The radiomic-clinical model using the SHAP method for assessing the treatment response of whole-brain radiotherapy: a multicentric study

Pilot study: radiomic analysis for predicting treatment response to whole-brain radiotherapy combined temozolomide in lung cancer brain metastases

Objective

The objective of this study is to assess the viability of utilizing radiomics for predicting the treatment response of lung cancer brain metastases (LCBM) to whole-brain radiotherapy (WBRT) combined with temozolomide (TMZ).

Methods

Fifty-three patients diagnosed with LCBM and undergoing WBRT combined with TMZ were enrolled. Patients were divided into responsive and non-responsive groups based on the RANO-BM criteria. Radiomic features were extracted from contrast-enhanced the whole brain tissue CT images. Feature selection was performed using t-tests, Pearson correlation coefficients, and Least Absolute Shrinkage And Selection (LASSO) regression. Logistic regression was employed to construct the radiomics model, which was then integrated with clinical data to develop the nomogram model. Model performance was evaluated using receiver operating characteristic (ROC) curves, and clinical utility was assessed using decision curve analysis (DCA).

Results

A total of 1834 radiomic features were extracted from each patient's images, and 3 features with predictive value were selected. Both the radiomics and nomogram models exhibited satisfactory predictive performance and clinical utility, with the nomogram model demonstrating superior predictive value. The ROC analysis revealed that the AUC of the radiomics model in the training and testing sets were 0.776 and 0.767, respectively, while the AUC of the nomogram model were 0.799 and 0.833, respectively. DCA curves demonstrated that both models provided benefits to patients across various thresholds.

Conclusion

Radiomic-defined image biomarkers can effectively predict the treatment response of WBRT combined with TMZ in patients with LCBM, offering potential to optimize treatment decisions for this condition.

Effects of environmental phenols on eGFR: machine learning modeling methods applied to cross-sectional studies

Purpose

Limited investigation is available on the correlation between environmental phenols' exposure and estimated glomerular filtration rate (eGFR). Our target is established a robust and explainable machine learning (ML) model that associates environmental phenols' exposure with eGFR.

Methods

Our datasets for constructing the associations between environmental phenols' and eGFR were collected from the National Health and Nutrition Examination Survey (NHANES, 2013-2016). Five ML models were contained and fine-tuned to eGFR regression by phenols' exposure. Regression evaluation metrics were used to extract the limitation of the models. The most effective model was then utilized for regression, with interpretation of its features carried out using shapley additive explanations (SHAP) and the game theory python package to represent the model's regression capacity.

Results

The study identified the top-performing random forest (RF) regressor with a mean absolute error of 0.621 and a coefficient of determination of 0.998 among 3,371 participants. Six environmental phenols with eGFR in linear regression models revealed that the concentrations of triclosan (TCS) and bisphenol S (BPS) in urine were positively correlated with eGFR, and the correlation coefficients were β = 0.010 (p = 0.026) and β = 0.007 (p = 0.004) respectively. SHAP values indicate that BPS (1.38), bisphenol F (BPF) (0.97), 2,5-dichlorophenol (0.87), TCS (0.78), BP3 (0.60), bisphenol A (BPA) (0.59) and 2,4-dichlorophenol (0.47) in urinary contributed to the model.

Conclusion

The RF model was efficient in identifying a correlation between phenols' exposure and eGFR among United States NHANES 2013-2016 participants. The findings indicate that BPA, BPF, and BPS are inversely associated with eGFR.

MRI radiomics-based interpretable model and nomogram for preoperative prediction of Ki-67 expression status in primary central nervous system lymphoma

Objectives

To investigate the value of interpretable machine learning model and nomogram based on clinical factors, MRI imaging features, and radiomic features to predict Ki-67 expression in primary central nervous system lymphomas (PCNSL).

Materials and methods

MRI images and clinical information of 92 PCNSL patients were retrospectively collected, which were divided into 53 cases in the training set and 39 cases in the external validation set according to different medical centers. A 3D brain tumor segmentation model was trained based on nnU-NetV2, and two prediction models, interpretable Random Forest (RF) incorporating the SHapley Additive exPlanations (SHAP) method and nomogram based on multivariate logistic regression, were proposed for the task of Ki-67 expression status prediction.

