Radiomic Features Research Articles

Magnetic Resonance Imaging-Based Radiomics of Axial and Sagittal Orientation in Pregnant Patients with Suspected Placenta Accreta Spectrum

Rationale and ObjectivesPlacenta accreta spectrum (PAS) is associated with significant morbidity and mortality. Current radiomic analysis of PAS magnetic resonance (MR) images is often performed on a single imaging plane. However, depending on the chosen imaging plane, radiomic features extracted from the same patient may vary due to the differing orientations and anatomical contexts, potentially leading to inconsistent results. In this study, we applied region of interest (ROI)-based radiomic analysis on axial and sagittal MR images in pregnant patients at high-risk for PAS. Our objective was to compare MR textural features extracted from these imaging planes and to correlate these findings with surgical outcomes, aiming to enhance the accuracy of PAS diagnosis and treatment planning. Materials and MethodsThis is a retrospective review of MR images of pregnancies with prenatally suspected PAS. Volumetric placental, uterus, and internal os of the cervix regions of interest (ROI) were manually segmented on axial and sagittal MR images for each patient. Radiomic features were extracted following the image biomarker standardization initiative guideline. Agreement in features extracted from axial and sagittal images were assessed using Spearman’s rank correlation coefficient. ResultsOf the 101 pregnant patients that met the study inclusion criteria, 65 underwent cesarean hysterectomy for PAS. Seventy seven percent of the radiomics features had strong Spearman rank correlations (>0.8) between axial and sagittal images, indicating that these imaging planes provide similar radiomics information. The diagnostic performance of features extracted from axial and sagittal planes was quantified under the receiver operating characteristics curve (AUC). We found that axial and sagittal planes have similar performance for the prediction of hysterectomy. Shape elongation, Placental Location within the Uterus (PLU), and heterogeneity features were significant predictors for hysterectomy regardless of the imaging plane. ConclusionOur study found that radiomics features extracted from axial and sagittal MR image plane in the same patient have excellent agreement and strong correlation. We identified several features present in both axial and sagittal images that were predictive in detecting PAS-suspected patient who required hysterectomy. These features may represent the underlying placental pathophysiology.

Predicting the severity of mycoplasma pneumoniae pneumonia in pediatric and adult patients: a multicenter study

The purpose of this study is to develop a nomogram model for early prediction of the severe mycoplasma pneumoniae pneumonia (SMPP) in Pediatric and Adult Patients. A retrospective analysis was conducted on patients with MPP, classifying them into SMPP and non-severe MPP (NSMPP) groups. A total of 550 patients (NSMPP 374 and SMPP 176) were enrolled in the study and allocated to training, validation cohorts. 278 patients (NSMPP 224 and SMPP 54) were retrospectively collected from two institutions and allocated to testing cohort. The risk factors for SMPP were identified using univariate analysis. For radiomic feature selection, Spearman’s correlation and the least absolute shrinkage and selection operator (LASSO) were utilized. Logistic regression was used to build different models, including clinical, imaging, radiomics, and integrated models (combining clinical, imaging, and radiomics features selected). The model’s discrimination was evaluated using a receiver operating characteristic curve, its calibration with a calibration curve, and the results were visualized using the Hosmer–Lemeshow goodness-of-fit test. Thirteen clinical features and fourteen imaging features were selected for constructing the clinical and imaging models. Simultaneously, a set of twenty-five radiomics features were utilized to build the radiomics model. The integrated model demonstrated good calibration and discrimination in the training cohorts (AUC, 0.922; 95% CI: 0.900, 0.942), validation cohorts (AUC, 0.879; 95% CI: 0.806, 0.920), and testing cohorts (AUC, 0.877; 95% CI: 0.836, 0.916). The discriminatory and predictive efficacy of the clinical model in testing cohorts increased further after clinical and radiological features were incorporated (AUC, 0.849 vs. 0.922, P = 0.002). The model demonstrated exemplary predictive efficacy for SMPP by leveraging a comprehensive set of inputs, encompassing clinical data, quantitative and qualitative radiological features, along with radiomics features. The integration of these three aspects in the predictive model further enhanced the performance of the clinical model, indicating the potential for extensive clinical applications.

