Scientific Status Quo of Small Renal Lesions: Diagnostic Assessment and Radiomics

Journal of Clinical Medicine, Journal Year: 2024, Volume and Issue: 13(2), P. 547 - 547

Published: Jan. 18, 2024

Background: Small renal masses (SRMs) are defined as contrast-enhanced lesions less than or equal to 4 cm in maximal diameter, which can be compatible with stage T1a cell carcinomas (RCCs). Currently, 50–61% of all tumors found incidentally. Methods: The characteristics the lesion influence choice type management, include several methods SRM including nephrectomy, partial ablation, observation, and also stereotactic body radiotherapy. Typical imaging available for differentiating benign from malignant ultrasound (US), (CEUS), computed tomography (CT), magnetic resonance (MRI). Results: Although is first technique used detect small lesions, it has limitations. CT main most widely characterization. advantages MRI compared better contrast resolution tissue characterization, use functional sequences, possibility performing examination patients allergic iodine-containing medium, absence exposure ionizing radiation. For a correct evaluation during follow-up, necessary reliable method assessment represented by Bosniak classification system. This was initially developed based on findings, 2019 revision proposed inclusion features; however, latest not yet received widespread validation. Conclusions: radiomics an emerging increasingly central field applications such characterizing masses, distinguishing RCC subtypes, monitoring response targeted therapeutic agents, prognosis metastatic context.

Language: Английский

---

Machine learning model based on enhanced CT radiomics for the preoperative prediction of lymphovascular invasion in esophageal squamous cell carcinoma

Frontiers in Oncology, Journal Year: 2024, Volume and Issue: 14

Published: Feb. 23, 2024

Objective To evaluate the value of a machine learning model using enhanced CT radiomics features in prediction lymphovascular invasion (LVI) esophageal squamous cell carcinoma (ESCC) before treatment. Methods We reviewed and analyzed images 258 ESCC patients from June 2017 to December 2019. randomly assigned ratio 7:3 training set (182 cases) validation (76 set. Clinical risk factors image characteristics were recorded, multifactor logistic regression was used screen independent LVI patients. extracted FAE software screened maximum relevance minimum redundancy (MRMR) least absolute shrinkage selection operator (LASSO) algorithms, finally, labels each patient established. Five namely, support vector (SVM), K-nearest neighbor (KNN), (LR), Gauss naive Bayes (GNB), multilayer perceptron (MLP), construct labels, its clinical screened. The predictive efficacy for evaluated receiver operating characteristic (ROC) curve. Results Tumor thickness [OR = 1.189, 95% confidence interval (CI) 1.060–1.351, P 0.005], tumor-to-normal wall enhancement (TNR) (OR 2.966, CI 1.174–7.894, 0.024), N stage 5.828, 1.752–20.811, 0.005) determined as LVI. 1,316 preoperative selected 14 MRMR LASSO labels. In test set, SVM, KNN, LR, GNB showed high performance, while MLP had poor performance. area under curve (AUC) values 0.945 0.905 KNN SVM models, but these decreased 0.866 0.867 indicating significant overfitting. LR models AUC 0.911 0.900 0.893 with stable performance good fitting ability. 0.658 0.674 sets, A multiscale combined constructed multivariate has an (0.870–0.951) (0.840–0.962), accuracy 84.4% 79.7%, sensitivity 90.8% 87.1%, specificity 80.5% 79.0% respectively. Conclusion Machine can preoperatively predict condition effectively based on features. exhibit stability may bring new way non-invasive

Language: Английский

---

Deep learning-based multi-model prediction for disease-free survival status of patients with clear cell renal cell carcinoma after surgery: a multicenter cohort study

International Journal of Surgery, Journal Year: 2024, Volume and Issue: unknown

Published: March 4, 2024

Background: Although separate analysis of individual factor can somewhat improve the prognostic performance, integration multimodal information into a single signature is necessary to stratify patients with clear cell renal carcinoma (ccRCC) for adjuvant therapy after surgery. Methods: A total 414 whole slide images, computed tomography and clinical data from three patient cohorts were retrospectively analyzed. The authors performed deep learning machine algorithm construct single-modality prediction models disease-free survival ccRCC based on segmentation, respectively. multimodel (MMPS) further developed by combining tumor stage/grade system. Prognostic performance model was also verified in two independent validation cohorts. Results: Single-modality well predicting status ccRCC. MMPS achieved higher area under curve value 0.742, 0.917, 0.900 cohorts, could distinguish worse survival, HR 12.90 (95% CI: 2.443–68.120, P <0.0001), 11.10 5.467–22.520, 8.27 1.482–46.130, <0.0001) different In addition, outperformed current factors, which provide complements risk stratification Conclusion: Our novel exhibited significant improvements After multiple centers regions, system be potential practical tool clinicians treatment patients.