Results

The mean dice Similarity Coefficient (DSC) score of the 3D segmentation model on the validation set was 0.85. On the Ki-67 expression prediction task, the AUC of the interpretable RF model on the validation set was 0.84 (95% CI:0.81, 0.86; p < 0.001), which was a 3% improvement compared to the AUC of the nomogram. The Delong test showed that the z statistic for the difference between the two models was 1.901, corresponding to a p value of 0.057. In addition, SHAP analysis showed that the Rad-Score made a significant contribution to the model decision.

Conclusion

In this study, we developed a 3D brain tumor segmentation model and used an interpretable machine learning model and nomogram for preoperative prediction of Ki-67 expression status in PCNSL patients, which improved the prediction of this medical task.

Clinical relevance statement

Ki-67 represents the degree of active cell proliferation and is an important prognostic parameter associated with clinical outcomes. Non-invasive and accurate prediction of Ki-67 expression level preoperatively plays an important role in targeting treatment selection and patient stratification management for PCNSL thereby improving prognosis.

An interactive web application to identify early Parkinsonian non-tremor-dominant subtypes

BACKGROUND

Parkinson's disease (PD) patients with tremor-dominant (TD) and non-tremor-dominant (NTD) subtypes exhibit heterogeneity. Rapid identification of different motor subtypes may help to develop personalized treatment plans.

METHODS

The data were acquired from the Parkinson's Disease Progression Marker Initiative (PPMI). Following the identification of predictors utilizing recursive feature elimination (RFE), seven classical machine learning (ML) models, including logistic regression, support vector machine, decision tree, random forest, extreme gradient boosting, etc., were trained to predict patients' motor subtypes, evaluating the performance of models through the area under the receiver operating characteristic curve (AUC) and validating by the follow-up data.

RESULTS

The feature subset engendered by RFE encompassed 20 features, comprising some clinical assessments and cerebrospinal fluid α-synuclein (CSF α-syn). ML models fitted in the RFE subset performed better in the test and validation sets. The best performing model was support vector machines with the polynomial kernel (P-SVM), achieving an AUC of 0.898. Five-fold repeated cross-validation showed the P-SVM model with CSF α-syn performed better than the model without CSF α-syn (P = 0.034). The Shapley additive explanation plot (SHAP) illustrated that how the levels of each feature affect the predicted probability as NTD subtypes.

CONCLUSION

An interactive web application was developed based on the P-SVM model constructed from feature subset by RFE. It can identify the current motor subtypes of PD patients, making it easier to understand the status of patients and develop personalized treatment plans.

Machine Learning Supported the Modified Gustafson's Criteria for Dental Age Estimation in Southwest China

Adult age estimation is one of the most challenging problems in forensic science and physical anthropology. In this study, we aimed to develop and evaluate machine learning (ML) methods based on the modified Gustafson's criteria for dental age estimation. In this retrospective study, a total of 851 orthopantomograms were collected from patients aged 15 to 40 years old. The secondary dentin formation (SE), periodontal recession (PE), and attrition (AT) of four mandibular premolars were analyzed according to the modified Gustafson's criteria. Ten ML models were generated and compared for age estimation. The partial least squares regressor outperformed other models in males with a mean absolute error (MAE) of 4.151 years. The support vector regressor (MAE = 3.806 years) showed good performance in females. The accuracy of ML models is better than the single-tooth model provided in the previous studies (MAE = 4.747 years in males and MAE = 4.957 years in females). The Shapley additive explanations method was used to reveal the importance of the 12 features in ML models and found that AT and PE are the most influential in age estimation. The findings suggest that the modified Gustafson method can be effectively employed for adult age estimation in the southwest Chinese population. Furthermore, this study highlights the potential of machine learning models to assist experts in achieving accurate and interpretable age estimation.