A CT-based machine learning model for using clinical-radiomics to predict malignant cerebral edema after stroke: a two-center study.

This research aimed to create a machine learning model for clinical-radiomics that utilizes unenhanced computed tomography images to assess the likelihood of malignant cerebral edema (MCE) in individuals suffering from acute ischemic stroke (AIS). The research included 179 consecutive patients with AIS from two different hospitals. These patients were randomly assigned to training (n = 143) and validation (n = 36) sets with an 8:2 ratio. Using 3DSlicer software, the radiomics features of regions impacted by infarction were derived from unenhanced CT scans. The radiomics features linked to MCE were pinpointed through a consistency test, Student's t test and the least absolute shrinkage and selection operator (LASSO) method for selecting features. Clinical parameters associated with MCE were also identified. Subsequently, machine learning models were constructed based on clinical, radiomics, and clinical-radiomics. Ultimately, the efficacy of these models was evaluated by measuring the operating characteristics of the subjects through their area under the curve (AUCs). Logistic regression (LR) was found to be the most effective machine learning algorithm, for forecasting the MCE. In the training and validation cohorts, the AUCs of clinical model were 0.836 and 0.773, respectively, for differentiating MCE patients; the AUCs of radiomics model were 0.849 and 0.818, respectively; the AUCs of clinical and radiomics model were 0.912 and 0.916, respectively. This model can assist in predicting MCE after acute ischemic stroke and can provide guidance for clinical treatment and prognostic assessment.

DSA Quantitative Analysis and Predictive Modeling of Obliteration in Cerebral AVM following Stereotactic Radiosurgery.

Stereotactic radiosurgery is a key treatment modality for cerebral AVMs, particularly for small lesions and those located in eloquent brain regions. Predicting obliteration remains challenging due to evolving treatment paradigms and complex AVM presentations. With digital subtraction angiography (DSA) being the gold standard for outcome evaluation, radiomic approaches offer potential for more objective and detailed analysis. We aimed to develop machine learning modeling using DSA quantitative features for post-SRS obliteration prediction. A prospective registry of patients with cerebral AVMs was screened to include patients with digital prestereotactic radiosurgery DSA. Anterior-posterior and lateral views were retrieved and manually segmented. Quantitative features were computed from the lesion ROI. Following feature selection, machine learning models were developed to predict unsuccessful 2-year total obliteration using processed radiomics features in comparison with clinical and radiosurgical features. When we evaluated through area under the receiver operating characteristic curve (AUROC), accuracy, area under the precision-recall curve F1, recall, and precision, the best performing model predictions on the test set were interpreted using the Shapley additive explanations approach. DSA images of 100 included patients were retrieved and analyzed. The best-performing clinical radiosurgical model was a gradient boosting classifier with an AUROC of 68% and a recall of 67%. When we used radiomics variables as input, the AdaBoost classifier had the best evaluation metrics with an AUROC of 79% and a recall of 75%. The most important clinico-radiosurgical features, ranked by model contribution, were lesion volume, patient age, treatment dose rate, the presence of seizure at presentation, and prior resection. The most important ranked radiomics features were the following: gray-level size zone matrix, gray-level nonuniformity, kurtosis, sphericity, skewness, and gray-level dependence matrix dependence nonuniformity. The combination of radiomics with machine learning is a promising approach for predicting cerebral AVM obliteration status following stereotactic radiosurgery. DSA could enhance prognostication of stereotactic radiosurgery-treated AVMs due to its high spatial resolution. Model interpretation is essential for building transparent models and establishing clinically valid radiomic signatures.

Machine learning and radiomics for ventricular tachyarrhythmia prediction in hypertrophic cardiomyopathy: insights from an MRI-based analysis.

