Comparison of deep-learning multimodality data fusion strategies in mandibular osteoradionecrosis NTCP modelling using clinical variables and radiation dose distribution volumes.

OBJECTIVE: Normal tissue complication probability (NTCP) modelling is rapidly embracing deep learning (DL) methods, acknowledging the importance of spatial dose information. Finding effective ways to combine information from radiation dose distribution maps (dosiomics) and clinical data involves technical challenges and requires domain knowledge. We propose different multi-modality data fusion strategies to facilitate future DL-based NTCP studies. Approach. Early, joint and late DL multi-modality fusion strategies were compared using clinical and mandibular radiation dose distribution volumes. These were contrasted with single-modality models: a random forest trained on non-image data (clinical, demographic and dose-volume metrics) and a 3D DenseNet-40 trained on image data (mandibular dose distribution maps). The study involved a matched cohort of 92 ORN cases and 92 controls from a single institution. Main results. The late fusion model exhibited superior discrimination and calibration performance, while the join fusion achieved a more balanced distribution of the predicted probabilities. Discrimination performance did not significantly differ between strategies. Late fusion, though less technically complex, lacks crucial inter-modality interactions for NTCP modelling. In contrast, joint fusion, despite its complexity, resulted in a single network training process which included intra- and inter-modality interactions in its model parameter optimisation. Significance. This study is a pioneering effort in comparing different strategies for including image data into DL-based NTCP models in combination with lower dimensional data such as clinical variables. The discrimination performance of such multi-modality NTCP models and the choice of fusion strategy will depend on the distribution and quality of both types of data. Multiple data fusion strategies should be compared and reported in multi-modality NTCP modelling using DL. .

Xerostomia prediction in patients with nasopharyngeal carcinoma during radiotherapy using segmental dose distribution in dosiomics and radiomics models.

OBJECTIVES: This study aimed to integrate radiomics and dosiomics features to develop a predictive model for xerostomia (XM) in nasopharyngeal carcinoma after radiotherapy. It explores the influence of distinct feature extraction methods and dose ranges on the performance. MATERIALS AND METHODS: Data from 363 patients with nasopharyngeal carcinoma were retrospectively analyzed. We pioneered a dose-segmentation strategy, where the overall dose distribution (OD) was divided into four segmental dose distributions (SDs) at intervals of 15 Gy. Features were extracted using manual definition and deep learning, applying OD or SD and integrating radiomics and dosiomics, yielding corresponding feature scores (manually defined radiomics, MDR; manually defined dosiomics, MDD; deep learning-based radiomics, DLR; deep learning-based dosiomics, DLD). Subsequently, 18 models were developed by combining features and model types (random forest and support vector machine). RESULTS AND CONCLUSION: Under OD, O(DLR_DLD) demonstrated exceptional performance, with an optimal area under the curve (AUC) of 0.81 and an average AUC of 0.71. Within SD, S(DLR_DLD) surpassed the other models, achieving an optimal AUC of 0.90 and an average AUC of 0.85. Therefore, the integration of dosiomics into radiomics can augment predictive efficacy. The dose-segmentation strategy can facilitate the extraction of more profound information. This indicates that ScoreDLR and ScoreMDR were negatively associated with XM, whereas ScoreDLD, derived from SD exceeding 15 Gy, displayed a positive association with XM. For feature extraction, deep learning was superior to manual definition.

Comparative Analysis of Repeatability in CT Radiomics and Dosiomics Features under Image Perturbation: A Study in Cervical Cancer Patients.

This study aims to evaluate the repeatability of radiomics and dosiomics features via image perturbation of patients with cervical cancer. A total of 304 cervical cancer patients with planning CT images and dose maps were retrospectively included. Random translation, rotation, and contour randomization were applied to CT images and dose maps before radiomics feature extraction. The repeatability of radiomics and dosiomics features was assessed using intra-class correlation of coefficient (ICC). Pearson correlation coefficient (r) was adopted to quantify the correlation between the image characteristics and feature repeatability. In general, the repeatability of dosiomics features was lower compared with CT radiomics features, especially after small-sigma Laplacian-of-Gaussian (LoG) and wavelet filtering. More repeatable features (ICC > 0.9) were observed when extracted from the original, Large-sigma LoG filtered, and LLL-/LLH-wavelet filtered images. Positive correlations were found between image entropy and high-repeatable feature number in both CT and dose (r = 0.56, 0.68). Radiomics features showed higher repeatability compared to dosiomics features. These findings highlight the potential of radiomics features for robust quantitative imaging analysis in cervical cancer patients, while suggesting the need for further refinement of dosiomics approaches to enhance their repeatability.

