MRI-based deep learning and radiomics for prediction of occult cervical lymph node metastasis and prognosis in early-stage oral and oropharyngeal squamous cell carcinoma: a diagnostic study

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

Published: May 9, 2024

The incidence of occult cervical lymph node metastases (OCLNM) is reported to be 20-30% in early-stage oral cancer and oropharyngeal cancer. There a lack an accurate diagnostic method predict metastasis help surgeons make precise treatment decisions.

Language: Английский

---

The application of radiomics in cancer imaging with a focus on lung cancer, renal cell carcinoma, gastrointestinal cancer, and head and neck cancer: A systematic review

Physica Medica, Journal Year: 2025, Volume and Issue: 130, P. 104891 - 104891

Published: Jan. 8, 2025

Language: Английский

---

Comparative analysis of radiomics and deep-learning algorithms for survival prediction in hepatocellular carcinoma

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: Jan. 5, 2024

Abstract To examine the comparative robustness of computed tomography (CT)-based conventional radiomics and deep-learning convolutional neural networks (CNN) to predict overall survival (OS) in HCC patients. Retrospectively, 114 patients with pretherapeutic CT liver were randomized into a development (n = 85) validation 29) cohort, including all tumor stages several applied therapies. In addition clinical parameters, image annotations parenchyma findings on available. Cox-regression based features CNN models established combined parameters OS. Model performance was assessed using concordance index (C-index). Log-rank tests used test model-based patient stratification high/low-risk groups. The model achieved best for OS (C-index [95% confidence interval (CI)] 0.74 [0.57–0.86]) significant difference between risk groups (p 0.03). analysis, (lowest C-index [CI] 0.63 [0.39–0.83]; highest 0.71 [0.49–0.88]) superior corresponding 0.51 [0.30–0.73]; 0.66 [0.48–0.79]). A not possible > 0.05). Under conditions, CNN-algorithms demonstrate prognostic potential compared approaches could therefore provide important information setting, especially when data is limited.

Language: Английский

---

High-risk habitat radiomics model based on ultrasound images for predicting lateral neck lymph node metastasis in differentiated thyroid cancer

BMC Medical Imaging, Journal Year: 2025, Volume and Issue: 25(1)

Published: Jan. 13, 2025

This study aims to evaluate the predictive usefulness of a habitat radiomics model based on ultrasound images for anticipating lateral neck lymph node metastasis (LLNM) in differentiated thyroid cancer (DTC), and pinpointing high-risk regions significant traits. A group 214 patients diagnosed with carcinoma (DTC) between August 2021 2023 were included, consisting 107 confirmed postoperative without or cervical involvement. An additional cohort 43 was recruited serve as an independent external testing this study. Patients randomly divided into training internal at 8:2 ratio. Region interest (ROI) manually outlined, analysis subregions defined using K-means method. The ideal number (n = 5) determined Calinski-Harabasz score, leading creation 5 identification model. Area under curve (AUC) values calculated all models assess their validity, nomograms created by integrating clinical features. dataset is employed performance stability In group, Habitat 3 identified study, showing best diagnostic efficacy among (AUC(CRM) vs. AUC(Habitat 3) AUC(CRM + 0.84(95%CI:0.71–0.97) 0.90(95%CI:0.80-1.00) 0.79(95%CI:0.65–0.93)). Moreover, features constructing enhanced capability combined (AUC 0.95(95%CI:0.88-1.00)). utilized model's accuracy, yielding AUC 0.88 (95%CI: 0.78–0.98). integration High-Risk Habitats (Habitat characteristics demonstrated high accuracy identifying LLNM. has potential offer valuable guidance surgeons deciding necessity LLNM dissection DTC. Not applicable.

