Machine Learning-Based MRI Texture Analysis to Predict the Histologic Grade of Oral Squamous Cell Carcinoma

Assessing the Reporting Quality of Machine Learning Algorithms in Head and Neck Oncology

OBJECTIVE

This study aimed to assess reporting quality of machine learning (ML) algorithms in the head and neck oncology literature using the TRIPOD-AI criteria.

DATA SOURCES

A comprehensive search was conducted using PubMed, Scopus, Embase, and Cochrane Database of Systematic Reviews, incorporating search terms related to "artificial intelligence," "machine learning," "deep learning," "neural network," and various head and neck neoplasms.

REVIEW METHODS

Two independent reviewers analyzed each published study for adherence to the 65-point TRIPOD-AI criteria. Items were classified as "Yes," "No," or "NA" for each publication. The proportion of studies satisfying each TRIPOD-AI criterion was calculated. Additionally, the evidence level for each study was evaluated independently by two reviewers using the Oxford Centre for Evidence-Based Medicine (OCEBM) Levels of Evidence. Discrepancies were reconciled through discussion until consensus was reached.

RESULTS

The study highlights the need for improvements in ML algorithm reporting in head and neck oncology. This includes more comprehensive descriptions of datasets, standardization of model performance reporting, and increased sharing of ML models, data, and code with the research community. Adoption of TRIPOD-AI is necessary for achieving standardized ML research reporting in head and neck oncology.

CONCLUSION

Current reporting of ML algorithms hinders clinical application, reproducibility, and understanding of the data used for model training. To overcome these limitations and improve patient and clinician trust, ML developers should provide open access to models, code, and source data, fostering iterative progress through community critique, thus enhancing model accuracy and mitigating biases.

LEVEL OF EVIDENCE

NA Laryngoscope, 135:687-694, 2025.

Image-based multi-omics analysis for oral science: Recent progress and perspectives

OBJECTIVES

The diagnosis and treatment of oral and dental diseases rely heavily on various types of medical imaging. Deep learning-mediated multi-omics analysis can extract more representative features than those identified through traditional diagnostic methods. This review aims to discuss the applications and recent advances in image-based multi-omics analysis in oral science and to highlight its potential to enhance traditional diagnostic approaches for oral diseases.

STUDY SELECTION, DATA, AND SOURCES

A systematic search was conducted in the PubMed, Web of Science, and Google Scholar databases, covering all available records. This search thoroughly examined and summarized advances in image-based multi-omics analysis in oral and maxillofacial medicine.

CONCLUSIONS

This review comprehensively summarizes recent advancements in image-based multi-omics analysis for oral science, including radiomics, pathomics, and photographic-based omics analysis. It also discusses the ongoing challenges and future perspectives that could provide new insights into exploiting the potential of image-based omics analysis in the field of oral science.

CLINICAL SIGNIFICANCE

This review article presents the state of image-based multi-omics analysis in stomatology, aiming to help oral clinicians recognize the utility of combining omics analyses with imaging during diagnosis and treatment, which can improve diagnostic accuracy, shorten times to diagnosis, save medical resources, and reduce disparity in professional knowledge among clinicians.

Development of a flexible feature selection framework in radiomics-based prediction modeling: Assessment with four real-world datasets

There are several important challenges in radiomics research; one of them is feature selection. Since many quantitative features are non-informative, feature selection becomes essential. Feature selection methods have been mixed with filter, wrapper, and embedded methods without a rule of thumb. This study aims to develop a framework for optimal feature selection in radiomics research. We developed the framework that the optimal features were selected to quickly through controlling relevance and redundancy among features. A 'FeatureMap' was generated containing information for each step and used as a platform. Through this framework, we can explore the optimal combination of radiomics features and evaluate the predictive performance using only selected features. We assessed the framework using four real datasets. The FeatureMap generated 6 combinations, with the number of features selected varying for each combination. The predictive models obtained high performances; the highest test area under the curves (AUCs) were 0.792, 0.820, 0.846 and 0.738 in the cross-validation method, respectively. We developed a flexible framework for feature selection methods in radiomics research and assessed its usefulness using various real-world data. Our framework can assist clinicians in efficiently developing predictive models based on radiomics.