Utilizing radiomics and dosiomics with AI for precision prediction of radiation dermatitis in breast cancer patients.

PURPOSE: This study explores integrating clinical features with radiomic and dosiomic characteristics into AI models to enhance the prediction accuracy of radiation dermatitis (RD) in breast cancer patients undergoing volumetric modulated arc therapy (VMAT). MATERIALS AND METHODS: This study involved a retrospective analysis of 120 breast cancer patients treated with VMAT at Kaohsiung Veterans General Hospital from 2018 to 2023. Patient data included CT images, radiation doses, Dose-Volume Histogram (DVH) data, and clinical information. Using a Treatment Planning System (TPS), we segmented CT images into Regions of Interest (ROIs) to extract radiomic and dosiomic features, focusing on intensity, shape, texture, and dose distribution characteristics. Features significantly associated with the development of RD were identified using ANOVA and LASSO regression (p-value < 0.05). These features were then employed to train and evaluate Logistic Regression (LR) and Random Forest (RF) models, using tenfold cross-validation to ensure robust assessment of model efficacy. RESULTS: In this study, 102 out of 120 VMAT-treated breast cancer patients were included in the detailed analysis. Thirty-two percent of these patients developed Grade 2+ RD. Age and BMI were identified as significant clinical predictors. Through feature selection, we narrowed down the vast pool of radiomic and dosiomic data to 689 features, distributed across 10 feature subsets for model construction. In the LR model, the J subset, comprising DVH, Radiomics, and Dosiomics features, demonstrated the highest predictive performance with an AUC of 0.82. The RF model showed that subset I, which includes clinical, radiomic, and dosiomic features, achieved the best predictive accuracy with an AUC of 0.83. These results emphasize that integrating radiomic and dosiomic features significantly enhances the prediction of Grade 2+ RD. CONCLUSION: Integrating clinical, radiomic, and dosiomic characteristics into AI models significantly improves the prediction of Grade 2+ RD risk in breast cancer patients post-VMAT. The RF model analysis demonstrates that a comprehensive feature set maximizes predictive efficacy, marking a promising step towards utilizing AI in radiation therapy risk assessment and enhancing patient care outcomes.

A dosiomics approach to treatment outcome modeling in carbon ion radiotherapy for skull base chordomas.

PURPOSE: To investigate the role of dosiomics features extracted from physical dose (DPHYS), RBE-weighted dose (DRBE) and dose-averaged Linear Energy Transfer (LETd), to predict the risk of local recurrence (LR) in skull base chordoma (SBC) treated with Carbon Ion Radiotherapy (CIRT). Thus, define and evaluate dosiomics-driven tumor control probability (TCP) models. MATERIALS AND METHODS: 54 SBC patients were retrospectively selected for this study. A regularized Cox proportional hazard model (r-Cox) and Survival Support Vector Machine (s-SVM) were tuned within a repeated Cross Validation (CV) and patients were stratified in low/high risk of LR. Models' performance was evaluated through Harrell's concordance statistic (C-index), and survival was represented through Kaplan-Meier (KM) curves. A multivariable logistic regression was fit to the selected feature sets to generate a dosiomics-driven TCP model for each map. These were compared to a reference model built with clinical parameters in terms of f-score and accuracy. RESULTS: The LETd maps reached a test C-index of 0.750 and 0.786 with r-Cox and s-SVM, and significantly separated KM curves. DPHYS maps and clinical parameters showed promising CV outcomes with C-index above 0.8, despite a poorer performance on the test set and patients stratification. The LETd-based TCP showed a significatively higher f-score (0.67[0.52-0.70], median[IQR]) compared to the clinical model (0.4[0.32-0.63], p < 0.025), while DPHYS achieved a significatively higher accuracy (DPHYS: 0.73[0.65-0.79], Clinical: 0.6 [0.52-0.72]). CONCLUSION: This analysis supports the role of LETd as relevant source of prognostic factors for LR in SBC treated with CIRT. This is reflected in the TCP modeling, where LETd and DPHYS showed an improved performance with respect to clinical models.

Dosiomics-based detection of dose distribution variations in helical tomotherapy for prostate cancer patients: influence of treatment plan parameters.

