: We compared and evaluated the ADC values of the observed areas for low-grade gliomas and high-grade gliomas. The ADC values of low-grade gliomas in the tumor parenchyma, peritumoral edema, and white matter around the edema area were significantly lower than those of high-grade gliomas, and the differences were statistically significant (P<0.05). The difference in ADC values of normal white matter between the two groups of patients was not statistically significant (P=0.125). Conclusions: DWI has prognostic predictive value in the preoperative differential diagnosis and pathological classification of gliomas. This advanced technology can verify the extent of glioma infiltration in the surrounding brain tissue. It can help clinicians formulate a safer and more effective therapeutic strategy by providing accurate information on prognostic evaluation before the successful surgical intervention of gliomas.

Monitoring Clinical-Pathological Grading of Hepatocellular Carcinoma Using MicroRNA-Guided Semiconducting Polymer Dots.

MicroRNAs (miRNAs) are a class of small, noncoding RNAs involved in nearly all genetic central dogma processes and human biological behavior, which also play a significant role in the pathological activity of tumors, such as gene transcription, protein translation, and exosome secretion. Therefore, through the navigation of certain specific miRNAs, we can trace the specific physiological processes or image some specific tissues. Designing and accurately positioning microRNA (miRNA)-sensitive fluorescent nanoprobes with benign specificity and recognition in cells or tissues are a challenging research field. To solve the difficulties, we introduce four semiconducting polymer dots (Pdots) as nanoprobes linked by specific miRNA antisense sequences for monitoring the pathological grading by the variation in miRNA expression. Based on the base pairing principle, these miRNA-sensitive Pdots could bind to specific miRNAs within the cancerous cells. As impacted by the background of different pathology gradings, the proportions of the four hepatocellular carcinoma (HCC)-specific miRNAs within the cancerous cell are different, and the pathological grading of the patient tissues can be determined by comparing the palette combinations. The short single-stranded RNA-functionalized Pdots, which have excellent microRNA sensitivity, are observed in an experimental cell model and a series of tissue specimens from HCC patients for the first time. Using the Förster (or fluorescence) resonance energy transfer (FRET) model of Pdots and Cy3dt tag to simulate in vivo miRNA detection, the superior sensitivity and specificity of these nanoprobes are verified. The interference of subjective factors in traditional single/bis-dye emission intensity detection is abandoned, and multiple label staining is used to enhance sensitivity further and reduce the false-positive rate. The feasibility exhibited by this novel staining method is verified in normal hepatocellular HCC cell lines and 16 frozen ultrathin tissue sections, which are employed to quantify pathological grading-related color presentation systems for clinical doctors and pathologists' use. The intelligently designed miRNA-guided Pdots will emerge as an ideal platform with promising biological imaging.

The value of MR amide proton transfer weighted imaging technique in predicting the pathological grade of brainstem glioma / 中华放射学杂志

Objective:To evaluate the clinical application value of MR amide proton transfer weighted imaging (APTWI) in predicting the pathological grade of brainstem glioma (BSG).Methods:The data of 41 BSG patients in Beijing Tiantan Hospital, Capital Medical University from August 2019 to June 2020 who underwent both MRI and APTWI 2 weeks before surgery and had pathological grading results were retrospectively analyzed. According to the pathological results, 41 patients were classified into high-grade BSG (20 patients) and low-grade BSG (21 patients). Combined with conventional MR images, the signal intensity (%) of amide proton transfer (APT) in the parenchymal area of the tumor was obtained on APTWI images. χ 2 test or independent sample t-test was used to analyze the differences in gender distribution, age and APT signal intensity between patients with high and low grade BSG. Receiver operating characteristic (ROC) curve was drawn to predict the efficacy of APT signal intensity in the differential diagnosis of high and low grade BSG, and Youden index was calculated to obtain the optimal diagnostic threshold; the predictive ability of APT signal intensity was analyzed in combination with Hosmer-Lemeshow goodness of fit test. Results:There was no significant difference in age [(23±18) years, (20±17) years, t=0.97, P=0.340] and gender distribution (9/11, 9/12 for males/females, χ 2=0.02, P=0.890) between high-grade and low-grade BSG patients. The APT signal intensity of high-grade BSG [(3.9±0.9)%] was significantly higher than that of low-grade BSG [(2.8±0.9)%], and the difference had statistical significance ( t=4.16, P<0.001). The area under the ROC curve of APT signal intensity to distinguish high-grade and low grade BSG was 0.836, and with 2.85% as the optimal diagnostic threshold of APT signal intensity, its sensitivity for the diagnosis of high-grade BSG was 90.0% and specificity was 71.4%. The Hosmer-Lemeshow test showed that APTWI had a good predictive ability for BSG grade (χ 2=13.33, P=0.101). Conclusion:APTWI can be applied in distinguishing high grade BSG from low grade BSG, and has clinical value in predicting glioma grading.

