Machine-learning with region-level radiomic and dosimetric features for predicting radiotherapy-induced rectal toxicities in prostate cancer patients.

BACKGROUND AND PURPOSE: This study aims to build machine learning models to predict radiation-induced rectal toxicities for three clinical endpoints and explore whether the inclusion of radiomic features calculated on radiotherapy planning computerised tomography (CT) scans combined with dosimetric features can enhance the prediction performance. MATERIALS AND METHODS: 183 patients recruited to the VoxTox study (UK-CRN-ID-13716) were included. Toxicity scores were prospectively collected after 2 years with grade ≥ 1 proctitis, haemorrhage (CTCAEv4.03); and gastrointestinal (GI) toxicity (RTOG) recorded as the endpoints of interest. The rectal wall on each slice was divided into 4 regions according to the centroid, and all slices were divided into 4 sections to calculate region-level radiomic and dosimetric features. The patients were split into a training set (75%, N = 137) and a test set (25%, N = 46). Highly correlated features were removed using four feature selection methods. Individual radiomic or dosimetric or combined (radiomic + dosimetric) features were subsequently classified using three machine learning classifiers to explore their association with these radiation-induced rectal toxicities. RESULTS: The test set area under the curve (AUC) values were 0.549, 0.741 and 0.669 for proctitis, haemorrhage and GI toxicity prediction using radiomic combined with dosimetric features. The AUC value reached 0.747 for the ensembled radiomic-dosimetric model for haemorrhage. CONCLUSIONS: Our preliminary results show that region-level pre-treatment planning CT radiomic features have the potential to predict radiation-induced rectal toxicities for prostate cancer. Moreover, when combined with region-level dosimetric features and using ensemble learning, the model prediction performance slightly improved.

Sub-regional analysis of the parotid glands: model development for predicting late xerostomia with radiomics features in head and neck cancer patients.

BACKGROUND: The irradiation of sub-regions of the parotid has been linked to xerostomia development in patients with head and neck cancer (HNC). In this study, we compared the xerostomia classification performance of radiomics features calculated on clinically relevant and de novo sub-regions of the parotid glands of HNC patients. MATERIAL AND METHODS: All patients (N = 117) were treated with TomoTherapy in 30-35 fractions of 2-2.167 Gy per fraction with daily mega-voltage-CT (MVCT) acquisition for image-guidance purposes. Radiomics features (N = 123) were extracted from daily MVCTs for the whole parotid gland and nine sub-regions. The changes in feature values after each complete week of treatment were considered as predictors of xerostomia (CTCAEv4.03, grade ≥ 2) at 6 and 12 months. Combinations of predictors were generated following the removal of statistically redundant information and stepwise selection. The classification performance of the logistic regression models was evaluated on train and test sets of patients using the Area Under the Curve (AUC) associated with the different sub-regions at each week of treatment and benchmarked with the performance of models solely using dose and toxicity at baseline. RESULTS: In this study, radiomics-based models predicted xerostomia better than standard clinical predictors. Models combining dose to the parotid and xerostomia scores at baseline yielded an AUCtest of 0.63 and 0.61 for xerostomia prediction at 6 and 12 months after radiotherapy while models based on radiomics features extracted from the whole parotid yielded a maximum AUCtest of 0.67 and 0.75, respectively. Overall, across sub-regions, maximum AUCtest was 0.76 and 0.80 for xerostomia prediction at 6 and 12 months. Within the first two weeks of treatment, the cranial part of the parotid systematically yielded the highest AUCtest. CONCLUSION: Our results indicate that variations of radiomics features calculated on sub-regions of the parotid glands can lead to earlier and improved prediction of xerostomia in HNC patients.

Assessing the generalisability of radiomics features previously identified as predictive of radiation-induced sticky saliva and xerostomia.