The stability of dosiomics features (DFs) and dose-volume histogram (DVH) parameters for detecting disparities in helical tomotherapy planned dose distributions was assessed. Treatment plans of 18 prostate patients were recalculated using the followings: field width (WF) (2.5 vs. 5), pitch factor (PF) (0.433 vs. 0.444), and modulation factor (MF) (2.5 vs. 3). From each of the eight plans per patient, ninety-three original and 744 wavelet-based DFs were extracted, using 3D-Slicer software, across six regions including: target volume (PTV), pelvic lymph nodes (PTV-LN), PTV + PTV-LN (PTV-All), one cm rind around PTV-All (PTV-Ring), rectum, and bladder. For the resulting DFs and DVH parameters, the coefficient of variation (CV) was calculated, and using hierarchical clustering, the features were classified into low/high variability. The significance of parameters on instability was analyzed by a three-way analysis of variance. All DF's were stable in PTV, PTV-LN, and PTV-Ring (average CV ( CV ¯ )  ≤ 0.36). Only one feature in the bladder ( CV ¯  = 0.9), rectum ( CV ¯  = 0.4), and PTV-All ( CV ¯  = 0.37) were considered unstable due to change in MF in the bladder and WF in the PTV-All. The value of CV ¯ for the wavelet features was much higher than that for the original features. Out of 225 unstable wavelet features, 84 features had CV ¯  ≥ 1. The CVs for all the DVHs remained very small ( CV ¯ < 0.06). This study highlights that the sensitivity of DFs to changes in tomotherapy planning parameters is influenced by the region and the DFs, particularly wavelet features, surpassing the effectiveness of DVHs.

Combined deep learning and radiomics in pretreatment radiation esophagitis prediction for patients with esophageal cancer underwent volumetric modulated arc therapy.

PURPOSE: To develop a combined radiomics and deep learning (DL) model in predicting radiation esophagitis (RE) of a grade ≥ 2 for patients with esophageal cancer (EC) underwent volumetric modulated arc therapy (VMAT) based on computed tomography (CT) and radiation dose (RD) distribution images. MATERIALS AND METHODS: A total of 273 EC patients underwent VMAT were retrospectively reviewed and enrolled from two centers and divided into training (n = 152), internal validation (n = 66), and external validation (n = 55) cohorts, respectively. Radiomic and dosiomic features along with DL features using convolutional neural networks were extracted and screened from CT and RD images to predict RE. The performance of these models was evaluated and compared using the area under curve (AUC) of the receiver operating characteristic curves (ROC). RESULTS: There were 5 and 10 radiomic and dosiomic features were screened, respectively. XGBoost achieved a best AUC of 0.703, 0.694 and 0.801, 0.729 with radiomic and dosiomic features in the internal and external validation cohorts, respectively. ResNet34 achieved a best prediction AUC of 0.642, 0.657 and 0.762, 0.737 for radiomics based DL model (DLR) and RD based DL model (DLD) in the internal and external validation cohorts, respectively. Combined model of DLD + Dosiomics + clinical factors achieved a best AUC of 0.913, 0.821 and 0.805 in the training, internal, and external validation cohorts, respectively. CONCLUSION: Although the dose was not responsible for the prediction accuracy, the combination of various feature extraction methods was a factor in improving the RE prediction accuracy. Combining DLD with dosiomic features was promising in the pretreatment prediction of RE for EC patients underwent VMAT.

Radiation pneumonitis prediction with dual-radiomics for esophageal cancer underwent radiotherapy.

BACKGROUND: To integrate radiomics and dosiomics features from multiple regions in the radiation pneumonia (RP grade ≥ 2) prediction for esophageal cancer (EC) patients underwent radiotherapy (RT). METHODS: Total of 143 EC patients in the authors' hospital (training and internal validation: 70%:30%) and 32 EC patients from another hospital (external validation) underwent RT from 2015 to 2022 were retrospectively reviewed and analyzed. Patients were dichotomized as positive (RP+) or negative (RP-) according to CTCAE V5.0. Models with radiomics and dosiomics features extracted from single region of interest (ROI), multiple ROIs and combined models were constructed and evaluated. A nomogram integrating radiomics score (Rad_score), dosiomics score (Dos_score), clinical factors, dose-volume histogram (DVH) factors, and mean lung dose (MLD) was also constructed and validated. RESULTS: Models with Rad_score_Lung&Overlap and Dos_score_Lung&Overlap achieved a better area under curve (AUC) of 0.818 and 0.844 in the external validation in comparison with radiomics and dosiomics models with features extracted from single ROI. Combining four radiomics and dosiomics models using support vector machine (SVM) improved the AUC to 0.854 in the external validation. Nomogram integrating Rad_score, and Dos_score with clinical factors, DVH factors, and MLD further improved the RP prediction AUC to 0.937 and 0.912 in the internal and external validation, respectively. CONCLUSION: CT-based RP prediction model integrating radiomics and dosiomics features from multiple ROIs outperformed those with features from a single ROI with increased reliability for EC patients who underwent RT.