Enhanced Diagnostic Precision: Assessing Tumor Differentiation in Head and Neck Squamous Cell Carcinoma Using Multi-Slice Spiral CT Texture Analysis

This study explores the efficacy of texture analysis by using preoperative multi-slice spiral computed tomography (MSCT) to non-invasively determine the grade of cellular differentiation in head and neck squamous cell carcinoma (HNSCC). In a retrospective study, MSCT scans of patients with HNSCC were analyzed and classified based on its histological grade as moderately differentiated, well-differentiated, or poorly differentiated. The location of the tumor was categorized as either in the bone or in soft tissues. Segmentation of the lesion areas was conducted, followed by texture analysis. Eleven GLCM parameters across five different distances were calculated. Median values and correlations of texture parameters were examined in relation to tumor differentiation grade by using Spearman's correlation coefficient and Kruskal-Wallis and Dunn tests. Forty-six patients were included, predominantly female (87%), with a mean age of 66.7 years. Texture analysis revealed significant parameter correlations with histopathological grades of tumor differentiation. The study identified no significant age correlation with tumor differentiation, which underscores the potential of texture analysis as an age-independent biomarker. The strong correlations between texture parameters and histopathological grades support the integration of this technique into the clinical decision-making process.

Machine learning-based MRI radiomics for assessing the level of tumor infiltrating lymphocytes in oral tongue squamous cell carcinoma: a pilot study

BACKGROUND

To investigate the value of machine learning (ML)-based magnetic resonance imaging (MRI) radiomics in assessing tumor-infiltrating lymphocyte (TIL) levels in patients with oral tongue squamous cell carcinoma (OTSCC).

METHODS

The study included 68 patients with pathologically diagnosed OTSCC (30 with high TILs and 38 with low TILs) who underwent pretreatment MRI. Based on the regions of interest encompassing the entire tumor, a total of 750 radiomics features were extracted from T2-weighted (T2WI) and contrast-enhanced T1-weighted (ceT1WI) imaging. To reduce dimensionality, reproducibility analysis by two radiologists and collinearity analysis were performed. The top six features were selected from each sequence alone, as well as their combination, using the minimum-redundancy maximum-relevance algorithm. Random forest, logistic regression, and support vector machine models were used to predict TIL levels in OTSCC, and 10-fold cross-validation was employed to assess the performance of the classifiers.

RESULTS

Based on the features selected from each sequence alone, the ceT1WI models outperformed the T2WI models, with a maximum area under the curve (AUC) of 0.820 versus 0.754. When combining the two sequences, the optimal features consisted of one T2WI and five ceT1WI features, all of which exhibited significant differences between patients with low and high TILs (all P < 0.05). The logistic regression model constructed using these features demonstrated the best predictive performance, with an AUC of 0.846 and an accuracy of 80.9%.

CONCLUSIONS

ML-based T2WI and ceT1WI radiomics can serve as valuable tools for determining the level of TILs in patients with OTSCC.

MRI radiomics to monitor therapeutic outcome of sorafenib plus IHA transcatheter NK cell combination therapy in hepatocellular carcinoma

BACKGROUND

Hepatocellular carcinoma (HCC) is a common liver malignancy with limited treatment options. Previous studies expressed the potential synergy of sorafenib and NK cell immunotherapy as a promising approach against HCC. MRI is commonly used to assess response of HCC to therapy. However, traditional MRI-based metrics for treatment efficacy are inadequate for capturing complex changes in the tumor microenvironment, especially with immunotherapy. In this study, we investigated potent MRI radiomics analysis to non-invasively assess early responses to combined sorafenib and NK cell therapy in a HCC rat model, aiming to predict multiple treatment outcomes and optimize HCC treatment evaluations.

METHODS

Sprague Dawley (SD) rats underwent tumor implantation with the N1-S1 cell line. Tumor progression and treatment efficacy were assessed using MRI following NK cell immunotherapy and sorafenib administration. Radiomics features were extracted, processed, and selected from both T1w and T2w MRI images. The quantitative models were developed to predict treatment outcomes and their performances were evaluated with area under the receiver operating characteristic (AUROC) curve. Additionally, multivariable linear regression models were constructed to determine the correlation between MRI radiomics and histology, aiming for a noninvasive evaluation of tumor biomarkers. These models were evaluated using root-mean-squared-error (RMSE) and the Spearman correlation coefficient.

