Machine learning and modeling: Data, validation, communication challenges

Advanced prediction of multi-leaf collimator leaf position using artificial neural network

BACKGROUND

Multi-leaf collimators (MLCs) are crucial for modern radiotherapy as they ensure precise target irradiation through accurate leaf positioning. Accurate prediction of MLC leaf positions is vital for the effectiveness and safety of treatments.

PURPOSE

This study aims to establish three neural network models for predicting the delivered positions of MLCs in radiotherapy.

METHODS

Fifty plans with sliding window dynamic intensity-modulated radiation therapy delivery were selected from an Elekta linear accelerator, which features a 160-leaf MLC system. The dose fraction, gantry angle, collimator angle, X1 and X2 jaw positions, Y1 and Y2 carriage positions, planned leaf positions, adjacent leaf positions, leaf gap, leaf velocity, and leaf acceleration were extracted from the planning data in the machine's log files and used as model inputs, with the delivered leaf positional serving as the target response. This establishes the input-output relationship for the neural network, and the predicted MLC positions are obtained through training. Particle Swarm Optimization Back Propagation Neural Network (PSOBPNN), Back Propagation Neural Network (BPNN), and Radial Basis Function Neural Network (RBFNN) architectures were developed to predict MLC leaf positional deviations during treatment. The training was conducted on 70% of the sample data, with the remaining 30% used for validation and testing. Model performance was assessed using metrics such as mean absolute error (MAE), mean squared error (MSE), regression plots, and error histograms.

RESULTS

The proposed neural network models demonstrated high accuracy in predicting MLC leaf positions. The PSOBPNN model demonstrated superior performance with an MAE of 0.0043 mm and an MSE of 0.00003 mm2. In comparison, the BPNN model achieved an MAE of 0.0241 mm and an MSE of 0.001 mm2, while the RBFNN model exhibited an MAE of 0.0331 mm and an MSE of 0.0019 mm2. The correlation coefficient (R = 0.9999) of models indicates a close match between predicted and delivered leaf positions for all MLC leaves.

CONCLUSION

Three models were evaluated for predicting the delivered MLC positions using data from an Elekta accelerator. The PSOBPNN model exhibited superior performance by achieving markedly lower MAE and MSE values while also demonstrating robust generalizability in predicting positions across various leaf indices, outperforming the conventional BPNN and RBFNN models.

Advancing Privacy-Preserving Health Care Analytics and Implementation of the Personal Health Train: Federated Deep Learning Study

BACKGROUND

The rapid advancement of deep learning in health care presents significant opportunities for automating complex medical tasks and improving clinical workflows. However, widespread adoption is impeded by data privacy concerns and the necessity for large, diverse datasets across multiple institutions. Federated learning (FL) has emerged as a viable solution, enabling collaborative artificial intelligence model development without sharing individual patient data. To effectively implement FL in health care, robust and secure infrastructures are essential. Developing such federated deep learning frameworks is crucial to harnessing the full potential of artificial intelligence while ensuring patient data privacy and regulatory compliance.

OBJECTIVE

The objective is to introduce an innovative FL infrastructure called the Personal Health Train (PHT) that includes the procedural, technical, and governance components needed to implement FL on real-world health care data, including training deep learning neural networks. The study aims to apply this federated deep learning infrastructure to the use case of gross tumor volume segmentation on chest computed tomography images of patients with lung cancer and present the results from a proof-of-concept experiment.

METHODS

The PHT framework addresses the challenges of data privacy when sharing data, by keeping data close to the source and instead bringing the analysis to the data. Technologically, PHT requires 3 interdependent components: "tracks" (protected communication channels), "trains" (containerized software apps), and "stations" (institutional data repositories), which are supported by the open source "Vantage6" software. The study applies this federated deep learning infrastructure to the use case of gross tumor volume segmentation on chest computed tomography images of patients with lung cancer, with the introduction of an additional component called the secure aggregation server, where the model averaging is done in a trusted and inaccessible environment.

RESULTS

We demonstrated the feasibility of executing deep learning algorithms in a federated manner using PHT and presented the results from a proof-of-concept study. The infrastructure linked 12 hospitals across 8 nations, covering 4 continents, demonstrating the scalability and global reach of the proposed approach. During the execution and training of the deep learning algorithm, no data were shared outside the hospital.