A nomogram based on pretreatment radiomics and dosiomics features for predicting overall survival associated with esophageal squamous cell cancer.

OBJECTIVES: To propose a nomogram-based survival prediction model for esophageal squamous cell carcinoma (ESCC) treated with definitive chemoradiotherapy using pretreatment computed tomography (CT), positron emission tomography (PET) radiomics and dosiomics features, and common clinical factors. METHODS: Radiomics and dosiomics features were extracted from CT and PET images and dose distribution from 2 institutions. The least absolute shrinkage and selection operator (LASSO) with logistic regression was used to select radiomics and dosiomics features by calculating the radiomics and dosiomics scores (Rad-score and Dos-score), respectively, in the training model. The model was trained in 81 patients and validated in 35 patients at Center 1 using 10-fold cross validation. The model was externally tested in 26 patients at Center 2. The predictive clinical factors, Rad-score, and Dos-score were identified to develop a nomogram model. RESULTS: Using LASSO Cox regression, 13, 11, and 19 CT, PET-based radiomics, and dosiomics features, respectively, were selected. The clinical factors T-stage, N-stage, and clinical stage were selected as significant prognostic factors by univariate Cox regression. In the external validation cohort, the C-index of the combined model of CT-based radiomics, PET-based radiomics, and dosiomics features with clinical factors were 0.74, 0.82, and 0.92, respectively. Significant differences in overall survival (OS) in the combined model of CT-based radiomics, PET-based radiomics, and dosiomics features with clinical factors were observed between the high- and low-risk groups (P = 0.019, 0.038, and 0.014, respectively). CONCLUSION: The dosiomics features have a better predicter for OS than CT- and PET-based radiomics features in ESCC treated with radiotherapy. CLINICAL RELEVANCE STATEMENT: The current study predicted the overall survival for esophageal squamous cell carcinoma patients treated with definitive chemoradiotherapy. The dosiomics features have a better predicter for overall survival than CT- and PET-based radiomics features.

Multi-omics deep learning for radiation pneumonitis prediction in lung cancer patients underwent volumetric modulated arc therapy.

BACKGROUND AND OBJECTIVE: To evaluate the feasibility and accuracy of radiomics, dosiomics, and deep learning (DL) in predicting Radiation Pneumonitis (RP) in lung cancer patients underwent volumetric modulated arc therapy (VMAT) to improve radiotherapy safety and management. METHODS: Total of 318 and 31 lung cancer patients underwent VMAT from First Affiliated Hospital of Wenzhou Medical University (WMU) and Quzhou Affiliated Hospital of WMU were enrolled for training and external validation, respectively. Models based on radiomics (R), dosiomics (D), and combined radiomics and dosiomics features (R+D) were constructed and validated using three machine learning (ML) methods. DL models trained with CT (DLR), dose distribution (DLD), and combined CT and dose distribution (DL(R+D)) images were constructed. DL features were then extracted from the fully connected layers of the best-performing DL model to combine with features of the ML model with the best performance to construct models of R+DLR, D+DLD, R+D+DL(R+D)) for RP prediction. RESULTS: The R+D model achieved a best area under curve (AUC) of 0.84, 0.73, and 0.73 in the internal validation cohorts with Support Vector Machine (SVM), XGBoost, and Logistic Regression (LR), respectively. The DL(R+D) model achieved a best AUC of 0.89 and 0.86 using ResNet-34 in training and internal validation cohorts, respectively. The R+D+DL(R+D) model achieved a best performance in the external validation cohorts with an AUC, accuracy, sensitivity, and specificity of 0.81(0.62-0.99), 0.81, 0.84, and 0.67, respectively. CONCLUSIONS: The integration of radiomics, dosiomics, and DL features is feasible and accurate for the RP prediction to improve the management of lung cancer patients underwent VMAT.

Radiomics-based hybrid model for predicting radiation pneumonitis: A systematic review and meta-analysis.