Language: Английский

---

Novel pre‐spatial data fusion deep learning approach for multimodal volumetric outcome prediction models in radiotherapy

Medical Physics, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 10, 2025

Given the recent increased emphasis on multimodal neural networks to solve complex modeling tasks, problem of outcome prediction for a course treatment can be framed as fundamentally in nature. A patient's response will vary based their specific anatomy and proposed plan-these factors are spatial closely related. However, additional may also have importance, such non-spatial descriptive clinical characteristics, which structured tabular data. It is critical provide models with comprehensive patient representation possible, but inputs differing data structures incompatible raw form; traditional that consider these require feature engineering prior modeling. In networks, organically integrated into model itself, under one governing optimization, rather than performed prescriptively beforehand. native incompatibility different must addressed. Methods reconcile structural incompatibilities called fusion. We present novel joint early pre-spatial (JEPS) fusion technique demonstrate differences approach produce significant performance even when identical. To volumetric its impact pretreatment overall survival (OS). From retrospective cohort 531 head neck patients treated at our clinic, we prepared an OS dataset 222 data-complete cases 2-year post-treatment time threshold. Each included CT imaging, dose array, approved structure set, summary demographics survey establish single-modality baselines, fit both Cox Proportional Hazards (CPH) dense network only data, then trained 3D convolutional (CNN) volume Then, five competing architectures modalities: two models, late model, JEPS, where merged training upstream most convolution operations. used standardized 10-fold cross validation directly compare all identical train/test splits patients, using area receiver-operator curve (AUC) primary metric. two-tailed Student t-test assess statistical significance (p-value threshold 0.05) any observed differences. The JEPS design scored highest, achieving mean AUC 0.779 ± 0.080. clinical-only CPH second third highest 0.746 0.066 0.720 0.091 AUC, respectively. between three were not statistically significant. All other comparison significantly worse top performing model. For evaluation, architecture achieves better integration improves predictive over common approaches. easily applied CNN.

Language: Английский

---

The prognostic value of pathologic lymph node imaging using deep learning-based outcome prediction in oropharyngeal cancer patients

Physics and Imaging in Radiation Oncology, Journal Year: 2025, Volume and Issue: 33, P. 100733 - 100733

Published: Jan. 1, 2025

Deep learning (DL) models can extract prognostic image features from pre-treatment PET/CT scans. The study objective was to explore the potential benefits of incorporating pathologic lymph node (PL) spatial information in addition that primary tumor (PT) DL-based for predicting local control (LC), regional (RC), distant-metastasis-free survival (DMFS), and overall (OS) oropharyngeal cancer (OPC) patients. included 409 OPC patients treated with definitive (chemo)radiotherapy between 2010 2022. Patient data, including scans, manually contoured PT (GTVp) PL (GTVln) structures, clinical variables, endpoints, were collected. Firstly, a method employed segment tumours PET/CT, resulting predicted probability maps (TPMp) (TPMln). Secondly, different combinations CT, PET, manual contours 300 used train outcome prediction each endpoint through 5-fold cross validation. Model performance, assessed by concordance index (C-index), evaluated using test set 100 Including improved C-index results all endpoints except LC. For LC, comparable C-indices (around 0.66) observed trained only those as additional structure. Models combined into single structure achieved highest 0.65 0.80 RC DMFS prediction, respectively. these target structures separate entities 0.70 OS. Incorporating performance RC, DMFS,

Language: Английский

---

Deep learning informed multimodal fusion of radiology and pathology to predict outcomes in HPV-associated oropharyngeal squamous cell carcinoma

EBioMedicine, Journal Year: 2025, Volume and Issue: 114, P. 105663 - 105663

Published: March 23, 2025

Language: Английский

---

Gd-EOB-DTPA-enhanced MRI radiomics and deep learning models to predict microvascular invasion in hepatocellular carcinoma: a multicenter study

BMC Medical Imaging, Journal Year: 2025, Volume and Issue: 25(1)