Artificial intelligence in neuro-oncology: advances and challenges in brain tumor diagnosis, prognosis, and precision treatment

This review delves into the most recent advancements in applying artificial intelligence (AI) within neuro-oncology, specifically emphasizing work on gliomas, a class of brain tumors that represent a significant global health issue. AI has brought transformative innovations to brain tumor management, utilizing imaging, histopathological, and genomic tools for efficient detection, categorization, outcome prediction, and treatment planning. Assessing its influence across all facets of malignant brain tumor management- diagnosis, prognosis, and therapy- AI models outperform human evaluations in terms of accuracy and specificity. Their ability to discern molecular aspects from imaging may reduce reliance on invasive diagnostics and may accelerate the time to molecular diagnoses. The review covers AI techniques, from classical machine learning to deep learning, highlighting current applications and challenges. Promising directions for future research include multimodal data integration, generative AI, large medical language models, precise tumor delineation and characterization, and addressing racial and gender disparities. Adaptive personalized treatment strategies are also emphasized for optimizing clinical outcomes. Ethical, legal, and social implications are discussed, advocating for transparency and fairness in AI integration for neuro-oncology and providing a holistic understanding of its transformative impact on patient care.

Development of Clinical Radiomics-Based Models to Predict Survival Outcome in Pancreatic Ductal Adenocarcinoma: A Multicenter Retrospective Study

PURPOSE

This multicenter retrospective study aims to identify reliable clinical and radiomic features to build machine learning models that predict progression-free survival (PFS) and overall survival (OS) in pancreatic ductal adenocarcinoma (PDAC) patients.

METHODS

Between 2010 and 2020 pre-treatment contrast-enhanced CT scans of 287 pathology-confirmed PDAC patients from two sites of the Hopital Universitaire de Bruxelles (HUB) and from 47 hospitals within the HUB network were retrospectively analysed. Demographic, clinical, and survival data were also collected. Gross tumour volume (GTV) and non-tumoral pancreas (RPV) were semi-manually segmented and radiomics features were extracted. Patients from two HUB sites comprised the training dataset, while those from the remaining 47 hospitals of the HUB network constituted the testing dataset. A three-step method was used for feature selection. Based on the GradientBoostingSurvivalAnalysis classifier, different machine learning models were trained and tested to predict OS and PFS. Model performances were assessed using the C-index and Kaplan-Meier curves. SHAP analysis was applied to allow for post hoc interpretability.

RESULTS

A total of 107 radiomics features were extracted from each of the GTV and RPV. Fourteen subgroups of features were selected: clinical, GTV, RPV, clinical & GTV, clinical & GTV & RPV, GTV-volume and RPV-volume both for OS and PFS. Subsequently, 14 Gradient Boosting Survival Analysis models were trained and tested. In the testing dataset, the clinical & GTV model demonstrated the highest performance for OS (C-index: 0.72) among all other models, while for PFS, the clinical model exhibited a superior performance (C-index: 0.70).

CONCLUSIONS

An integrated approach, combining clinical and radiomics features, excels in predicting OS, whereas clinical features demonstrate strong performance in PFS prediction.

Comparative performance analysis of Boruta, SHAP, and Borutashap for disease diagnosis: A study with multiple machine learning algorithms

Interpretable machine learning models are instrumental in disease diagnosis and clinical decision-making, shedding light on relevant features. Notably, Boruta, SHAP (SHapley Additive exPlanations), and BorutaShap were employed for feature selection, each contributing to the identification of crucial features. These selected features were then utilized to train six machine learning algorithms, including LR, SVM, ETC, AdaBoost, RF, and LR, using diverse medical datasets obtained from public sources after rigorous preprocessing. The performance of each feature selection technique was evaluated across multiple ML models, assessing accuracy, precision, recall, and F1-score metrics. Among these, SHAP showcased superior performance, achieving average accuracies of 80.17%, 85.13%, 90.00%, and 99.55% across diabetes, cardiovascular, statlog, and thyroid disease datasets, respectively. Notably, the LGBM emerged as the most effective algorithm, boasting an average accuracy of 91.00% for most disease states. Moreover, SHAP enhanced the interpretability of the models, providing valuable insights into the underlying mechanisms driving disease diagnosis. This comprehensive study contributes significant insights into feature selection techniques and machine learning algorithms for disease diagnosis, benefiting researchers and practitioners in the medical field. Further exploration of feature selection methods and algorithms holds promise for advancing disease diagnosis methodologies, paving the way for more accurate and interpretable diagnostic models.