PURPOSE: This study reviewed and meta-analyzed evidence on radiomics-based hybrid models for predicting radiation pneumonitis (RP). These models are crucial for improving thoracic radiotherapy plans and mitigating RP, a common complication of thoracic radiotherapy. We examined and compared the RP prediction models developed in these studies with the radiomics features employed in RP models. METHODS: We systematically searched Google Scholar, Embase, PubMed, and MEDLINE for studies published up to April 19, 2024. Sixteen studies met the inclusion criteria. We compared the RP prediction models developed in these studies and the radiomics features employed. RESULTS: Radiomics, as a single-factor evaluation, achieved an area under the receiver operating characteristic curve (AUROC) of 0.73, accuracy of 0.69, sensitivity of 0.64, and specificity of 0.74. Dosiomics achieved an AUROC of 0.70. Clinical and dosimetric factors showed lower performance, with AUROCs of 0.59 and 0.58. Combining clinical and radiomic factors yielded an AUROC of 0.78, while combining dosiomic and radiomics factors produced an AUROC of 0.81. Triple combinations, including clinical, dosimetric, and radiomics factors, achieved an AUROC of 0.81. The study identifies key radiomics features, such as the Gray Level Co-occurrence Matrix (GLCM) and Gray Level Size Zone Matrix (GLSZM), which enhance the predictive accuracy of RP models. CONCLUSIONS: Radiomics-based hybrid models are highly effective in predicting RP. These models, combining traditional predictive factors with radiomic features, particularly GLCM and GLSZM, offer a clinically feasible approach for identifying patients at higher RP risk. This approach enhances clinical outcomes and improves patient quality of life. PROTOCOL REGISTRATION: The protocol of this study was registered on PROSPERO (CRD42023426565).

A dosiomics model for prediction of radiation-induced acute skin toxicity in breast cancer patients: machine learning-based study for a closed bore linac.

BACKGROUND: Radiation induced acute skin toxicity (AST) is considered as a common side effect of breast radiation therapy. The goal of this study was to design dosiomics-based machine learning (ML) models for prediction of AST, to enable creating optimized treatment plans for high-risk individuals. METHODS: Dosiomics features extracted using Pyradiomics tool (v3.0.1), along with treatment plan-derived dose volume histograms (DVHs), and patient-specific treatment-related (PTR) data of breast cancer patients were used for modeling. Clinical scoring was done using the Common Terminology Criteria for Adverse Events (CTCAE) V4.0 criteria for skin-specific symptoms. The 52 breast cancer patients were grouped into AST 2 + (CTCAE ≥ 2) and AST 2 - (CTCAE < 2) toxicity grades to facilitate AST modeling. They were randomly divided into training (70%) and testing (30%) cohorts. Multiple prediction models were assessed through multivariate analysis, incorporating different combinations of feature groups (dosiomics, DVH, and PTR) individually and collectively. In total, seven unique combinations, along with seven classification algorithms, were considered after feature selection. The performance of each model was evaluated on the test group using the area under the receiver operating characteristic curve (AUC) and f1-score. Accuracy, precision, and recall of each model were also studied. Statistical analysis involved features differences between AST 2 - and AST 2 + groups and cutoff value calculations. RESULTS: Results showed that 44% of the patients developed AST 2 + after Tomotherapy. The dosiomics (DOS) model, developed using dosiomics features, exhibited a noteworthy improvement in AUC (up to 0.78), when spatial information is preserved in the dose distribution, compared to DVH features (up to 0.71). Furthermore, a baseline ML model created using only PTR features for comparison with DOS models showed the significance of dosiomics in early AST prediction. By employing the Extra Tree (ET) classifiers, the DOS + DVH + PTR model achieved a statistically significant improved performance in terms of AUC (0.83; 95% CI 0.71-0.90), accuracy (0.70), precision (0.74) and sensitivity (0.72) compared to other models. CONCLUSIONS: This study confirmed the benefit of dosiomics-based ML in the prediction of AST. However, the combination of dosiomics, DVH, and PTR yields significant improvement in AST prediction. The results of this study provide the opportunity for timely interventions to prevent the occurrence of radiation induced AST.

A machine learning approach for predicting radiation-induced hypothyroidism in patients with nasopharyngeal carcinoma undergoing tomotherapy.

