MRI texture-based machine learning models for the evaluation of renal function on different segmentations: a proof-of-concept study

Multi-parametric MRI-based machine learning model for prediction of pathological grade of renal injury in a rat kidney cold ischemia-reperfusion injury model

BACKGROUND

Renal cold ischemia-reperfusion injury (CIRI), a pathological process during kidney transplantation, may result in delayed graft function and negatively impact graft survival and function. There is a lack of an accurate and non-invasive tool for evaluating the degree of CIRI. Multi-parametric MRI has been widely used to detect and evaluate kidney injury. The machine learning algorithms introduced the opportunity to combine biomarkers from different MRI metrics into a single classifier.

OBJECTIVE

To evaluate the performance of multi-parametric magnetic resonance imaging for grading renal injury in a rat model of renal cold ischemia-reperfusion injury using a machine learning approach.

METHODS

Eighty male SD rats were selected to establish a renal cold ischemia -reperfusion model, and all performed multiparametric MRI scans (DWI, IVIM, DKI, BOLD, T1mapping and ASL), followed by pathological analysis. A total of 25 parameters of renal cortex and medulla were analyzed as features. The pathology scores were divided into 3 groups using K-means clustering method. Lasso regression was applied for the initial selecting of features. The optimal features and the best techniques for pathological grading were obtained. Multiple classifiers were used to construct models to evaluate the predictive value for pathology grading.

RESULTS

All rats were categorized into mild, moderate, and severe injury group according the pathologic scores. The 8 features that correlated better with the pathologic classification were medullary and cortical Dp, cortical T2*, cortical Fp, medullary T2*, ∆T1, cortical RBF, medullary T1. The accuracy(0.83, 0.850, 0.81, respectively) and AUC (0.95, 0.93, 0.90, respectively) for pathologic classification of the logistic regression, SVM, and RF are significantly higher than other classifiers. For the logistic model and combining logistic, RF and SVM model of different techniques for pathology grading, the stable and perform are both well. Based on logistic regression, IVIM has the highest AUC (0.93) for pathological grading, followed by BOLD(0.90).

CONCLUSION

The multi-parametric MRI-based machine learning model could be valuable for noninvasive assessment of the degree of renal injury.

Non-contrast-enhanced magnetic resonance urography for measuring split kidney function in pediatric patients with hydronephrosis: comparison with renal scintigraphy

BACKGROUND

Split kidney function (SKF) is critical for treatment decision in pediatric patients with hydronephrosis and is commonly measured using renal scintigraphy (RS). Non-contrast-enhanced magnetic resonance urography (NCE-MRU) is increasingly used in clinical practice. This study aimed to investigate the feasibility of using NCE-MRU as an alternative to estimate SKF in pediatric patients with hydronephrosis, compared to RS.

METHODS

Seventy-five pediatric patients with hydronephrosis were included in this retrospective study. All patients underwent NCE-MRU and RS within 2 weeks. Kidney parenchyma volume (KPV) and texture analysis parameters were obtained from T2-weighted (T2WI) in NCE-MRU. The calculated split KPV (SKPV) percent and texture analysis parameters percent of left kidney were compared with the RS-determined SKF.

RESULTS

SKPV showed a significant positive correlation with SKF (r = 0.88, p < 0.001), while inhomogeneity was negatively correlated with SKF (r =  - 0.68, p < 0.001). The uncorrected and corrected prediction models of SKF were established using simple and multiple linear regression. Bland-Altman plots demonstrated good agreement of both predictive models. The residual sum of squares of the corrected prediction model was lower than that of the uncorrected model (0.283 vs. 0.314) but not statistically significant (p = 0.662). Subgroup analysis based on different MR machines showed correlation coefficients of 0.85, 0.95, and 0.94 between SKF and SKPV for three different scanners, respectively (p < 0.05 for all).