Preoperative prediction of pathological grading of hepatocellular carcinoma using machine learning-based ultrasomics: A multicenter study.

PURPOSE: The present study investigated the value of ultrasomics signatures in the preoperative prediction of the pathological grading of hepatocellular carcinoma (HCC) via machine learning. METHODS: A total of 193 patients were collected from three hospitals. The patients from two hospitals (n = 160) were randomly divided into training set (n = 128) and test set (n = 32) at a 8:2 ratio. The patients from a third hospital were used as an independent validation set (n = 33). The ultrasomics features were extracted from the tumor lesions on the ultrasound images. Support vector machine (SVM) was used to construct three preoperative pathological grading models for HCC on each dataset. The performance of the three models was evaluated by area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and accuracy. RESULTS: The ultrasomics signatures extracted from the grayscale ultrasound images could successfully differentiate between high- and low-grade HCC lesions on the training set, test set, and the independent validation set (p < 0.05). On the test set and the validation set, the combined model's performance was the highest, followed by the ultrasomics model and the clinical model successively (p < 0.05). Their AUC (along with 95 %CI) of these models was 0.874(0.709-0.964), 0.789(0.608-0.912), 0.720(0.534-0.863) and 0.849(0.682-0.949), 0.825(0.654-0.935), 0.770(0.591-0.898), respectively. CONCLUSION: Machine learning-based ultrasomics signatures could be used for noninvasive preoperative prediction of pathological grading of HCC. The combined model displayed a better predictive performance for pathological grading of HCC and had a stronger generalization ability.

Pathological evaluation of tumor grade for salivary adenoid cystic carcinoma: A proposal of an objective grading system.

Three pathological grading systems advocated by Perzin/Szanto, Spiro, and van Weert are currently used for adenoid cystic carcinoma (AdCC). In these systems, the amount or presence of the solid tumor component in AdCC specimens is an important index. However, the "solid tumor component" has not been well defined. Salivary AdCC cases (N = 195) were collected after a central pathology review. We introduced a novel criterion for solid tumor component, minAmax (minor axis maximum). The largest solid tumor nest in each AdCC case was histologically screened, the maximum oval fitting the solid nest was estimated, and the length of the minor axis of the oval (minAmax) was measured. The prognostic cutoff for the minAmax was determined using training and validation cohorts. All cases were evaluated for the four grading systems, and their prognostic impact and interobserver variability were examined. The cutoff value for the minAmax was set at 0.20 mm. Multivariate prognostic analyses showed the minAmax and van Weert systems to be independent prognostic tools for overall, disease-free, and distant metastasis-free survival while the Perzin/Szanto and Spiro systems were selected for overall survival but not for disease-free or distant metastasis-free survival. The highest hazard ratio for overall survival (11.9) was obtained with the minAmax system. The reproducibility of the minAmax system (kappa coefficient of 0.81) was scored as very good while those of the other three systems were scored as moderate. In conclusion, the minAmax is a simple, objective, and highly reproducible grading system useful for prognostic stratification for salivary AdCC.

Rapid visualizing and pathological grading of bladder tumor tissues by simple nanodiagnostics.