CONCLUSIONS

The findings of the proof-of-concept study, as well as the implications and limitations of the infrastructure and the results, are discussed. The application of federated deep learning to unstructured medical imaging data, facilitated by the PHT framework and Vantage6 platform, represents a significant advancement in the field. The proposed infrastructure addresses the challenges of data privacy and enables collaborative model development, paving the way for the widespread adoption of deep learning-based tools in the medical domain and beyond. The introduction of the secure aggregation server implied that data leakage problems in FL can be prevented by careful design decisions of the infrastructure.

TRIAL REGISTRATION

ClinicalTrials.gov NCT05775068; https://clinicaltrials.gov/study/NCT05775068.

A bibliometrics analysis based on the application of artificial intelligence in the field of radiotherapy from 2003 to 2023

BACKGROUND

Recent research has demonstrated that the use of artificial intelligence (AI) in radiotherapy (RT) has significantly streamlined the process for physicians to treat patients with tumors; however, bibliometric studies examining the correlation between AI and RT are not available. Providing a thorough overview of the knowledge structure and research hotspots between AI and RT was the main goal of the current study.

METHOD

A search was conducted on the Web of Science Core Collection (WoSCC) database for publications pertaining to AI and RT between 2003 and 2023. VOSviewers, CiteSpace, and the R program "bibliometrix" were used to do the bibliometric analysis.

RESULTS

The analysis comprised 615 publications from 64 countries, with USA and China leading the pack. Since 2017, there have been more and more publications about RT and AI every year. The research center that made the biggest contribution to this topic was Maastricht University. The most articles published journal in this field was Frontiers in Oncology, while Medical Physics received the greatest number of citations. Dekker Andre is the author with the greatest number of published articles, while Philippe Lambin was the most often co-cited author. In the newly identified research hotspots, "autocontouring algorithm", "deep learning", and "machine learning" stand out as the main terms.

CONCLUSION

In fact, our bibliometric analysis offers insightful information on current research directions and advancements pertaining to the use of AI in RT. For academics looking to understand the connection between AI and RT, this study is a great resource because it highlights current research frontiers and hot trends.

Prospective validation of a machine learning model for applicator and hybrid interstitial needle selection in high-dose-rate (HDR) cervical brachytherapy

PURPOSE

To Demonstrate the clinical validation of a machine learning (ML) model for applicator and interstitial needle prediction in gynecologic brachytherapy through a prospective clinical study in a single institution.

METHODS

The study included cervical cancer patients receiving high-dose-rate brachytherapy using intracavitary (IC) or hybrid interstitial (IC/IS) applicators. For each patient, the primary radiation oncologist contoured the high-risk clinical target volume on a pre-brachytherapy MRI, indicated the approximate applicator location, and made a clinical determination of the first fraction applicator. A pre-trained ML model predicted the applicator and IC/IS needle arrangement using tumor geometry. Following the first fraction, ML and radiation oncologist predictions were compared and a replanning study determined the applicator providing optimal organ-at-risk (OAR) dosimetry. The ML-predicted applicator and needle arrangement and the clinical determination were compared to this dosimetric ground truth.

RESULTS

Ten patients were accrued from December 2020 to October 2022. Compared to the dosimetrically optimal applicator, both the radiation oncologist and ML had an accuracy of 70%. ML demonstrated better identification of patients requiring IC/IS applicators and provided balanced IC and IC/IS predictions. The needle selection model achieved an average accuracy of 82.5%. ML-predicted needle arrangements matched or improved plan quality when compared to clinically selected arrangements. Overall, ML predictions led to an average total improvement of 2.0 Gy to OAR doses over three treatment fractions when compared to clinical predictions.

CONCLUSION

In the context of a single institution study, the presented ML model demonstrates valuable decision-support for the applicator and needle selection process with the potential to provide improved dosimetry. Future work will include a multi-center study to assess generalizability.

Efficacy of stereotactic body radiotherapy and response prediction using artificial intelligence in oligometastatic gynaecologic cancer

PURPOSE

We present a large real-world multicentric dataset of ovarian, uterine and cervical oligometastatic lesions treated with SBRT exploring efficacy and clinical outcomes. In addition, an exploratory machine learning analysis was performed.