CONCLUSIONS

NCE-MRU can be used as an alternative method for estimating SKF in pediatric patients with hydronephrosis when comparing with RS. Specifically, SKPV proves to be a simple and universally applicable indicator for predicting SKF.

Radiomic texture analysis based on neurite orientation dispersion and density imaging to differentiate glioblastoma from solitary brain metastasis

BACKGROUND

We created discriminative models of different regions of interest (ROIs) using radiomic texture features of neurite orientation dispersion and density imaging (NODDI) and evaluated the feasibility of each model in differentiating glioblastoma multiforme (GBM) from solitary brain metastasis (SBM).

METHODS

We conducted a retrospective study of 204 patients with GBM (n = 146) or SBM (n = 58). Radiomic texture features were extracted from five ROIs based on three metric maps (intracellular volume fraction, orientation dispersion index, and isotropic volume fraction of NODDI), including necrosis, solid tumors, peritumoral edema, tumor bulk volume (TBV), and abnormal bulk volume. Four feature selection methods and eight classifiers were used for the radiomic texture feature selection and model construction. Receiver operating characteristic (ROC) curve analysis was used to evaluate the diagnostic performance of the models. Routine magnetic resonance imaging (MRI) radiomic texture feature models generated in the same manner were used for the horizontal comparison.

RESULTS

NODDI-radiomic texture analysis based on TBV subregions exhibited the highest accuracy (although nonsignificant) in differentiating GBM from SBM, with area under the ROC curve (AUC) values of 0.918 and 0.882 in the training and test datasets, respectively, compared to necrosis (AUCtraining:0.845, AUCtest:0.714), solid tumor (AUCtraining:0.852, AUCtest:0.821), peritumoral edema (AUCtraining:0.817, AUCtest:0.762), and ABV (AUCtraining:0.834, AUCtest:0.779). The performance of the five ROI radiomic texture models in routine MRI was inferior to that of the NODDI-radiomic texture model.

CONCLUSION

Preoperative NODDI-radiomic texture analysis based on TBV subregions shows great potential for distinguishing GBM from SBM.

Kidney Cancer Diagnosis and Surgery Selection by Machine Learning from CT Scans Combined with Clinical Metadata

Kidney cancers are one of the most common malignancies worldwide. Accurate diagnosis is a critical step in the management of kidney cancer patients and is influenced by multiple factors including tumor size or volume, cancer types and stages, etc. For malignant tumors, partial or radical surgery of the kidney might be required, but for clinicians, the basis for making this decision is often unclear. Partial nephrectomy could result in patient death due to cancer if kidney removal was necessary, whereas radical nephrectomy in less severe cases could resign patients to lifelong dialysis or need for future transplantation without sufficient cause. Using machine learning to consider clinical data alongside computed tomography images could potentially help resolve some of these surgical ambiguities, by enabling a more robust classification of kidney cancers and selection of optimal surgical approaches. In this study, we used the publicly available KiTS dataset of contrast-enhanced CT images and corresponding patient metadata to differentiate four major classes of kidney cancer: clear cell (ccRCC), chromophobe (chRCC), papillary (pRCC) renal cell carcinoma, and oncocytoma (ONC). We rationalized these data to overcome the high field of view (FoV), extract tumor regions of interest (ROIs), classify patients using deep machine-learning models, and extract/post-process CT image features for combination with clinical data. Regardless of marked data imbalance, our combined approach achieved a high level of performance (85.66% accuracy, 84.18% precision, 85.66% recall, and 84.92% F1-score). When selecting surgical procedures for malignant tumors (RCC), our method proved even more reliable (90.63% accuracy, 90.83% precision, 90.61% recall, and 90.50% F1-score). Using feature ranking, we confirmed that tumor volume and cancer stage are the most relevant clinical features for predicting surgical procedures. Once fully mature, the approach we propose could be used to assist surgeons in performing nephrectomies by guiding the choices of optimal procedures in individual patients with kidney cancer.