Differentiation Between Phyllodes Tumor and Fibroadenoma of the Breast: A Radiomics Prediction Model Based on Full-Field Digital Mammography & Digital Tomosynthesis

OBJECTIVE

To assess the diagnostic performance of FFDM-based and DBT-based radiomics models to differentiate breast phyllodes tumors from fibroadenomas.

METHODS

192 patients (93 phyllodes tumors and 99 fibroadenomas) who underwent mammography were retrospectively enrolled. Radiomic features were respectively extracted from FFDM and the clearest slice of DBT images. A least absolute shrinkage and selection operator (LASSO) regression was used to select radiomics features. A combined model was constructed by radiomics and radiological signatures. Machine learning classification was done using logistic regression based on radiomics or radiological signatures (clinical model). Four radiologists were tested on phyllodes tumors and fibroadenomas with and without optimal model assistance. The area under the receiver operating characteristic (ROC) curve (AUC) was computed to assess the performance of each model or radiologist. The Delong test and McNemar's test were performed to compare the performance.

RESULTS

The combined model yielded the highest performance with an AUC of 0.948 (95%CI: 0.889-1.000) in the testing set, slightly higher than the FFDM-radiomics model (AUC of 0.937, 95%CI: 0.841-0.984) and the DBT-radiomics model (AUC of 0.860, 95%CI: 0.742-0.936) and significantly superior to the clinical model (AUC of 0.719, 95%CI: 0.585-0.829). With the combined model aid, the AUCs of four radiologists were improved from 0.808 to 0.914 (p=0.079), 0.759 to 0.888 (p=0.015), 0.717 to 0.846 (p=0.004), and 0.629 to 0.803 (p=0.001).

CONCLUSION

Radiomics analysis based on FFDM and DBT shows promise in differentiating phyllodes tumors from fibroadenomas.

Prediction of nonsentinel lymph node metastasis in breast cancer patients based on machine learning

BACKGROUND

Develop the best machine learning (ML) model to predict nonsentinel lymph node metastases (NSLNM) in breast cancer patients.

METHODS

From June 2016 to August 2022, 1005 breast cancer patients were included in this retrospective study. Univariate and multivariate analyses were performed using logistic regression. Six ML models were introduced, and their performance was compared.

RESULTS

NSLNM occurred in 338 (33.6%) of 1005 patients. The best ML model was XGBoost, whose average area under the curve (AUC) based on 10-fold cross-verification was 0.722. It performed better than the nomogram, which was based on logistic regression (AUC: 0.764 vs. 0.706).

CONCLUSIONS

The ML model XGBoost can well predict NSLNM in breast cancer patients.

Radiomics and Machine Learning in Brain Tumors and Their Habitat: A Systematic Review

Radiomics is a rapidly evolving field that involves extracting and analysing quantitative features from medical images, such as computed tomography or magnetic resonance images. Radiomics has shown promise in brain tumor diagnosis and patient-prognosis prediction by providing more detailed and objective information about tumors' features than can be obtained from the visual inspection of the images alone. Radiomics data can be analyzed to determine their correlation with a tumor's genetic status and grade, as well as in the assessment of its recurrence vs. therapeutic response, among other features. In consideration of the multi-parametric and high-dimensional space of features extracted by radiomics, machine learning can further improve tumor diagnosis, treatment response, and patients' prognoses. There is a growing recognition that tumors and their microenvironments (habitats) mutually influence each other-tumor cells can alter the microenvironment to increase their growth and survival. At the same time, habitats can also influence the behavior of tumor cells. In this systematic review, we investigate the current limitations and future developments in radiomics and machine learning in analysing brain tumors and their habitats.