The purpose of this study was to establish an integrated predictive model that combines clinical features, DVH, radiomics, and dosiomics features to predict RIHT in patients receiving tomotherapy for nasopharyngeal carcinoma. Data from 219 patients with nasopharyngeal carcinoma were randomly divided into a training cohort (n = 175) and a test cohort (n = 44) in an 8:2 ratio. RIHT is defined as serum thyroid-stimulating hormone (TSH) greater than 5.6 µU/mL, with or without a decrease in free thyroxine (FT4). Clinical features, 27 DVH features, 107 radiomics features and 107 dosiomics features were extracted for each case and included in the model construction. The least absolute shrinkage and selection operator (LASSO) regression method was used to select the most relevant features. The eXtreme Gradient Boosting (XGBoost) was then employed to train separate models using the selected features from clinical, DVH, radiomics and dosiomics data. Finally, a combined model incorporating all features was developed. The models were evaluated using receiver operating characteristic (ROC) curves and decision curve analysis. In the test cohort, the area under the receiver operating characteristic curve (AUC) for the clinical, DVH, radiomics, dosiomics and combined models were 0.798 (95% confidence interval [CI], 0.656-0.941), 0.673 (0.512-0.834), 0.714 (0.555-0.873), 0.698 (0.530-0.848) and 0.842 (0.724-0.960), respectively. The combined model exhibited higher AUC values compared to other models. The decision curve analysis demonstrated that the combined model had superior clinical utility within the threshold probability range of 1% to 79% when compared to the other models. This study has successfully developed a predictive model that combines multiple features. The performance of the combined model is superior to that of single-feature models, allowing for early prediction of RIHT in patients with nasopharyngeal carcinoma after tomotherapy.

A deep learning-based method for the prediction of temporal lobe injury in patients with nasopharyngeal carcinoma.

PURPOSE: To establish a deep learning-based model to predict radiotherapy-induced temporal lobe injury (TLI). MATERIALS AND METHODS: Spatial features of dose distribution within the temporal lobe were extracted using both the three-dimensional convolution (C3D) network and the dosiomics method. The Minimal Redundancy-Maximal-Relevance (mRMR) method was employed to rank the extracted features and select the most relevant ones. Four machine learning (ML) classifiers, including logistic regression (LR), k-nearest neighbors (kNN), support vector machines (SVM) and random forest (RF), were used to establish prediction models. Nested sampling and hyperparameter tuning methods were applied to train and validate the prediction models. For comparison, a prediction model base on the conventional D0.5cc of the temporal lobe obtained from dose volume (DV) histogram was established. The area under the receiver operating characteristic (ROC) curve (AUC) was utilized to compare the predictive performance of the different models. RESULTS: A total of 127 nasopharyngeal carcinoma (NPC) patients were included in the study. In the model based on C3D deep learning features, the highest AUC value of 0.843 was achieved with 5 features. For the dosiomics features model, the highest AUC value of 0.715 was attained with 1 feature. Both of these models demonstrated superior performance compared to the prediction model based on DV parameters, which yielded an AUC of 0.695. CONCLUSION: The prediction model utilizing C3D deep learning features outperformed models based on dosiomics features or traditional parameters in predicting the onset of TLI. This approach holds promise for predicting radiation-induced toxicities and guide individualized radiotherapy.

Radiomic- and dosiomic-based clustering development for radio-induced neurotoxicity in pediatric medulloblastoma.

BACKGROUND: Texture analysis extracts many quantitative image features, offering a valuable, cost-effective, and non-invasive approach for individual medicine. Furthermore, multimodal machine learning could have a large impact for precision medicine, as texture biomarkers can underlie tissue microstructure. This study aims to investigate imaging-based biomarkers of radio-induced neurotoxicity in pediatric patients with metastatic medulloblastoma, using radiomic and dosiomic analysis. METHODS: This single-center study retrospectively enrolled children diagnosed with metastatic medulloblastoma (MB) and treated with hyperfractionated craniospinal irradiation (CSI). Histological confirmation of medulloblastoma and baseline follow-up magnetic resonance imaging (MRI) were mandatory. Treatment involved helical tomotherapy (HT) delivering a dose of 39 Gray (Gy) to brain and spinal axis and a posterior fossa boost up to 60 Gy. Clinical outcomes, such as local and distant brain control and neurotoxicity, were recorded. Radiomic and dosiomic features were extracted from tumor regions on T1, T2, FLAIR (fluid-attenuated inversion recovery) MRI-maps, and radiotherapy dose distribution. Different machine learning feature selection and reduction approaches were performed for supervised and unsupervised clustering. RESULTS: Forty-eight metastatic medulloblastoma patients (29 males and 19 females) with a mean age of 12 ± 6 years were enrolled. For each patient, 332 features were extracted. Greater level of abstraction of input data by combining selection of most performing features and dimensionality reduction returns the best performance. The resulting one-component radiomic signature yielded an accuracy of 0.73 with sensitivity, specificity, and precision of 0.83, 0.64, and 0.68, respectively. CONCLUSIONS: Machine learning radiomic-dosiomic approach effectively stratified pediatric medulloblastoma patients who experienced radio-induced neurotoxicity. Strategy needs further validation in external dataset for its potential clinical use in ab initio management paradigms of medulloblastoma.

Synergizing the interaction of single nucleotide polymorphisms with dosiomics features to build a dual-omics model for the prediction of radiation pneumonitis.