Background and purpose: While core to the scientific approach, reproducibility of experimental results is challenging in radiomics studies. A recent publication identified radiomics features that are predictive of late irradiation-induced toxicity in head and neck cancer (HNC) patients. In this study, we assessed the generalisability of these findings. Materials and Methods: The procedure described in the publication in question was applied to a cohort of 109 HNC patients treated with 50-70 Gy in 20-35 fractions using helical radiotherapy although there were inherent differences between the two patient populations and methodologies. On each slice of the planning CT with delineated parotid and submandibular glands, the imaging features that were previously identified as predictive of moderate-to-severe xerostomia and sticky saliva 12 months post radiotherapy (Xer12m and SS12m) were calculated. Specifically, Short Run Emphasis (SRE) and maximum CT intensity (maxHU) were evaluated for improvement in prediction of Xer12m and SS12m respectively, compared to models solely using baseline toxicity and mean dose to the salivary glands. Results: None of the associations previously identified as statistically significant and involving radiomics features in univariate or multivariate models could be reproduced on our cohort. Conclusion: The discrepancies observed between the results of the two studies delineate limits to the generalisability of the previously reported findings. This may be explained by the differences in the approaches, in particular the imaging characteristics and subsequent methodological implementation. This highlights the importance of external validation, high quality reporting guidelines and standardisation protocols to ensure generalisability, replication and ultimately clinical implementation.

Gender-related and geographic trends in interactions between radiotherapy professionals on Twitter.

Background and purpose: Twitter presence in academia has been linked to greater research impact which influences career progression. The purpose of this study was to analyse Twitter activity of the radiotherapy community around ESTRO congresses with a focus on gender-related and geographic trends. Materials and methods: Tweets, re-tweets and replies, here designated as interactions, around the ESTRO congresses held in 2012-2021 were collected. Twitter activity was analysed temporally and, for the period 2016-2021, the geographical span of the ESTRO Twitter network was studied. Tweets and Twitter users collated during the 10 years analysed were ranked based on number of 'likes', 're-tweets' and followers, considered as indicators of leadership/influence. Gender representation was assessed for the top-end percentiles. Results: Twitter activity around ESTRO congresses was multiplied by 60 in 6 years growing from 150 interactions in 2012 to a peak of 9097 in 2018. In 2020, during the SARS-CoV-2 pandemic, activity dropped by 60 % to reach 2945 interactions and recovered to half the pre-pandemic level in 2021. Europe, North America and Oceania were strongly connected and remained the main contributors. While overall, 58 % of accounts were owned by men, this proportion increased towards top liked/re-tweeted tweets and most-followed profiles to reach up to 84 % in the top-percentiles. Conclusion: During the SARS-CoV-2 pandemic, Twitter activity around ESTRO congresses substantially decreased. Men were over-represented on the platform and in most popular tweets and influential accounts. Given the increasing importance of social media presence in academia the gender-based biases observed may help in understanding the gender gap in career progression.

Predicting radiotherapy-induced xerostomia in head and neck cancer patients using day-to-day kinetics of radiomics features.

Background and purpose: The images acquired during radiotherapy for image-guidance purposes could be used to monitor patient-specific response to irradiation and improve treatment personalisation. We investigated whether the kinetics of radiomics features from daily mega-voltage CT image-guidance scans (MVCT) improve prediction of moderate-to-severe xerostomia compared to dose/volume parameters in radiotherapy of head-and-neck cancer (HNC). Materials and Methods: All included HNC patients (N = 117) received 30 or more fractions of radiotherapy with daily MVCTs. Radiomics features were calculated on the contra-lateral parotid glands of daily MVCTs. Their variations over time after each complete week of treatment were used to predict moderate-to-severe xerostomia (CTCAEv4.03 grade ≥ 2) at 6, 12 and 24 months post-radiotherapy. After dimensionality reduction, backward/forward selection was used to generate combinations of predictors.Three types of logistic regression model were generated for each follow-up time: 1) a pre-treatment reference model using dose/volume parameters, 2) a combination of dose/volume and radiomics-based predictors, and 3) radiomics-based predictors. The models were internally validated by cross-validation and bootstrapping and their performance evaluated using Area Under the Curve (AUC) on separate training and testing sets. Results: Moderate-to-severe xerostomia was reported by 46 %, 33 % and 26 % of the patients at 6, 12 and 24 months respectively. The selected models using radiomics-based features extracted at or before mid-treatment outperformed the dose-based models with an AUCtrain/AUCtest of 0.70/0.69, 0.76/0.74, 0.86/0.86 at 6, 12 and 24 months, respectively. Conclusion: Our results suggest that radiomics features calculated on MVCTs from the first half of the radiotherapy course improve prediction of moderate-to-severe xerostomia in HNC patients compared to a dose-based pre-treatment model.

Development and Optimization of a Machine-Learning Prediction Model for Acute Desquamation After Breast Radiation Therapy in the Multicenter REQUITE Cohort.