RESULTS

A total of 743 radiomics features were extracted from T1w and T2w MRI data separately. Subsequently, a feature selection process was conducted to identify a subset of five features for modeling. For therapeutic prediction, four classification models were developed. Support vector machine (SVM) model, utilizing combined T1w + T2w MRI data, achieved 96% accuracy and an AUROC of 1.00 in differentiating the control and treatment groups. For multi-class treatment outcome prediction, Linear regression model attained 85% accuracy and an AUC of 0.93. Histological analysis showed that combination therapy of NK cell and sorafenib had the lowest tumor cell viability and the highest NK cell activity. Correlation analyses between MRI features and histological biomarkers indicated robust relationships (r = 0.94).

CONCLUSIONS

Our study underscored the significant potential of texture-based MRI imaging features in the early assessment of multiple HCC treatment outcomes.

An MRI-based radiomics nomogram for detecting cervical esophagus invasion in hypopharyngeal squamous cell carcinoma

BACKGROUND

Accurate detection of cervical esophagus invasion (CEI) in HPSCC is challenging but crucial. We aimed to investigate the value of magnetic resonance imaging (MRI)-based radiomics for detecting CEI in patients with HPSCC.

METHODS

This retrospective study included 151 HPSCC patients with or without CEI, which were randomly assigned into a training (n = 101) or validation (n = 50) cohort. A total of 750 radiomics features were extracted from T2-weighted imaging (T2WI) and contrast-enhanced T1-weighted imaging (ceT1WI), respectively. A radiomics signature was constructed using the least absolute shrinkage and selection operator method. Multivariable logistic regression analyses were adopted to establish a clinical model and a radiomics nomogram. Two experienced radiologists evaluated the CEI status based on morphological findings. Areas under the curve (AUCs) of the models and readers were compared using the DeLong method. The performance of the nomogram was also assessed by its calibration and clinical usefulness.

RESULTS

The radiomics signature, consisting of five T2WI and six ceT1WI radiomics features, was significantly associated with CEI in both cohorts (all p < 0.001). The radiomics nomogram combining the radiomics signature and clinical T stage achieved significantly higher predictive value than the clinical model and pooled readers in the training (AUC 0.923 vs. 0.723 and 0.621, all p < 0.001) and validation (AUC 0.888 vs. 0.754 and 0.647, all p < 0.05) cohorts. The radiomics nomogram showed favorable calibration in both cohorts and provided better net benefit than the clinical model.

CONCLUSIONS

The MRI-based radiomics nomogram is a promising method for detecting CEI in HPSCC.

Early Predicting Tribocorrosion Rate of Dental Implant Titanium Materials Using Random Forest Machine Learning Models

Early detection and prediction of bio-tribocorrosion can avert unexpected damage that may lead to secondary revision surgery and associated risks of implantable devices. Therefore, this study sought to develop a state-of-the-art prediction technique leveraging machine learning(ML) models to classify and predict the possibility of mechanical degradation in dental implant materials. Key features considered in the study involving pure titanium and titanium-zirconium (zirconium = 5, 10, and 15 in wt%) alloys include corrosion potential, acoustic emission(AE) absolute energy, hardness, and weight-loss estimates. ML prototype models deployed confirms its suitability in tribocorrosion prediction with an accuracy above 90%. Proposed system can evolve as a continuous structural-health monitoring as well as a reliable predictive modeling technique for dental implant monitoring.

Whole-tumor histogram analysis of apparent diffusion coefficient maps with machine learning algorithms for predicting histologic grade of sinonasal squamous cell carcinoma: a preliminary study

PURPOSE

Accurate histologic grade assessment is helpful for clinical decision making and prognostic assessment of sinonasal squamous cell carcinoma (SNSCC). This research aimed to explore whether whole-tumor histogram analysis of apparent diffusion coefficient (ADC) maps with machine learning algorithms can predict histologic grade of SNSCC.