Myocardial fibrosis is often detected in patients with hypertrophic cardiomyopathy (HCM), which causes left ventricular (LV) dysfunction and tachyarrhythmias. To evaluate the potential value of a machine learning (ML) approach that uses radiomic features from late gadolinium enhancement (LGE) and cine images for the prediction of ventricular tachyarrhythmia (VT) in patients with HCM. Hyperenhancing areas of LV myocardium on LGE images were manually segmented, and the segmentation was propagated to corresponding areas on cine images. Radiomic features were extracted using the PyRadiomics library. The least absolute shrinkage and selection operator (LASSO) method was employed for radiomic feature selection. Our model development employed the TabPFN algorithm, an adapted Prior-Data Fitted Network design. Model performance was evaluated graphically and numerically over five-repeat fivefold cross-validation. SHapley Additive exPlanations (SHAP) were employed to determine the relative importance of selected radiomic features. Our cohort consisted of 60 patients with HCM (73.3% male; median age = 51.5 years), among whom 17 had documented VT during the follow-up. A total of 1612 radiomic features were extracted for each patient. The LASSO algorithm led to a final selection of 18 radiomic features. The model achieved a mean area under the receiver operating characteristic curve of 0.877, demonstrating good discrimination, and a mean Brier score of 0.119, demonstrating good calibration. Radiomics-based ML models are promising for predicting VT in patients with HCM during the follow-up period. Developing predictive models as clinically useful decision-making tools may significantly improve risk assessment and prognosis.

Rapid patient-specific organ dose estimation in computed tomography scans via integration of radiomics features and neural networks.

Computed tomography (CT) offers detailed cross-sectional images of internal anatomy for disease detection but carries a risk of solid cancer or blood malignancies due to exposure to X-ray radiation. This study aimed to develop a new method to quickly predict patient-specific organ doses from CT examinations by training neural networks (NNs) based on radiomics features. CT Digital Imaging and Communications in Medicine (DICOM) image data were exported to DeepViewer, a clinical autosegmentation software, to segment the regions of interest (ROIs) for patient organs. Radiomics feature extraction was performed based on the selected CT data and ROIs. Reference organ doses were computed using Monte Carlo (MC) simulations. Patient-specific organ doses were predicted by training a NN model based on radiomics features and reference doses. For the dose prediction performance, the relative root mean squared error (RRMSE), mean absolute percentage error (MAPE), and coefficient of determination (R2) were evaluated on the test sets. The robustness of the NN model was evaluated via the random rearrangement of patient samples in the training and test sets. The maximal difference between the reference and predicted doses was less than 1 mGy for all investigated organs. The range of MAPE was 1.68% to 5.2% for head organs, 11.42% to 15.2% for chest organs, and 5.0% to 8.0% for abdominal organs; the maximal R2 values were 0.93, 0.86, and 0.89 for the head, chest, and abdominal organs, respectively. The radiomics feature-based NN model can achieve accurate prediction of patient-specific organ doses at a high speed of less than 1 second using a single central processing unit, which supports its use as a user-friendly online clinical application.

Prospective external validation of radiomics-based predictive model of distant metastasis after dynamic tumor tracking stereotactic body radiation therapy in patients with non-small-cell lung cancer: A multi-institutional analysis.

This study aims to externally validate a predictive model for distant metastasis (DM) with computed tomography (CT)-based radiomics features in prospectively enrolled non-small-cell lung cancer patients undergoing dynamic tumor-tracking stereotactic body radiation therapy (DTT-SBRT). The study collected retrospective data from 567 patients across 11 institutions as the training dataset and prospectively enrolled 42 patients from four institutions asthe external test dataset. Four clinical features were collected, and 944 CT-based radiomic features were extracted from gross tumor volumes. After standardization and feature selection, DM predictive models were developed using fine and gray regression (FG) and random survival forest (RSF), incorporating clinical and radiomic features, and their combinations within the training dataset. Then, the model was applied to the test dataset, dividing patients into high- and low-risk groups based on medians of risk scores. Model performance was assessed using the concordance index (C-index), and the statistical significance between groups was evaluated using Gray's test. In the training dataset, 122 of 567 patients (21.5%) developed DM, compared to 9 of 42 patients (21.4%) in the test dataset. In the test dataset, the C-indices of the clinical, radiomics, and hybrid models with FG were 0.559, 0.544, and 0.560, respectively, whereas those with RSF were 0.576, 0.604, and 0.627, respectively. The hybrid model with RSF, which exhibited the best predictive performance of all models, identified 7 of 23 patients (30.4%) as high risk and 2 of 19 patients (10.5%) as low risk for DM incidence in the test dataset (p=0.116). Although predictive models for DM lack significance when applied to prospectively enrolled cases undergoing DTT-lung SBRT, the model with RSF exhibits a consistent capacity to effectively classify patients at a high risk of developing DM.