Purpose: Some patients with breast cancer treated by surgery and radiation therapy experience clinically significant toxicity, which may adversely affect cosmesis and quality of life. There is a paucity of validated clinical prediction models for radiation toxicity. We used machine learning (ML) algorithms to develop and optimise a clinical prediction model for acute breast desquamation after whole breast external beam radiation therapy in the prospective multicenter REQUITE cohort study. Methods and Materials: Using demographic and treatment-related features (m = 122) from patients (n = 2058) at 26 centers, we trained 8 ML algorithms with 10-fold cross-validation in a 50:50 random-split data set with class stratification to predict acute breast desquamation. Based on performance in the validation data set, the logistic model tree, random forest, and naïve Bayes models were taken forward to cost-sensitive learning optimisation. Results: One hundred and ninety-two patients experienced acute desquamation. Resampling and cost-sensitive learning optimisation facilitated an improvement in classification performance. Based on maximising sensitivity (true positives), the "hero" model was the cost-sensitive random forest algorithm with a false-negative: false-positive misclassification penalty of 90:1 containing m = 114 predictive features. Model sensitivity and specificity were 0.77 and 0.66, respectively, with an area under the curve of 0.77 in the validation cohort. Conclusions: ML algorithms with resampling and cost-sensitive learning generated clinically valid prediction models for acute desquamation using patient demographic and treatment features. Further external validation and inclusion of genomic markers in ML prediction models are worthwhile, to identify patients at increased risk of toxicity who may benefit from supportive intervention or even a change in treatment plan.

50 years of radiotherapy research: Evolution, trends and lessons for the future.

Rapid and relentless technological advances in an ever-more globalized world have shaped the field of radiation oncology in which we practise today. These developments have drastically modified the habitus1 of health professionals and researchers at an individual and organisational level. In this article we present an analysis of trends in radiation oncology research over the last half a century. To do so, the data from >350,000 scientific publications pertaining to a yearly search of the PubMed database with the keywords cancer radiotherapy was analysed. This analysis revealed that, over the years, radiotherapy research output has declined relative to alternative cancer therapies, representing 64% in 1970 it decreased to 31% in 2019. Also, the pace of research has significantly accelerated with, in the last 15 years, a doubling in the number of articles published by the 10% most productive researchers. Researchers are also facing stronger competition today with a proportion of first authors that will never get to publish as a last author increasing steadily from 58% in 1970 to 84% in 2000. Additionally, radiotherapy research output is extremely unequally distributed in the world, with Africa and South America contributing to ∼3% of radiotherapy articles in 2019 while representing 23% of the world's population. This disparity, reflecting economic situations and radiotherapy capabilities, has a knock-on effect for the provision of routine clinical treatment. Since research activity is inherent to delivery of high quality clinical care, this contributes to the global inequity of radiotherapy services. Learning from these trends is crucial for the future not only of radiation oncology research but also for effective and equitable cancer care.

A data science approach for early-stage prediction of Patient's susceptibility to acute side effects of advanced radiotherapy.

The prediction by classification of side effects incidence in a given medical treatment is a common challenge in medical research. Machine Learning (ML) methods are widely used in the areas of risk prediction and classification. The primary objective of such algorithms is to use several features to predict dichotomous responses (e.g., disease positive/negative). Similar to statistical inference modelling, ML modelling is subject to the class imbalance problem and is affected by the majority class, increasing the false-negative rate. In this study, seventy-nine ML models were built and evaluated to classify approximately 2000 participants from 26 hospitals in eight different countries into two groups of radiotherapy (RT) side effects incidence based on recorded observations from the international study of RT related toxicity "REQUITE". We also examined the effect of sampling techniques and cost-sensitive learning methods on the models when dealing with class imbalance. The combinations of such techniques used had a significant impact on the classification. They resulted in an improvement in incidence status prediction by shifting classifiers' attention to the minority group. The best classification model for RT acute toxicity prediction was identified based on domain experts' success criteria. The Area Under Receiver Operator Characteristic curve of the models tested with an isolated dataset ranged from 0.50 to 0.77. The scale of improved results is promising and will guide further development of models to predict RT acute toxicities. One model was optimised and found to be beneficial to identify patients who are at risk of developing acute RT early-stage toxicities as a result of undergoing breast RT ensuring relevant treatment interventions can be appropriately targeted. The design of the approach presented in this paper resulted in producing a preclinical-valid prediction model. The study was developed by a multi-disciplinary collaboration of data scientists, medical physicists, oncologists and surgeons in the UK Radiotherapy Machine Learning Network.