METHODS

One hundred and forty-seven patients with pathologically diagnosed SNSCC formed this retrospective study. Sixty-six patients were low-grade (grade I/II) and eighty-one patients were high-grade (grade III). Eighteen histogram features were obtained from quantitative ADC maps. Additionally, the mean ADC value and clinical features were analyzed for comparison with histogram features. Machine learning algorithms were applied to build the best diagnostic model for predicting histological grade. The receiver operating characteristic (ROC) curve was used to evaluate the performance of each model prediction, and the area under the ROC curve (AUC) were analyzed.

RESULTS

The histogram model based on three features (10th Percentile, Mean, and 90th Percentile) with support vector machine (SVM) classifier demonstrated excellent diagnostic performance, with an AUC of 0.947 on the testing dataset. The AUC of the histogram model was similar to that of the mean ADC value model (0.947 vs 0.957; P = 0.7029). The poor diagnostic performance of the clinical model (AUC = 0.692) was improved by the combined model incorporating histogram features or mean ADC value (P < 0.05).

CONCLUSION

ADC histogram analysis improved the projection of SNSCC histologic grade, compared with clinical model. The complex histogram model had comparable but not better performance than mean ADC value model.

Integration of IDPC Clustering Analysis and Interpretable Machine Learning for Survival Risk Prediction of Patients with ESCC

Precise forecasting of survival risk plays a pivotal role in comprehending and predicting the prognosis of patients afflicted with esophageal squamous cell carcinoma (ESCC). The existing methods have the problems of insufficient fitting ability and poor interpretability. To address this issue, this work proposes a novel interpretable survival risk prediction method for ESCC patients based on extreme gradient boosting improved by whale optimization algorithm (WOA-XGBoost) and shapley additive explanations (SHAP). Given the imbalanced nature of the data set, the adaptive synthetic sampling (ADASYN) is first used to generate the samples with high survival risk. Then, an improved clustering by fast search and find of density peaks (IDPC) algorithm based on cosine distance and K nearest neighbors is used to cluster the patients. Next, the prediction model for each cluster is obtained by WOA-XGBoost and the constructed model is visualized with SHAP to uncover the factors hidden in the structured model and improve the interpretability of the black-box model. Finally, the effectiveness of the proposed scheme is demonstrated by analyzing the data collected from the First Affiliated Hospital of Zhengzhou University. The results of the analysis reveal that the proposed methodology exhibits superior performance, as indicated by the area under the receiver operating characteristic curve (AUROC) of 0.918 and accuracy of 0.881.

Data-centric artificial intelligence in oncology: a systematic review assessing data quality in machine learning models for head and neck cancer

Machine learning models have been increasingly considered to model head and neck cancer outcomes for improved screening, diagnosis, treatment, and prognostication of the disease. As the concept of data-centric artificial intelligence is still incipient in healthcare systems, little is known about the data quality of the models proposed for clinical utility. This is important as it supports the generalizability of the models and data standardization. Therefore, this study overviews the quality of structured and unstructured data used for machine learning model construction in head and neck cancer. Relevant studies reporting on the use of machine learning models based on structured and unstructured custom datasets between January 2016 and June 2022 were sourced from PubMed, EMBASE, Scopus, and Web of Science electronic databases. Prediction model Risk of Bias Assessment (PROBAST) tool was used to assess the quality of individual studies before comprehensive data quality parameters were assessed according to the type of dataset used for model construction. A total of 159 studies were included in the review; 106 utilized structured datasets while 53 utilized unstructured datasets. Data quality assessments were deliberately performed for 14.2% of structured datasets and 11.3% of unstructured datasets before model construction. Class imbalance and data fairness were the most common limitations in data quality for both types of datasets while outlier detection and lack of representative outcome classes were common in structured and unstructured datasets respectively. Furthermore, this review found that class imbalance reduced the discriminatory performance for models based on structured datasets while higher image resolution and good class overlap resulted in better model performance using unstructured datasets during internal validation. Overall, data quality was infrequently assessed before the construction of ML models in head and neck cancer irrespective of the use of structured or unstructured datasets. To improve model generalizability, the assessments discussed in this study should be introduced during model construction to achieve data-centric intelligent systems for head and neck cancer management.