A DTI-based radiomics model for predicting epidermal growth factor receptor (EGFR) amplification in adult IDH1-wild glioblastomas.

Molecular alteration events are common in glioblastomas, the isocitrate dehydrogenase (IDH)-wild of which have had poor survival results so far. The progress of radiomics-based model provides novel sights for its preoperatively noninvasive prediction. To develop a radiomics-based model for predicting epidermal growth factor receptor (EGFR) amplification status in IDH1-wild glioblastomas of adults by pretreatment diffusion tensor imaging (DTI). A total of 124 patients with diagnosed glioblastomas were retrospectively collected. Six conventional magnetic resonance imaging (MRI) features of all the tumors were evaluated visually. Patients were divided into the training (n = 87) and the test set (n = 37) with a ratio of 7:3. Radiomics features were extracted from two regions of the glioblastomas, which were the total tumor (ROI_1) and the solid portion of tumor (ROI_2). The radiomics features extracted from the DTI and T1-contrast-enhanced (T1C) images were selected using the least absolute shrinkage and selection operator (LASSO) regression algorithm. Logistic regression analysis was conducted to develop models for EGFR amplification prediction in the training set. The radiomics model based on ROI_1 demonstrated favorable discrimination in both the training (area under the curve [AUC] = 0.86) and the test set (AUC = 0.82) (P < 0.05). Combining the radiomics features and the conventional feature tumor location, no significant improvement of AUCs was achieved (AUC = 0.86 and 0.81). The radiomics model derived from pretreatment DTI may have potential in differentiating the EGFR mutation status in glioblastomas.

Tumor Growth Pattern and Intra- and Peritumoral Radiomics Combined for Prediction of Initial TACE Outcome in Patients with Primary Hepatocellular Carcinoma.

Non-invasive methods are urgently needed to assess the efficacy of transarterial chemoembolization (TACE) and to identify patients with hepatocellular carcinoma (HCC) who may benefit from this procedure. This study, therefore, aimed to investigate the predictive ability of tumor growth patterns and radiomics features from contrast-enhanced magnetic resonance imaging (CE-MRI) in predicting tumor response to TACE among patients with HCC. A retrospective study was conducted on 133 patients with HCC who underwent TACE at three centers between January 2015 and April 2023. Enrolled patients were divided into training, testing, and validation cohorts. Rim arterial phase hyperenhancement (Rim APHE), tumor growth patterns, nonperipheral washout, markedly low apparent diffusion coefficient (ADC) value, intratumoral arteries, and clinical baseline features were documented for all patients. Radiomics features were extracted from the intratumoral and peritumoral regions across the three phases of CE-MRI. Seven prediction models were developed, and their performances were evaluated using receiver operating characteristic (ROC) and decision curve analysis (DCA). Tumor growth patterns and albumin-bilirubin (ALBI) score were significantly correlated with tumor response. Tumor growth patterns also showed a positive correlation with tumor burden (r = 0.634, P = 0.000). The Peritumor (AUC = 0.85, 0.71, and 0.77), Clinics_Peritumor (AUC = 0.86, 0.77, and 0.81), and Tumor_Peritumor (AUC = 0.87, 0.77, and 0.80) models significantly outperformed the Clinics and Tumor models (P < 0.05), while the Clinics_Tumor_Peritumor model (AUC = 0.88, 0.81, and 0.81) outperformed the Clinics (AUC = 0.67, 0.77, and 0.75), Tumor (AUC = 0.78, 0.72, and 0.68), and Clinics_Tumor (AUC = 0.82, 0.83, and 0.78) models (P < 0.05 or 0.053, respectively). The DCA curve demonstrated better predictive performance within a specific threshold probability range for Clinics_Tumor_Peritumor. Combining tumor growth patterns, intra- and peri-tumoral radiomics features, and ALBI score could be a robust tool for non-invasive and personalized prediction of treatment response to TACE in patients with HCC.

Machine learning models based on CT radiomics features for distinguishing benign and malignant vertebral compression fractures in patients with malignant tumors.