METHODS

A pooled analysis of gynecological oligometastases in terms of efficacy and clinical outcomes as well an exploratory machine learning model to predict the CR to SBRT were carried out. The CR rate following radiotherapy (RT) was the study main endpoint. The secondary endpoints included the 2-year actuarial LC, DMFS, PFS, and OS.

RESULTS

501 patients from 21 radiation oncology institutions with 846 gynecological metastases were analyzed, mainly ovarian (53.1%) and uterine metastases(32.1%).Multiple fraction radiotherapy was used in 762 metastases(90.1%).The most frequent schedule was 24 Gy in 3 fractions(13.4%). CR was observed in 538(63.7%) lesions. The Machine learning analysis showed a poor ability to find covariates strong enough to predict CR in the whole series. Analyzing them separately, in uterine cancer, if RT dose≥78.3Gy, the CR probability was 75.4%; if volume was <13.7 cc, the CR probability became 85.1%. In ovarian cancer, if the lesion was a lymph node, the CR probability was 71.4%; if volume was <17 cc, the CR probability rose to 78.4%. No covariate predicted the CR for cervical lesions. The overall 2-year actuarial LC was 79.2%, however it was 91.5% for CR and 52.5% for not CR lesions(p < 0.001). The overall 2-year DMFS, PFS and OS rate were 27.3%, 24.8% and 71.0%, with significant differences between CR and not CR.

CONCLUSIONS

CR was substantially associated to patient outcomes in our series of gynecological cancer oligometastatic lesions. The ability to predict a CR through artificial intelligence could also drive treatment choices in the context of personalized oncology.

Motion compensated cone-beam CT reconstruction using ana priorimotion model from CT simulation: a pilot study

Objective. To combat the motion artifacts present in traditional 4D-CBCT reconstruction, an iterative technique known as the motion-compensated simultaneous algebraic reconstruction technique (MC-SART) was previously developed. MC-SART employs a 4D-CBCT reconstruction to obtain an initial model, which suffers from a lack of sufficient projections in each bin. The purpose of this study is to demonstrate the feasibility of introducing a motion model acquired during CT simulation to MC-SART, coined model-based CBCT (MB-CBCT).Approach. For each of 5 patients, we acquired 5DCTs during simulation and pre-treatment CBCTs with a simultaneous breathing surrogate. We cross-calibrated the 5DCT and CBCT breathing waveforms by matching the diaphragms and employed the 5DCT motion model parameters for MC-SART. We introduced the Amplitude Reassignment Motion Modeling technique, which measures the ability of the model to control diaphragm sharpness by reassigning projection amplitudes with varying resolution. We evaluated the sharpness of tumors and compared them between MB-CBCT and 4D-CBCT. We quantified sharpness by fitting an error function across anatomical boundaries. Furthermore, we compared our MB-CBCT approach to the traditional MC-SART approach. We evaluated MB-CBCT's robustness over time by reconstructing multiple fractions for each patient and measuring consistency in tumor centroid locations between 4D-CBCT and MB-CBCT.Main results. We found that the diaphragm sharpness rose consistently with increasing amplitude resolution for 4/5 patients. We observed consistently high image quality across multiple fractions, and observed stable tumor centroids with an average 0.74 ± 0.31 mm difference between the 4D-CBCT and MB-CBCT. Overall, vast improvements over 3D-CBCT and 4D-CBCT were demonstrated by our MB-CBCT technique in terms of both diaphragm sharpness and overall image quality.Significance. This work is an important extension of the MC-SART technique. We demonstrated the ability ofa priori5DCT models to provide motion compensation for CBCT reconstruction. We showed improvements in image quality over both 4D-CBCT and the traditional MC-SART approach.