Texture Feature-Based Machine Learning Classification on MRI Image for Sepsis-Associated Encephalopathy Detection: A Pilot Study

Objective

The objective of this study was to assess the performance of combining MRI-based texture analysis with machine learning for differentiating sepsis-associated encephalopathy (SAE) from sepsis alone.

Method

Sixty-six MRI-T1WI images of an SAE patient and 125 images of patients with sepsis alone were collected. Frontal lobe, brain stem, hippocampus, and amygdala were selected as regions of interest (ROIs). 279 texture features of each ROI were obtained using MaZda software. After the dimension reduction, 30 highly discriminative features of each ROI were adopted to differentiate SAE from sepsis alone using the CatBoost model.

Results

The classification models of frontal, brain stem, hippocampus, and amygdala were constructed. The classification accuracy was above 0.83, and the area under the curve (AUC) exceeded 0.90 in the validation set.

Conclusion

The texture features differed between SAE patients and patients with sepsis alone in different anatomical locations, suggesting that MRI-based texture analysis with machine learning might be helpful in differentiating SAE from sepsis alone.

Efficacy of magnetic resonance imaging texture features of the lateral pterygoid muscle in distinguishing rheumatoid arthritis and osteoarthritis of the temporomandibular joint

OBJECTIVES

The aim of this study was to assess whether magnetic resonance imaging (MRI) texture features of the lateral pterygoid muscle can distinguish between rheumatoid arthritis (RA) and osteoarthritis (OA) of the temporomandibular joint (TMJ).

METHODS

The authors extracted 279 texture features from 36 patients with RA and OA from the region of interest set for the lateral pterygoid muscle on short tau inversion recovery (STIR) images using MaZda Ver.3.3. A total of 10 texture features were selected using Fisher's coefficients, as well as probability of error and average correlation coefficients. Data observed to have a non-normal distribution using the Kolmogorov-Smirnov test were compared using the Mann-Whitney U-test. Receiver operating characteristic (ROC) curves were used to assess the ability of the 10 texture features to distinguish RA and OA of the TMJ.

RESULTS

A total of 10 features (5 Correlation, 3 Run Length Nonuniformity, 1 Sigma, and 1 Teta) were selected from 279 texture features. These texture features revealed significant differences between the RA and OA groups (p < 0.01). The sensitivity, specificity, accuracy, and area under the ROC curve of the texture features for distinguishing RA from OA were 0.78-0.94, 0.89-1.0, 0.86-0.92, and 0.89-0.95, respectively.

CONCLUSION

MRI texture analysis of the lateral pterygoid muscle may be useful for distinguishing between RA and OA of the TMJ.

Machine-Learning Applications in Oral Cancer: A Systematic Review

Over the years, several machine-learning applications have been suggested to assist in various clinical scenarios relevant to oral cancer. We offer a systematic review to identify, assess, and summarize the evidence for reported uses in the areas of oral cancer detection and prevention, prognosis, pre-cancer, treatment, and quality of life. The main algorithms applied in the context of oral cancer applications corresponded to SVM, ANN, and LR, comprising 87.71% of the total published articles in the field. Genomic, histopathological, image, medical/clinical, spectral, and speech data were used most often to predict the four areas of application found in this review. In conclusion, our study has shown that machine-learning applications are useful for prognosis, diagnosis, and prevention of potentially malignant oral lesions (pre-cancer) and therapy. Nevertheless, we strongly recommended the application of these methods in daily clinical practice.

BID-Net: An Automated System for Bone Invasion Detection Occurring at Stage T4 in Oral Squamous Carcinoma Using Deep Learning

Detection of the presence and absence of bone invasion by the tumor in oral squamous cell carcinoma (OSCC) patients is very significant for their treatment planning and surgical resection. For bone invasion detection, CT scan imaging is the preferred choice of radiologists because of its high sensitivity and specificity. In the present work, deep learning algorithm based model, BID-Net, has been proposed for the automation of bone invasion detection. BID-Net performs the binary classification of CT scan images as the images with bone invasion and images without bone invasion. The proposed BID-Net model has achieved an outstanding accuracy of 93.62%. The model is also compared with six Transfer Learning models like VGG16, VGG19, ResNet-50, MobileNetV2, DenseNet-121, ResNet-101 and BID-Net outperformed over the other models. As there exists no previous studies on bone invasion detection using Deep Learning models, so the results of the proposed model have been validated from the experts of practitioner radiologists, S.M.S. hospital, Jaipur, India.