Radiomics has become an important tool for distinguishing benign and malignant vertebral compression fractures (VCFs). It is more clinically significant to concentrate on patients who have malignant tumors and differentiate between benign and malignant VCFs. To explore the value of multiple machine learning (ML) models based on CT radiomics features for differentiating benign and malignant VCFs in patients with malignant tumors. This study retrospectively analyzed 78 patients with malignant tumors accompanied by VCFs, 45 patients with benign VCFs, and 33 patients with malignant VCFs. A total of 140 lesions (86 benign lesions, 54 malignant lesions) were ultimately included in this study. All patients were divided into training sets (n = 98) and validation sets (n = 42) according to the 7:3 ratio. The radiomics features were screened and dimensioned, and multiple radiomics ML models were constructed. The receiver operating characteristic (ROC) curve was performed to assess the diagnostic performance. Five radiomics features were included in the model. All the ML models built have good diagnostic efficiency, among which the support vector machine (SVM) model performs better. The area under the curve (AUC), sensitivity, specificity, and accuracy in the training set were 0.908, 0.816, 0.883, and 0.857, respectively, while those in the validation set were 0.911, 0.647, 0.92, and 0.81, respectively. A variety of ML models built based on CT radiomics features have good value for differentiating benign and malignant VCFs in malignant tumor patients, and the SVM model has a better performance.

A machine learning model based on results of a comprehensive radiological evaluation can predict the prognosis of basal ganglia cerebral hemorrhage treated with neuroendoscopy.

Spontaneous intracerebral hemorrhage is the second most common subtype of stroke. Therefore, this study aimed to investigate the risk factors affecting the prognosis of patients with basal ganglia cerebral hemorrhage after neuroendoscopy. Between January 2020 and January 2024, 130 patients with basal ganglia cerebral hemorrhage who underwent neuroendoscopy were recruited from two independent centers. We split this dataset into training (n = 79), internal validation (n = 22), and external validation (n = 29) sets. The least absolute shrinkage and selection operator-regression algorithm was used to select the top 10 important radiomic features of different regions (perioperative hemorrhage area [PRH], perioperative surround area [PRS], postoperative hemorrhage area [PSH], and postoperative edema area [PSE]). The black hole, island, blend, and swirl signs were evaluated. The top 10 radiomic features and 4 radiological features were combined to construct the k-nearest neighbor classification (KNN), logistic regression (LR), and support vector machine (SVM) models. Finally, the performance of the perioperative hemorrhage and postoperative edema machine learning models was validated using another independent dataset (n = 29). The primary outcome is mRS at 6 months after discharge. The mRS score greater than 3 defined as functional independence. A total of 12 models were built: PRH-KNN, PRH-LR, PRH-SVM, PRS-KNN, PRS-LR, PRS-SVM, PSH-KNN, PSH-LR, PSH-SVM, PSE-KNN, PSE-LR, and PSE-SVM, with corresponding areas under the curve (AUC) values in the internal validation set of 0.95, 0.91, 0.94, 0.52, 0.91, 0.54, 0.67, 0.9, 0.72, 0.92, 0.92, and 0.95, respectively. The AUC values of the PRH-KNN, PRH-LR, PRH-SVM, PSE-KNN, PSE-LR, and PSE-SVM in the external validation were 0.9, 0.92, 0.89, 0.91, 0.92, and 0.88, respectively. The model built based on computed tomography images of different regions accurately predicted the prognosis of patients with basal ganglia cerebral hemorrhage treated with neuroendoscopy. The models built based on the preoperative hematoma area and postoperative edema area showed excellent predictive efficacy in external verification, which has important clinical significance.