OBJECTIVE: Radiation pneumonitis (RP) is the major dose-limiting toxicity of thoracic radiotherapy. This study aimed to developed a dual-omics (single nucleotide polymorphisms, SNP and dosiomics) prediction model for symptomatic RP. MATERIALS AND METHODS: The potential SNPs, which are of significant difference between the RP grade ≥ 3 group and the RP grade ≤ 1 group, were selected from the whole exome sequencing SNPs using the Fisher's exact test. Patients with lung cancer who received thoracic radiotherapy at our institution from 2009 to 2016 were enrolled for SNP selection and model construction. The factorization machine (FM) method was used to model the SNP epistasis effect, and to construct the RP prediction model (SNP-FM). The dosiomics features were extracted, and further selected using the minimum redundancy maximum relevance (mRMR) method. The selected dosiomics features were added to the SNP-FM model to construct the dual-omics model. RESULTS: For SNP screening, peripheral blood samples of 28 patients with RP grade ≥ 3 and the matched 28 patients with RP grade ≤ 1 were sequenced. 81 SNPs were of significant difference (P < 0.015) and considered as potential SNPs. In addition, 21 radiation toxicity related SNPs were also included. For model construction, 400 eligible patients (including 108 RP grade ≥ 2) were enrolled. Single SNP showed no strong correlation with RP. On the other hand, the SNP-SNP interaction (epistasis effect) of 19 SNPs were modeled by the FM method, and achieved an area under the curve (AUC) of 0.76 in the testing group. In addition, 4 dosiomics features were selected and added to the model, and increased the AUC to 0.81. CONCLUSIONS: A novel dual-omics model by synergizing the SNP epistasis effect with dosiomics features was developed. The enhanced the RP prediction suggested its promising clinical utility in identifying the patients with severe RP during thoracic radiotherapy.

Dosiomics and radiomics-based prediction of pneumonitis after radiotherapy and immune checkpoint inhibition: The relevance of fractionation.

OBJECTIVES: Post-therapy pneumonitis (PTP) is a relevant side effect of thoracic radiotherapy and immunotherapy with checkpoint inhibitors (ICI). The influence of the combination of both, including dose fractionation schemes on PTP development is still unclear. This study aims to improve the PTP risk estimation after radio(chemo)therapy (R(C)T) for lung cancer with and without ICI by investigation of the impact of dose fractionation on machine learning (ML)-based prediction. MATERIALS AND METHODS: Data from 100 patients who received fractionated R(C)T were collected. 39 patients received additional ICI therapy. Computed Tomography (CT), RT segmentation and dose data were extracted and physical doses were converted to 2-Gy equivalent doses (EQD2) to account for different fractionation schemes. Features were reduced using Pearson intercorrelation and the Boruta algorithm within 1000-fold bootstrapping. Six single (clinics, Dose Volume Histogram (DVH), ICI, chemotherapy, radiomics, dosiomics) and four combined models (radiomics + dosiomics, radiomics + DVH + Clinics, dosiomics + DVH + Clinics, radiomics + dosiomics + DVH + Clinics) were trained to predict PTP. Dose-based models were tested using physical dose and EQD2. Four ML-algorithms (random forest (rf), logistic elastic net regression, support vector machine, logitBoost) were trained and tested using 5-fold nested cross validation and Synthetic Minority Oversampling Technique (SMOTE) for resampling in R. Prediction was evaluated using the area under the receiver operating characteristic curve (AUC) on the test sets of the outer folds. RESULTS: The combined model of all features using EQD2 surpassed all other models (AUC = 0.77, Confidence Interval CI 0.76-0.78). DVH, clinical data and ICI therapy had minor impact on PTP prediction with AUC values between 0.42 and 0.57. All EQD2-based models outperformed models based on physical dose. CONCLUSIONS: Radiomics + dosiomics based ML models combined with clinical and dosimetric models were found to be suited best for PTP prediction after R(C)T and could improve pre-treatment decision making. Different RT dose fractionation schemes should be considered for dose-based ML approaches.

Development and validation of survival prognostic models for head and neck cancer patients using machine learning and dosiomics and CT radiomics features: a multicentric study.