Published: March 31, 2025

Microvascular invasion (MVI) is an important risk factor for early postoperative recurrence of hepatocellular carcinoma (HCC). Based on gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid (Gd-EOB-DTPA)-enhanced magnetic resonance imaging (MRI) images, we developed a novel radiomics model. It combined bi-regional features and two machine learning algorithms. The aim this study was to validate its potential value preoperative prediction MVI. This retrospective included 304 HCC patients (training cohort, 216 patients; testing 88 patients) from three hospitals. Intratumoral peritumoral volumes interest were delineated in arterial phase, portal venous hepatobiliary phase images. Conventional (CR) deep (DLR) extracted based FeAture Explorer software the 3D ResNet-18 extractor, respectively. Clinical variables selected using univariate multivariate analyses. Clinical, CR, DLR, CR-DLR, clinical-radiomics (Clin-R) models built support vector machines. predictive capacity assessed by area under receiver operating characteristic curve (AUC), accuracy, sensitivity, specificity. CR-DLR model showed more gains gave better performance than single-regional or single-machine models. Its AUC, specificity 0.844, 76.9%, 87.8%, 69.1% training cohort 0.740, 73.9%, 50%, 84.5% cohort. Alpha-fetoprotein (odds ratio 0.32) maximum tumor diameter 1.270) independent predictors. AUCs clinical Clin-R 0.655 0.672, There no significant difference between all (P > 0.005). Gd-EOB-DTPA-enhanced MRI focused developing that combines algorithms (CR DLR). application new will provide accurate non-invasive diagnostic solution medical imaging. valuable information personalized treatment, thereby improving patient prognosis. Not applicable.

Language: Английский

---

Deep Texture Analysis—Enhancing CT Radiomics Features for Prediction of Head and Neck Cancer Treatment Outcomes: A Machine Learning Approach

Radiation, Journal Year: 2024, Volume and Issue: 4(1), P. 50 - 68

Published: Feb. 28, 2024

(1) Background: Some cancer patients do not experience tumour shrinkage but are still at risk of experiencing unwanted treatment side effects. Radiomics refers to mining biomedical images quantify textural characterization. When radiomics features labelled with response, retrospectively, they can train predictive machine learning (ML) models. (2) Methods: were determined from lymph node (LN) segmentations treatment-planning CT scans head and neck (H&N) patients. Binary outcomes (complete response versus partial or no response) for n = 71 used support vector (SVM) k-nearest neighbour (k-NN) classifier models 1–7 features. A deep texture analysis (DTA) methodology was proposed evaluated second- third-layer features, based on common metrics (sensitivity (%Sn), specificity (%Sp), accuracy (%Acc), precision (%Prec), balanced (%Bal Acc)). (3) Results: Models created both classifiers found be able predict the results suggest that inclusion deeper layer enhanced model performance. The best a seven-feature multivariable k-NN trained using three layers %Sn 74%, %Sp 68%, %Acc 72%, %Prec 81%, %Bal Acc 71% an area under curve (AUC) receiver operating characteristic (ROC) 0.700. (4) Conclusions: H&N Cancer patient LN contain phenotypic information regarding DTA improve performance by enhancing feature sets is worth consideration in future studies.

Language: Английский

---

Predicting Response to Exclusive Combined Radio-Chemotherapy in Naso-Oropharyngeal Cancer: The Role of Texture Analysis

Diagnostics, Journal Year: 2024, Volume and Issue: 14(10), P. 1036 - 1036

Published: May 17, 2024

The aim of this work is to identify MRI texture features able predict the response radio-chemotherapy (RT-CHT) in patients with naso-oropharyngeal carcinoma (NPC-OPC) before treatment order help clinical decision making. Textural were derived from ADC maps and post-gadolinium T1-images on a single machine for 37 NPC-OPC. Patients divided into two groups (responders/non-responders) according results scans 18F-FDG-PET/CT performed at follow-up 3–4 12 months after therapy biopsy. Pre-RT-CHT lesions segmented, radiomic extracted. A non-parametric Mann–Whitney test was performed. p-value < 0.05 considered significant. Receiver operating characteristic curves area-under-the-curve values generated; 95% confidence interval (CI) reported. model constructed using LASSO algorithm. After feature selection T1 post-contrast sequences, six statistically significant: gldm_DependenceEntropy DependenceNonUniformity, glrlm_RunEntropy RunLengthNonUniformity, glszm_SizeZoneNonUniformity ZoneEntropy, significant cut-off between responder non-responder group. With algorithm, showed an AUC 0.89 CI: 0.78–0.99. In ADC, five selected 0.84 0.68–1. Texture analysis could potentially NPC-OPC who will undergo exclusive RT-CHT, being, therefore, useful tool therapeutical–clinical

Language: Английский

---