Integrating Clinical Variables, Radiomics, and Tumor-derived Cell-free DNA for Enhanced Prediction of Resectable Esophageal Adenocarcinoma Outcomes

The value of integrating clinical variables, radiomics, and tumor-derived cell-free DNA (cfDNA) for the prediction of survival and response to chemoradiation of resectable esophageal adenocarcinoma (rEAC) patients is not yet known. Our aim was to investigate if radiomics and cfDNA metrics combined with clinical variables can improve personalized predictions. A cohort of 111 rEAC patients from two centers treated with neoadjuvant chemoradiotherapy was used for exploratory retrospective analyses. Models combining the clinical variables of the SOURCE survival model with radiomic features and cfDNA, were built using elastic net regression and internally validated using 5-fold cross validation. Model performance for overall survival (OS) and time to progression (TTP) were evaluated with the C-index and the area under the curve (AUC) for pathological complete response (pCR) RESULTS: The best performing baseline models for OS and TTP were based on the combination of SOURCE-cfDNA which reached a C-index of 0.55 and 0.59 compared to 0.44-0.45 with SOURCE alone. The addition of re-staging PET radiomics to SOURCE was the most promising addition for predicting OS (C-index: 0.65) and TTP (C-index: 0.60). Baseline risk-stratification was achieved for OS and TTP by combining SOURCE with radiomics or cfDNA, log-rank p<0.01. The best performing combination model for the prediction of pCR reached an AUC of 0.61 compared to 0.47 with SOURCE variables alone. The addition of radiomics and cfDNA can improve the performance of an established survival model. External validity needs to be further assessed in future studies together with the optimization of radiomic pipelines.

Identification of a rank-based radiomic signature with individualized prognostic value for lung adenocarcinoma in a multi-cohort study

ObjectivesRadiomics provides an opportunity to evaluate cancer prognosis noninvasively. However, the susceptibility of the radiomic quantitative features to multicenter effects, leads to the clinical dilemma of the radiomic signatures. This study aimed to develop a radiomic signature to circumvent multicenter effects, achieving the individualized prognostic assessment of lung adenocarcinoma (LUAD). MethodsUsing computed tomography (CT) imaging of 234 stage I–IIIA LUAD patients derived from three public multicenter cohorts, we proposed a rank-based method that utilized the relative rank patterns of quantitative values between radiomic feature pairs within individual patients and established a feature pair signature for LUAD prognosis. We collected a new clinical cohort with 162 LUAD patients for independent validation. ResultsA rank-based radiomic signature, consisting of 12 feature pairs, was developed, and it could determine the mortality risk for an individual according to the rank patterns of 12 feature pairs within the patient’s CT imaging. The prognostic performance of the rank-based signature was effectively validated in the new clinical cohort (log-rank P = 0.0051, C-index = 0.73), whereas other signatures lost their prognostic ability across centers. The novel proposed radiomic nomogram significantly improved the prognostic performance of clinicopathological factors. The further radiogenomic analyses revealed the underlying biological characteristics (e.g., Stemness, Ferroptosis,’ECM’) reflected by the rank-based radiomic signature. ConclusionsThis multicenter study illustrates the accuracy and stability of the rank-based radiomic signature for LUAD prognosis, and demonstrates a unique advantage of clinical individualized application. The biological characteristics underlying the rank-based radiomic signature would accelerate its clinical application.

A combined model based on radiomics features of Sonazoid contrast-enhanced ultrasound in the Kupffer phase for the diagnosis of well-differentiated hepatocellular carcinoma and atypical focal liver lesions: a prospective, multicenter study.

The objective of this study was to develop a combined model based on radiomics features of Sonazoid contrast-enhanced ultrasound (CEUS) during the Kupffer phase and to evaluate its value in differentiating well-differentiated hepatocellular carcinoma (w-HCC) from atypical benign focal liver lesions (FLLs). A total of 116 patients with preoperatively Sonazoid-CEUS confirmed w-HCC or benign FLL were selected from a prospective multiple study on the clinical application of Sonazoid in FLLs conducted from August 2020 to March 2021. According to the randomization principle, the patients were divided into a training cohort and a test cohort in a 7:3 ratio. Seventy-nine patients were used for establishing and training the radiomics model and combined model. In comparison, 37 patients were used for validating and comparing the performance of the models. The diagnostic efficacy of the models for w-HCC and atypical benign FLLs was evaluated using ROCs curves and decision curves. A combined model nomogram was created to assess its value in reducing unnecessary biopsies. Among the patients, there were 55 cases of w-HCC and 61 cases of atypical benign FLLs, including 28 cases of early liver abscess, 16 cases of atypical hepatic hemangioma, 8 cases of hepatocellular dysplastic nodules (DN), and 9 cases of focal nodular hyperplasia (FNH). The radiomics model and combined model we established had AUCs of 0.905 and 0.951, respectively, in the training cohort, and the AUCs of the two models in the test cohort were 0.826 and 0.912, respectively. The combined model outperformed the radiomics feature model significantly. Decision curve analysis demonstrated that the combined model achieved a higher net benefit within a specific threshold probability range (0.25 to 1.00). A nomogram of the combined model was developed. The combined model based on the radiomics features of Sonazoid-CEUS in the Kupffer phase showed satisfactory performance in diagnosing w-HCC and atypical benign FLLs. It can assist clinicians in timely detecting malignant FLLs and reducing unnecessary biopsies for benign diseases.