BACKGROUND: This study aimed to investigate the value of clinical, radiomic features extracted from gross tumor volumes (GTVs) delineated on CT images, dose distributions (Dosiomics), and fusion of CT and dose distributions to predict outcomes in head and neck cancer (HNC) patients. METHODS: A cohort of 240 HNC patients from five different centers was obtained from The Cancer Imaging Archive. Seven strategies, including four non-fusion (Clinical, CT, Dose, DualCT-Dose), and three fusion algorithms (latent low-rank representation referred (LLRR),Wavelet, weighted least square (WLS)) were applied. The fusion algorithms were used to fuse the pre-treatment CT images and 3-dimensional dose maps. Overall, 215 radiomics and Dosiomics features were extracted from the GTVs, alongside with seven clinical features incorporated. Five feature selection (FS) methods in combination with six machine learning (ML) models were implemented. The performance of the models was quantified using the concordance index (CI) in one-center-leave-out 5-fold cross-validation for overall survival (OS) prediction considering the time-to-event. RESULTS: The mean CI and Kaplan-Meier curves were used for further comparisons. The CoxBoost ML model using the Minimal Depth (MD) FS method and the glmnet model using the Variable hunting (VH) FS method showed the best performance with CI = 0.73 ± 0.15 for features extracted from LLRR fused images. In addition, both glmnet-Cindex and Coxph-Cindex classifiers achieved a CI of 0.72 ± 0.14 by employing the dose images (+ incorporated clinical features) only. CONCLUSION: Our results demonstrated that clinical features, Dosiomics and fusion of dose and CT images by specific ML-FS models could predict the overall survival of HNC patients with acceptable accuracy. Besides, the performance of ML methods among the three different strategies was almost comparable.

Multi-omics and Multi-VOIs to predict esophageal fistula in esophageal cancer patients treated with radiotherapy.

OBJECTIVE: This study aimed to develop a prediction model for esophageal fistula (EF) in esophageal cancer (EC) patients treated with intensity-modulated radiation therapy (IMRT), by integrating multi-omics features from multiple volumes of interest (VOIs). METHODS: We retrospectively analyzed pretreatment planning computed tomographic (CT) images, three-dimensional dose distributions, and clinical factors of 287 EC patients. Nine groups of features from different combination of omics [Radiomics (R), Dosiomics (D), and RD (the combination of R and D)], and VOIs [esophagus (ESO), gross tumor volume (GTV), and EG (the combination of ESO and GTV)] were extracted and separately selected by unsupervised (analysis of variance (ANOVA) and Pearson correlation test) and supervised (Student T test) approaches. The final model performance was evaluated using five metrics: average area under the receiver-operator-characteristics curve (AUC), accuracy, precision, recall, and F1 score. RESULTS: For multi-omics using RD features, the model performance in EG model shows: AUC, 0.817 ± 0.031; 95% CI 0.805, 0.825; p < 0.001, which is better than single VOI (ESO or GTV). CONCLUSION: Integrating multi-omics features from multi-VOIs enables better prediction of EF in EC patients treated with IMRT. The incorporation of dosiomics features can enhance the model performance of the prediction.

Dosiomics for intensity-modulated radiotherapy in patients with prostate cancer: survival analysis stratified by baseline prostate-specific antigen and Gleason grade group in a 2-institutional retrospective study.

OBJECTIVE: This study evaluated the prognostic impact of the quality of dose distribution using dosiomics in patients with prostate cancer, stratified by pretreatment prostate-specific antigen (PSA) levels and Gleason grade (GG) group. METHODS: A total of 721 patients (Japanese Foundation for Cancer Research [JFCR] cohort: N = 489 and Tokyo Radiation Oncology Clinic [TROC] cohort: N = 232) with localized prostate cancer treated by intensity-modulated radiation therapy were enrolled. Two predictive dosiomic features for biochemical recurrence (BCR) were selected and patients were divided into certain groups stratified by pretreatment PSA levels and GG. Freedom from biochemical failure (FFBF) was estimated using the Kaplan-Meier method based on each dosiomic feature and univariate discrimination was evaluated using the log-rank test. As an exploratory analysis, a dosiomics hazard (DH) score was developed and its prognostic power for BCR was examined. RESULTS: The dosiomic feature extracted from planning target volume (PTV) significantly distinguished the high- and low-risk groups in patients with PSA levels >10 ng/mL (7-year FFBF: 86.7% vs 76.1%, P < .01), GG 4 (92.2% vs 76.9%, P < .01), and GG 5 (83.1% vs 77.8%, P = .04). The DH score showed significant association with BCR (hazard score: 2.04; 95% confidence interval: 1.38-3.01; P < .001). CONCLUSION: The quality of planned dose distribution on PTV may affect the prognosis of patients with poor prognostic factors, such as PSA levels >10 ng/mL and higher GGs. ADVANCES IN KNOWLEDGE: The effects of planned dose distribution on prognosis differ depending on the patient's clinical background.