Developing a tissue diagnosis technology to avoid the complicated processes and the usage of expensive reagents while achieving rapid pathological grading diagnosis to provide a better strategy for clinical treatment is an important strategy of tumor diagnose. Herein, we selected the integrin αvß3 as target based on the analysis of clinical data, and then designed a stable and cancer-targeted selenium nanosystem (RGD@SeNPs) by using RGD polypeptide as the targeting modifier. In vitro experiments showed that RGD@SeNPs could specifically recognized tumor cells, especially in co-culture cells model. The RGD@SeNPs can be used for clinical samples staining without the use of primary and secondary antibody. Fluorescence difference of the tissue specimens staining with RGD@SeNPs could be used to distinguish normal tissues and tumor tissues or estimate different pathological grades of cancer at tissue level. 132 clinical tumor specimens with three types of tumor and 76 non-tumor specimens were examined which verified that the nanoparticles could specific and sensitive distinguish tumor tissue from normal tissue with a specificity of 92% and sensitivity of 96%. These results demonstrate the potential of cancer-targeted RGD@SeNPs as translational nanodiagnostics for rapid visualizing and pathological grading of bladder tumor tissues in clinical specimens.

Predictive Value of the Texture Analysis of Enhanced Computed Tomographic Images for Preoperative Pancreatic Carcinoma Differentiation.

PURPOSE: To assess the utility of texture analysis for predicting the pathological degree of differentiation of pancreatic carcinoma (PC). METHODS: Eighty-three patients with PC who went through postoperative pathology diagnose and CT examination were selected at Anhui Provincial Hospital. Among them, 34 cases were moderately differentiated, 13 cases were poorly differentiated, and 36 cases were moderately poorly differentiated. The images in the arterial and venous phase (VP) with the lesions at their largest cross section were selected to manually outline the region of interest (ROI) to delineate lesions using open-source software. A total of 396 features were extracted from the ROI using AK software. Spearman correlation analysis and random forest selection by filter (rfSBF) in the caret package of R studio were used to select the discriminating features. The receiver operating characteristic ROC analysis was used to evaluate their discriminative performance. RESULTS: Twelve and six features were selected in the arterial and VPs, respectively. The areas under the ROC curve (AUC) in the arterial phase (AP) for diagnosing poorly differentiated, moderately differentiated and moderate-poorly differentiated cases were 0.80, 1, and 0.80 in the training group and 0.77, 1, and 0.77 in the test group; in the VP, the values were 0.81, 1, and 0.82 in the training group and 0.74, 1, and 0.74 in the test group. CONCLUSION: Texture analysis based on contrast-enhanced CT images can be used as an adjunct for the preoperative assessment of the pathological degrees of differentiation of PC.

Pathological prognostic factors of pseudomyxoma peritonei: comprehensive clinicopathological analysis of 155 cases.

BACKGROUND: Pseudomyxoma peritonei (PMP) is an extremely rare malignancy, characterized by extensive peritoneal implantation and colloidal ascites. This study was to explore the pathological prognostic factors of PMP. METHODS: Specimens from 155 PMP patients were analyzed by H&E and immunohistochemistry. Parameters included primary tumor location, histological grade, lymph node metastasis, tumor emboli in the blood and lymph vessels, perineural invasion, Ki67 labeling index, p53, mismatch repair (MMR) gene mutations, MUC1, MUC2, MUC5AC, and MUC6. Clinicopathological and follow-up data were subjected to univariate and multivariate analyses. RESULTS: The patients included 63.2% (n = 98) low-grade mucinous carcinoma peritonei, 31.6% (n = 49) high-grade mucinous carcinoma peritonei and 5.2% (n = 8) high-grade mucinous carcinoma peritonei with signet ring cells. There were 9.7% (n = 15) with lymph node metastasis; 11.6% (n = 18) with angiolymphatic invasion; 6.3% (n = 8) with defective MMR (dMMR); 35.5% (n = 55) with Ki67 labeling index ≥ 50%; 36.1% (n = 56) with p53 mutation. For PMP from appendiceal origin (n = 140), univariate analysis identified 10 potential prognostic factors. But Multivariate analysis identified only histologic grade was the independent prognostic factor for OS. Mortality risk of high-grade peritoneal mucinous carcinoma or high-grade peritoneal mucinous carcinoma with signet ring cells was 7.056 times (P < .0001, 95% CI: 2.701-18.435) or 27.224 times (P < .0001, 95% CI: 6.207-119.408), respectively, higher than low-grade. CONCLUSIONS: For PMP from the appendiceal origin, histological grade could be the only independent prognostic factor.