Artificial intelligence model for enhancing the accuracy of transvaginal ultrasound in detecting endometrial cancer and endometrial atypical hyperplasia

ObjectivesTransvaginal ultrasound is typically the initial diagnostic approach in patients with postmenopausal bleeding for detecting endometrial atypical hyperplasia/cancer. Although transvaginal ultrasound demonstrates notable sensitivity, its specificity remains limited. The objective...

Prostate-Specific Membrane Antigen-Positron Emission Tomography-Guided Radiomics and Machine Learning in Prostate Carcinoma.

Positron emission tomography (PET) using radiolabeled prostate-specific membrane antigen targeting PET-imaging agents has been increasingly used over the past decade for imaging and directing prostate carcinoma treatment. Here, we summarize the available literature data on radiomics and machine learning using these imaging agents in prostate carcinoma. Gleason scores derived from biopsy and after resection are discordant in a large number of prostate carcinoma patients. Available studies suggest that radiomics and machine learning applied to PSMA-radioligand avid primary prostate carcinoma might be better performing than biopsy-based Gleason-scoring and could serve as an alternative for non-invasive GS characterization. Furthermore, it may allow for the prediction of biochemical recurrence with a net benefit for clinical utilization. Machine learning based on PET/CT radiomics features was also shown to be able to differentiate benign from malignant increased tracer uptake on PSMA-targeting radioligand PET/CT examinations, thus paving the way for a fully automated image reading in nuclear medicine. As for prediction to treatment outcome following 177Lu-PSMA therapy and overall survival, a limited number of studies have reported promising results on radiomics and machine learning applied to PSMA-targeting radioligand PET/CT images for this purpose. Its added value to clinical parameters warrants further exploration in larger datasets of patients.

Prediction of feeding difficulties in neonates with hypoxic-ischemic encephalopathy using magnetic resonance imaging-derived radiomics features.

The mechanisms behind brain and spinal cord injuries in hypoxic-ischemic encephalopathy (HIE) and associated feeding difficulties are unclear, with previous magnetic resonance imaging (MRI) attempts yielding inconclusive results. We aim to evaluate an MRI radiomics model for predicting feeding difficulties in HIE infants. Additionally, we investigate changes in predictive capability after incorporating the duration of mechanical ventilation and the timing of MRI examination. Retrospective study with 151 HIE infants (January 2013 to December 2021), randomly divided into training and validation sets. Radiomics features extracted from basal ganglia-thalamus and brainstem in T1-weighted and T2-weighted MRI. Established single-modality, single-site, and multimodality/multisite models. Receiver operating characteristic analysis and area under the curve evaluated models. Decision curve analysis assessed changes in predictive capability. The combined radiomics model of the basal ganglia-thalamus and brainstem regions on the T2-weighted imaging demonstrated superior performance (area under the curve: 0.958 and 0.875 for training and validation, respectively). Combining scores with duration of mechanical ventilation and MRI examination time in a calibration plot model improved and stabilized performance, showing high fitting and clinical utility. Decision curve analysis favored the combined calibration plot model. The MRI-based radiomics model predicts feeding difficulties in HIE infants, with basal ganglia-thalamus and brainstem as relevant factors. The combined calibration plot model exhibits the highest clinical predictive efficacy.

Multi-instance learning for identifying high-risk subregions associated with synchronous distant metastasis in clear cell renal cell carcinoma.