Application of machine learning in oral and maxillofacial surgery

Oral and maxillofacial anatomy is extremely complex, and medical imaging is critical in the diagnosis and treatment of soft and bone tissue lesions. Hence, there exists accumulating imaging data without being properly utilized over the last decades. As a result, problems are emerging regarding how to integrate and interpret a large amount of medical data and alleviate clinicians’ workload. Recently, artificial intelligence has been developing rapidly to analyze complex medical data, and machine learning is one of the specific methods of achieving this goal, which is based on a set of algorithms and previous results. Machine learning has been considered useful in assisting early diagnosis, treatment planning, and prognostic estimation through extracting key features and building mathematical models by computers. Over the past decade, machine learning techniques have been applied to the field of oral and maxillofacial surgery and increasingly achieved expert-level performance. Thus, we hold a positive attitude towards developing machine learning for reducing the number of medical errors, improving the quality of patient care, and optimizing clinical decision-making in oral and maxillofacial surgery. In this review, we explore the clinical application of machine learning in maxillofacial cysts and tumors, maxillofacial defect reconstruction, orthognathic surgery, and dental implant and discuss its current problems and solutions.

Prediction models applying machine learning to oral cavity cancer outcomes: A systematic review

OBJECTIVES

Machine learning platforms are now being introduced into modern oncological practice for classification and prediction of patient outcomes. To determine the current status of the application of these learning models as adjunctive decision-making tools in oral cavity cancer management, this systematic review aims to summarize the accuracy of machine-learning based models for disease outcomes.

METHODS

Electronic databases including PubMed, Scopus, EMBASE, Cochrane Library, LILACS, SciELO, PsychINFO, and Web of Science were searched up until December 21, 2020. Pertinent articles detailing the development and accuracy of machine learning prediction models for oral cavity cancer outcomes were selected in a two-stage process. Quality assessment was conducted using the Quality in Prognosis Studies (QUIPS) tool and results of base studies were qualitatively synthesized by all authors. Outcomes of interest were malignant transformation of precancer lesions, cervical lymph node metastasis, as well as treatment response, and prognosis of oral cavity cancer.

RESULTS

Twenty-seven articles out of 950 citations identified from electronic and manual searching were included in this study. Five studies had low bias concerns on the QUIPS tool. Prediction of malignant transformation, cervical lymph node metastasis, treatment response, and prognosis were reported in three, six, eight, and eleven articles respectively. Accuracy of these learning models on the internal or external validation sets ranged from 0.85 to 0.97 for malignant transformation prediction, 0.78-0.91 for cervical lymph node metastasis prediction, 0.64-1.00 for treatment response prediction, and 0.71-0.99 for prognosis prediction. In general, most trained algorithms predicting these outcomes performed better than alternate methods of prediction. We also found that models including molecular markers in training data had better accuracy estimates for malignant transformation, treatment response, and prognosis prediction.

CONCLUSION

Machine learning algorithms have a satisfactory to excellent accuracy for predicting three of four oral cavity cancer outcomes i.e., malignant transformation, nodal metastasis, and prognosis. However, considering the training approach of many available classifiers, these models may not be streamlined enough for clinical application currently.

Machine learning in dental, oral and craniofacial imaging: a review of recent progress

Artificial intelligence has been emerging as an increasingly important aspect of our daily lives and is widely applied in medical science. One major application of artificial intelligence in medical science is medical imaging. As a major component of artificial intelligence, many machine learning models are applied in medical diagnosis and treatment with the advancement of technology and medical imaging facilities. The popularity of convolutional neural network in dental, oral and craniofacial imaging is heightening, as it has been continually applied to a broader spectrum of scientific studies. Our manuscript reviews the fundamental principles and rationales behind machine learning, and summarizes its research progress and its recent applications specifically in dental, oral and craniofacial imaging. It also reviews the problems that remain to be resolved and evaluates the prospect of the future development of this field of scientific study.