Associations between voxel-level accumulated dose and rectal toxicity in prostate radiotherapy.

BACKGROUND AND PURPOSE: Associations between dose and rectal toxicity in prostate radiotherapy are generally poorly understood. Evaluating spatial dose distributions to the rectal wall (RW) may lead to improvements in dose-toxicity modelling by incorporating geometric information, masked by dose-volume histograms. Furthermore, predictive power may be strengthened by incorporating the effects of interfraction motion into delivered dose calculations.Here we interrogate 3D dose distributions for patients with and without toxicity to identify rectal subregions at risk (SRR), and compare the discriminatory ability of planned and delivered dose. MATERIAL AND METHODS: Daily delivered dose to the rectum was calculated using image guidance scans, and accumulated at the voxel level using biomechanical finite element modelling. SRRs were statistically determined for rectal bleeding, proctitis, faecal incontinence and stool frequency from a training set (n = 139), and tested on a validation set (n = 47). RESULTS: SRR patterns differed per endpoint. Analysing dose to SRRs improved discriminative ability with respect to the full RW for three of four endpoints. Training set AUC and OR analysis produced stronger toxicity associations from accumulated dose than planned dose. For rectal bleeding in particular, accumulated dose to the SRR (AUC 0.76) improved upon dose-toxicity associations derived from planned dose to the RW (AUC 0.63). However, validation results could not be considered significant. CONCLUSIONS: Voxel-level analysis of dose to the RW revealed SRRs associated with rectal toxicity, suggesting non-homogeneous intra-organ radiosensitivity. Incorporating spatial features of accumulated delivered dose improved dose-toxicity associations. This may be an important tool for adaptive radiotherapy in the future.

Anatomical change during radiotherapy for head and neck cancer, and its effect on delivered dose to the spinal cord.

BACKGROUND AND PURPOSE: The impact of weight loss and anatomical change during head and neck (H&N) radiotherapy on spinal cord dosimetry is poorly understood, limiting evidence-based adaptive management strategies. MATERIALS AND METHODS: 133 H&N patients treated with daily mega-voltage CT image-guidance (MVCT-IG) on TomoTherapy, were selected. Elastix software was used to deform planning scan SC contours to MVCT-IG scans, and accumulate dose. Planned (DP) and delivered (DA) spinal cord D2% (SCD2%) were compared. Univariate relationships between neck irradiation strategy (unilateral vs bilateral), T-stage, N-stage, weight loss, and changes in lateral separation (LND) and CT slice surface area (SSA) at C1 and the superior thyroid notch (TN), and ΔSCD2% [(DA - DP) D2%] were examined. RESULTS: The mean value for (DA - DP) D2% was -0.07 Gy (95%CI -0.28 to 0.14, range -5.7 Gy to 3.8 Gy), and the mean absolute difference between DP and DA (independent of difference direction) was 0.9 Gy (95%CI 0.76-1.04 Gy). Neck treatment strategy (p = 0.39) and T-stage (p = 0.56) did not affect ΔSCD2%. Borderline significance (p = 0.09) was seen for higher N-stage (N2-3) and higher ΔSCD2%. Mean reductions in anatomical metrics were substantial: weight loss 6.8 kg; C1LND 12.9 mm; C1SSA 12.1 cm2; TNLND 5.3 mm; TNSSA 11.2 cm2, but no relationship between weight loss or anatomical change and ΔSCD2% was observed (all r2 < 0.1). CONCLUSIONS: Differences between delivered and planned spinal cord D2% are small in patients treated with daily IG. Even patients experiencing substantial weight loss or anatomical change during treatment do not require adaptive replanning for spinal cord safety.

Applying physical science techniques and CERN technology to an unsolved problem in radiation treatment for cancer: the multidisciplinary 'VoxTox' research programme.

The VoxTox research programme has applied expertise from the physical sciences to the problem of radiotherapy toxicity, bringing together expertise from engineering, mathematics, high energy physics (including the Large Hadron Collider), medical physics and radiation oncology. In our initial cohort of 109 men treated with curative radiotherapy for prostate cancer, daily image guidance computed tomography (CT) scans have been used to calculate delivered dose to the rectum, as distinct from planned dose, using an automated approach. Clinical toxicity data have been collected, allowing us to address the hypothesis that delivered dose provides a better predictor of toxicity than planned dose.