Application of Radiomics Analysis Based on CT Combined With Machine Learning in Diagnostic of Pancreatic Neuroendocrine Tumors Patient's Pathological Grades.

PURPOSE: To evaluate the value of multiple machine learning methods in classifying pathological grades (G1,G2, and G3), and to provide the best machine learning method for the identification of pathological grades of pancreatic neuroendocrine tumors (PNETs) based on radiomics. MATERIALS AND METHODS: A retrospective study was conducted on 82 patients with Pancreatic Neuroendocrine tumors. All patients had definite pathological diagnosis and grading results. Using Lifex software to extract the radiomics features from CT images manually. The sensitivity, specificity, area under the curve (AUC) and accuracy were used to evaluate the performance of the classification model. RESULT: Our analysis shows that the CT based radiomics features combined with multi algorithm machine learning method has a strong ability to identify the pathological grades of pancreatic neuroendocrine tumors. DC + AdaBoost, DC + GBDT, and Xgboost+RF were very valuable for the differential diagnosis of three pathological grades of PNET. They showed a strong ability to identify the pathological grade of pancreatic neuroendocrine tumors. The validation set AUC of DC + AdaBoost is 0.82 (G1 vs G2), 0.70 (G2 vs G3), and 0.85 (G1 vs G3), respectively. CONCLUSION: In conclusion, based on enhanced CT radiomics features could differentiate between different pathological grades of pancreatic neuroendocrine tumors. Feature selection method Distance Correlation + classifier method Adaptive Boosting show a good application prospect.

[Quantitative analysis of hepatocellular carcinomas pathological grading in non-contrast magnetic resonance images].

In order to solve the pathological grading of hepatocellular carcinomas (HCC) which depends on biopsy or surgical pathology invasively, a quantitative analysis method based on radiomics signature was proposed for pathological grading of HCC in non-contrast magnetic resonance imaging (MRI) images. The MRI images were integrated to predict clinical outcomes using 328 radiomics features, quantifying tumour image intensity, shape and text, which are extracted from lesion by manual segmentation. Least absolute shrinkage and selection operator (LASSO) were used to select the most-predictive radiomics features for the pathological grading. A radiomics signature, a clinical model, and a combined model were built. The association between the radiomics signature and HCC grading was explored. This quantitative analysis method was validated in 170 consecutive patients (training dataset: n = 125; validation dataset, n = 45), and cross-validation with receiver operating characteristic (ROC) analysis was performed and the area under the ROC curve (AUC) was employed as the prediction metric. Through the proposed method, AUC was 0.909 in training dataset and 0.800 in validation dataset, respectively. Overall, the prediction performances by radiomics features showed statistically significant correlations with pathological grading. The results showed that radiomics signature was developed to be a significant predictor for HCC pathological grading, which may serve as a noninvasive complementary tool for clinical doctors in determining the prognosis and therapeutic strategy for HCC.

Multi-level features combined end-to-end learning for automated pathological grading of breast cancer on digital mammograms.

We propose to discriminate the pathological grades directly on digital mammograms instead of pathological images. An end-to-end learning algorithm based on the combined multi-level features is proposed. Low-level features are extracted and selected by supervised LASSO logistic regression. Convolutional Neural Network (CNN) is designed to extract high-level semantic features. These extracted multi-level features are combined to optimize the new CNN end to end to make different parts of the network learn to pay attention to different level of features. Results demonstrate that our proposed algorithm is superior to other CNN models and obtain comparable performance compared with pathological images.