Clear cell renal cell carcinoma (ccRCC) is one of the most common histological subtypes of renal tumors. To identify high-risk subregions associated with synchronous distant metastasis. This study enrolled a total of 277 patients with ccRCC. Voxel intensity and local entropy values were compiled within the region of interest for all patients. Unsupervised k-means clustering yielded three subregions per tumor. Radiomic features were extracted, and random forest-based feature selection was conducted. The selected features were used in a multi-instance support vector machine (mi-SVM) model for training, and predictions were made on the validation cohort. Model performance was evaluated using five-fold cross-validation. The subregion with the highest score for patients with synchronous distant metastasis was identified across all cohorts. The mi-SVM model yielded an average area under the curve (AUC) of 0.812 in the training cohort and 0.805 in the validation cohort. In the entire cohort of patients with synchronous distant metastasis, subregion 2, characterized by tumor periphery and intratumoral transitional components, accounted for the highest proportion (48.57%, 30.6/63) among all subregions. It represents a high-risk subregion for synchronous distant metastasis of clear cell renal cell carcinoma. The peripheral and intratumoral transition zones of clear cell renal cell carcinoma are high-risk subregions associated with synchronous distant metastasis.

Lung nodule classification using radiomics model trained on degraded SDCT images

Background and ObjectiveLow-dose computed tomography (LDCT) screening has shown promise in reducing lung cancer mortality; however, it suffers from high false positive rates and a scarcity of available annotated datasets. To overcome these challenges, we propose a novel approach using synthetic LDCT images generated from standard-dose CT (SDCT) scans from the LIDC-IDRI dataset. Our objective is to develop and validate an interpretable radiomics-based model for distinguishing likely benign from likely malignant pulmonary nodules. MethodsFrom a total of 1010 CT images (695 SDCTs and 315 LDCTs), we degraded SDCTs in the sinogram domain and obtained 1950 nodules as the training set. The 675 nodules from the LDCTs were stratified into 50%-50% partitions for validation and testing. Radiomic features were extracted from nodules, and three feature sets were assessed using: a) only shape and size (SS) features, b) all features but SS features, and c) all features. A systematic pipeline was developed to optimize the feature set and evaluate multiple machine learning models. Models were trained using degraded SDCT, validated and tested on the LDCT nodules. ResultsTraining a logistic regression model using three SS features yielded the most promising results, achieving on the test set mean balanced accuracy, sensitivity, specificity, and AUC-ROC scores of 0.81, 0.76, 0.85, and 0.87, respectively. ConclusionsOur study demonstrates the feasibility and effectiveness of using synthetic LDCT images for developing a relatively accurate radiomics-based model in lung nodule classification. This approach addresses challenges associated with LDCT screening, offering potential implications for improving lung cancer detection and reducing false positives.

Radiomics model for predicting distant metastasis in soft tissue sarcoma of the extremities and trunk treated with surgery.

The aim of this study was to develop a radiomics model based on magnetic resonance imaging (MRI) for predicting metastasis in soft tissue sarcomas (STSs) treated with surgery. MRI and clinical data of 73 patients with STSs of the extremities and trunk were obtained from TCIA database and Jiangsu Cancer Hospital as the training set, data of other 40 patients were retrospectively collected at our institution as the external validation set. Radiomics features were extracted from both intratumoral and peritumoral regions of fat-suppressed T2-weighted images (FS-T2WIs) of patients, and 3D ResNet10 was used to extract deep learning features. Recursive feature elimination (RFE) and least absolute shrinkage and selection operator (LASSO) algorithms were used for the selection of features. Based on 4 different sets of features, 5 machine learning algorithms were used to construct intratumor, peritumor, combined intratumor and peritumor radiomics models and deep learning radiomics (DLR) model. The area under the ROC curve (AUC) and Decision curve analysis (DCA) were used to evaluate the ability of models to predict metastasis. Based on 20 selected features from the deep-learning and radiomics features set, the DLR model was able to predict metastasis in the validation dataset, with an AUC of 0.9770. The DCA and Hosmer-Lemeshow test revealed that the DLR model had good clinical benefit and consistency. By getting richer information from MRI, The DLR model is a noninvasive, low-cost method for predicting the risk of metastasis in STSs, and can help develop appropriate treatment programs.