Language: Английский

---

Multicenter evaluation of CT deep radiomics model in predicting Leibovich score risk groups for non-metastatic clear cell renal cell carcinoma

Displays, Journal Year: 2024, Volume and Issue: unknown, P. 102867 - 102867

Published: Oct. 1, 2024

Language: Английский

---

Development and Validation of a Prediction Model for Differentiation of Benign and Malignant Fat-Poor Renal Tumors Using CT Radiomics

Applied Sciences, Journal Year: 2023, Volume and Issue: 13(20), P. 11345 - 11345

Published: Oct. 16, 2023

Objectives: To develop and validate a machine learning-based CT radiomics classification model for distinguishing benign renal tumors from malignant tumors. Methods: We reviewed 499 patients who underwent nephrectomy solid at our institution between 2003 2021. In this retrospective study, had undergone computed tomography (CT) scan within 3 months before surgery were included. randomly divided the dataset in stratified manner as follows: 75% training set 25% test set. By using various feature selection methods dimensionality reduction method exclusively set, we selected 160 radiomic features out of 1,288 to classify Results: The included 396 patients, 103 patients. percentage extracted was 32% (385/1218) after reproducibility test. terms average Area Under Receiver Operating Characteristic Curve (AU-ROC) Precision-Recall (AU-PRC), Random Forest achieved better performance (AU-ROC = 0.725; AU-PRC 0.899). An accuracy 0.778 obtained on evaluation with hold-out At optimal threshold, showed an F1 score 0.746, precision 0.862, sensitivity 0.657, specificity 0.651, Negative Predictive Value (NPV) 0.364. Conclusions: Our performed well independent indicating that it could be useful tool discriminating

Language: Английский

---

Research Progress on Renal Cell Carcinoma and Radiomics

Advances in Clinical Medicine, Journal Year: 2024, Volume and Issue: 14(04), P. 1706 - 1712

Published: Jan. 1, 2024

Language: Английский

---

AI Predictive Modeling of Survival Outcomes for Renal Cancer Patients Undergoing Targeted Therapy

Research Square (Research Square), Journal Year: 2024, Volume and Issue: unknown

Published: July 3, 2024

Background: Renal clear cell cancer (RCC) is a complex and heterogeneous disease, posing significant challenges in predicting patient outcomes. The introduction of targeted drug therapy has improved treatment outcomes, but there still pressing need for personalized effective planning. Artificial intelligence (AI) emerged as promising tool addressing this challenge, enabling the development predictive models that can accurately forecast survival periods. By harnessing power AI, clinicians be empowered with decision support, patients to receive more tailored plans enhance both efficacy quality life. Methods: To achieve goal, we conducted retrospective analysis clinical data from Cancer Imaging Archive (TCIA) categorized RCC receiving into two groups: Group 1 (anticipated lifespan exceeding 3 years) 2 less than years). We utilized UPerNet algorithm extract pertinent features CT markers tumors validate their efficacy. extracted were then used develop an AI-based model was trained on dataset. Results: developed AI demonstrated remarkable accuracy, achieving rate 93.66% 94.14% 2. Conclusions: In conclusion, our study demonstrates potential technology time undergoing therapy. established prediction exhibits high accuracy stability, serving valuable facilitate patients. This highlights importance integrating decision-making, overall

Language: Английский

---

Radiomics predicts the prognosis of patients with clear cell renal cell carcinoma by reflecting the tumor heterogeneity and microenvironment

Cancer Imaging, Journal Year: 2024, Volume and Issue: 24(1)

Published: Sept. 16, 2024

Abstract Purpose We aimed to develop and externally validate a CT-based deep learning radiomics model for predicting overall survival (OS) in clear cell renal carcinoma (ccRCC) patients, investigate the association of with tumor heterogeneity microenvironment. Methods The clinicopathological data contrast-enhanced CT images 512 ccRCC patients from three institutions were collected. A total 3566 features extracted 3D regions interest. generated score (DLRS), validated this using an external cohort TCIA. Patients divided into high low-score groups by DLRS. Sequencing corresponding TCGA used reveal differences microenvironment between different groups. What’s more, univariate multivariate Cox regression identify independent risk factors poor OS after operation. combined was developed incorporating DLRS features. SHapley Additive exPlanation method interpretation predictive results. Results At analysis, identified as factor OS. genomic landscape investigated. significantly varied both In test cohort, had great performance, AUCs (95%CI) 1, 3 5-year 0.879(0.868–0.931), 0.854(0.819–0.899) 0.831(0.813–0.868), respectively. There significant difference time stratified model. This showed discrimination calibration, outperforming existing prognostic models (all p values < 0.05). Conclusion allowed prediction clinicopathologic could reflect

Language: Английский

---