Establishment of WHO/ISUP grading prediction model for renal clear cell carcinoma based on CT radiomics / 中华泌尿外科杂志

Objective@#A predictive model of WHO/ISUP grading of renal clear cell carcinoma was constructed based on CT radiomics.@*Methods@#The clinical data of 104 patients with ccRCC confirmed by operation or biopsy from March 2014 to December 2018 in the Affiliated Hospital of Shaanxi University of Traditional Chinese Medicine were retrospectively analyzed. There were 70 males and 34 females, and the age was 61.2±11.7 years. The patients were randomly divided into development cohort (73 cases) and validation cohort (31 cases) by stratified sampling according to 7∶3 ratio. According to the WHO/ISUP pathological grading criteria of renal cancer in 2016, Ⅰ and Ⅱ were defined as low-grade group, Ⅲ and Ⅳ were defined as high-grade group. The radiomics features of ccRCC were calculated in cortical phase images of CT enhanced scanning. LASSO regression was used to reduce the radiomics feature dimensionality in the training group, and to establish radiomics risk scores. The binary logistic regression was used to build the prediction model, which was used in the validation group. Bootstrap method was used to validate the model of training and validation group. AUC, sensitivity and specificity were calculated respectively. Hosmer-Lemeshow goodness-of-fit test was used to evaluate model calibration degree.@*Results@#After dimensionality reduction, the radiomics risk score of ccRCC was established. The low and high-level risk scores of the training group were -2.49±1.73 and 1.23±2.17, with significant difference (t=-7.785, P < 0.01). The binary logistic regression multivariate analysis showed that the radiomics risk score was an independent risk factor in identifying low or high-grade ccRCC with odds ratio of (OR=3.576, 95%CI 1.964~6.513). The predictive model was Y=1/[1+ exp(-Z)], Z=1.274×radiomics risk score+ 0.072. The AUC of radiomics risk score in training group was 0.940 (95%CI 0.883-0.998) with 95.5% sensitivity and 88.2% specificity after internal verification by Bootstrap method, and good Hosmer-Lemeshow goodness-of-fit test (χ2=4.463, P>0.05). The low and high-level risk scores of the Validation group were -2.27±2.02 and 0.82±2.08, with significant difference (t=-3.832, P<0.01). The AUC in validation group was 0.859(95%CI 0.723-0.995) with 77.8% sensitivity and 81.8% specificity, and with good Hosmer-Lemeshow goodness-of-fit test (χ2=14.554, P=0.068) as well.@*Conclusions@#The prediction model based on CT radiomics has high accuracy in predicting high or low grade of ccRCC.

Quantitative analysis of hepatocellular carcinomas pathological grading in non-contrast magnetic resonance images / 生物医学工程学杂志

In order to solve the pathological grading of hepatocellular carcinomas (HCC) which depends on biopsy or surgical pathology invasively, a quantitative analysis method based on radiomics signature was proposed for pathological grading of HCC in non-contrast magnetic resonance imaging (MRI) images. The MRI images were integrated to predict clinical outcomes using 328 radiomics features, quantifying tumour image intensity, shape and text, which are extracted from lesion by manual segmentation. Least absolute shrinkage and selection operator (LASSO) were used to select the most-predictive radiomics features for the pathological grading. A radiomics signature, a clinical model, and a combined model were built. The association between the radiomics signature and HCC grading was explored. This quantitative analysis method was validated in 170 consecutive patients (training dataset: = 125; validation dataset, = 45), and cross-validation with receiver operating characteristic (ROC) analysis was performed and the area under the ROC curve (AUC) was employed as the prediction metric. Through the proposed method, AUC was 0.909 in training dataset and 0.800 in validation dataset, respectively. Overall, the prediction performances by radiomics features showed statistically significant correlations with pathological grading. The results showed that radiomics signature was developed to be a significant predictor for HCC pathological grading, which may serve as a noninvasive complementary tool for clinical doctors in determining the prognosis and therapeutic strategy for HCC.

Establishment of WHO/ISUP grading prediction model for renal clear cell carcinoma based on CT radiomics / 中华泌尿外科杂志

Objective A predictive model of WHO/ISUP grading of renal clear cell carcinoma was constructed based on CT radiomics.Methods The clinical data of 104 patients with ccRCC confirmed by operation or biopsy from March 2014 to December 2018 in the Mfiliated Hospital of Shaanxi University of Traditional Chinese Medicine were retrospectively analyzed.There were 70 males and 34 females,and the age was 61.2 ± 11.7 years.The patients were randomly divided into development cohort (73 cases) and validation cohort (31 cases) by stratified sampling according to 7∶3 ratio.According to the WHO/ISUP pathological grading criteria of renal cancer in 2016,Ⅰ and Ⅱ were defined as low-grade group,Ⅲ and Ⅳ were defined as high-grade group.The radiomics features of ccRCC were calculated in cortical phase images of CT enhanced scanning.LASSO regression was used to reduce the radiomics feature dimensionality in the training group,and to establish radiomics risk scores.The binary logistic regression was used to build the prediction model,which was used in the validation group.Bootstrap method was used to validate the model of training and validation group.AUC,sensitivity and specificity were calculated respectively.Hosmer-Lemeshow goodness-of-fit test was used to evaluate model calibration degree.Results After dimensionality reduction,the radiomics risk score of ccRCC was established.The low and high-level risk scores of the training group were-2.49 ± 1.73 and 1.23 ± 2.17,with significant difference (t =-7.785,P ＜ 0.01).The binary logistic regression multivariate analysis showed that the radiomics risk score was an independent risk factor in identifying low or high-grade ccRCC with odds ratio of (OR =3.576,95％ CI 1.964 ～ 6.513).The predictive model was Y =1/[1 + exp(-Z)],Z =1.274 × radiomics risk score + 0.072.The AUC of radiomics risk score in training group was 0.940 (95％ CI 0.883-0.998) with 95.5％ sensitivity and 88.2％ specificity after internal verification by Bootstrap method,and good Hosmer-Lemeshow goodness-of-fit test (x2 =4.463,P ＞ 0.05).The low and high-level risk scores of the Validation group were-2.27 ± 2.02 and 0.82 ± 2.08,with significant difference (t =-3.832,P ＜ 0.01).The AUC in validation group was 0.859(95％ CI 0.723-0.995) with 77.8％ sensitivity and 81.8％ specificity,and with good Hosmer-Lemeshow goodness-of-fit test (x2 =14.554,P =0.068) as well.Conclusions The prediction model based on CT radiomics has high accuracy in predicting high or low grade of ccRCC.

Prediction of ESRD in IgA Nephropathy Patients from an Asian Cohort: A Random Forest Model.

BACKGROUND/AIMS: There is an increasing risk of end-stage renal disease (ESRD) among Asian people with immunoglobulin A nephropathy (IgAN). A computer-aided system for ESRD prediction in Asian IgAN patients has not been well studied. METHODS: We retrospectively reviewed biopsy-proven IgAN patients treated at the Department of Nephrology of the Second Xiangya Hospital from January 2009 to November 2013. Demographic and clinicopathological data were obtained within 1 month of renal biopsy. A random forest (RF) model was employed to predict the ESRD status in IgAN patients. All cases were initially trained and validated, taking advantage of the out-of-bagging(OOB) error. Predictors used in the model were selected according to the Gini impurity index in the RF model and verified by logistic regression analysis. The area under the receiver operating characteristic(ROC) curve (AUC) and F-measure were used to evaluate the RF model. RESULTS: A total of 262 IgAN patients were enrolled in this study with a median follow-up time of 4.66 years. The importance rankings of predictors of ESRD in the RF model were first obtained, indicating some of the most important predictors. Logistic regression also showed that these factors were statistically associated with ESRD status. We first trained an initial RF model using gender, age, hypertension, serum creatinine, 24-hour proteinuria and histological grading suggested by the Clinical Decision Support System for IgAN (CDSS, www.IgAN.net). This 6-predictor model achieved a F-measure of 0.8 and an AUC of 92.57%. By adding Oxford-MEST scores, this model outperformed the initial model with an improved AUC (96.1%) and F-measure (0.823). When C3 staining was incorporated, the AUC was 97.29% and F-measure increased to 0.83. Adding the estimated glomerular filtration rate (eGFR) improved the AUC to 95.45%. We also observed improved performance of the model with additional inputs of blood urea nitrogen (BUN), uric acid, hemoglobin and albumin. CONCLUSION: In addition to the predictors in the CDSS, Oxford-MEST scores, C3 staining and eGFR conveyed additional information for ESRD prediction in Chinese IgAN patients using a RF model.