A generalized model for monitor units determination in ocular proton therapy using machine learning: A proof-of-concept study

Objective.Determining and verifying the number of monitor units is crucial to achieving the desired dose distribution in radiotherapy and maintaining treatment efficacy. However, current commercial treatment planning system(s) dedicated to ocular passive eyelines in proton therapy do not provide the number of monitor units for patient-specific plan delivery. Performing specific pre-treatment field measurements, which is time and resource consuming, is usually gold-standard practice. This proof-of-concept study reports on the development of a multi-institutional-based generalized model for monitor units determination in proton therapy for eye melanoma treatments.Approach.To cope with the small number of patients being treated in proton centers, three European institutes participated in this study. Measurements data were collected to address output factor differences across the institutes, especially as function of field size, spread-out Bragg peak modulation width, residual range, and air gap. A generic model for monitor units prediction using a large number of 3748 patients and broad diversity in tumor patterns, was evaluated using six popular machine learning algorithms: (i) decision tree; (ii) random forest, (iii) extra trees, (iv) K-nearest neighbors, (v) gradient boosting, and (vi) the support vector regression. Features used as inputs into each machine learning pipeline were: Spread-out Bragg peak width, range, air gap, fraction and calibration doses. Performance measure was scored using the mean absolute error, which was the difference between predicted and real monitor units, as collected from institutional gold-standard methods.Main results.Predictions across algorithms were accurate within 3% uncertainty for up to 85.2% of the plans and within 10% uncertainty for up to 98.6% of the plans with the extra trees algorithm.Significance.A proof-of-concept of using machine learning-based generic monitor units determination in ocular proton therapy has been demonstrated. This could trigger the development of an independent monitor units calculation tool for clinical use.

Machine-learning prediction of treatment response to stereotactic body radiation therapy in oligometastatic gynecological cancer: A multi-institutional study

BACKGROUND AND PURPOSE

We aimed to develop and validate different machine-learning (ML) prediction models for the complete response of oligometastatic gynecological cancer after SBRT.

MATERIAL AND METHODS

One hundred fifty-seven patients with 272 lesions from 14 different institutions and treated with SBRT with radical intent were included. Thirteen datasets including 222 lesions were combined for model training and internal validation purposes, with an 80:20 ratio. The external testing dataset was selected as the fourteenth Institution with 50 lesions. Lesions that achieved complete response (CR) were defined as responders. Prognostic clinical and dosimetric variables were selected using the LASSO algorithm. Six supervised ML models, including logistic regression (LR), classification and regression tree analysis (CART) and support vector machine (SVM) using four different kernels, were trained and tested to predict the complete response of uterine lesions after SBRT. The performance of models was assessed by receiver operating characteristic curves (ROC), area under the curve (AUC) and calibration curves. An explainable approach based on SHapley Additive exPlanations (SHAP) method was deployed to generate individual explanations of the model's decisions.

RESULTS

63.6% of lesions had a complete response and were used as ground truth for the supervised models. LASSO strongly associated complete response with three variables, namely the lesion volume (PTV), the type of lesions (lymph-nodal versus parenchymal), and the biological effective dose (BED10), that were used as input for ML modeling. In the training set, the AUCs for complete response were 0.751 (95% CI: 0.716-0.786), 0.766 (95% CI: 0.729-0.802) and 0.800 (95% CI: 0.742-0.857) for the LR, CART and SVM with a radial basis function kernel, respectively. These models achieve AUC values of 0.727 (95% CI: 0.669-0.795), 0.734 (95% CI: 0.649-0.815) and 0.771 (95% CI: 0.717-0.824) in the external testing set, demonstrating excellent generalizability.

CONCLUSION

ML models enable a reliable prediction of the treatment response of oligometastatic lesions receiving SBRT. This approach may assist radiation oncologists to tailor more individualized treatment plans for oligometastatic patients.

Proof-of-concept study of artificial intelligence-assisted review of CBCT image guidance

PURPOSE

Automation and computer assistance can support quality assurance tasks in radiotherapy. Retrospective image review requires significant human resources, and automation of image review remains a noteworthy missing element in previous work. Here, we present initial findings from a proof-of-concept clinical implementation of an AI-assisted review of CBCT registrations used for patient setup.

METHODS

An automated pipeline was developed and executed nightly, utilizing python scripts to interact with the clinical database through DICOM networking protocol and automate data retrieval and analysis. A previously developed artificial intelligence (AI) algorithm scored CBCT setup registrations based on misalignment likelihood, using a scale from 0 (most unlikely) through 1 (most likely). Over a 45-day period, 1357 pre-treatment CBCT registrations from 197 patients were retrieved and analyzed by the pipeline. Daily summary reports of the previous day's registrations were produced. Initial action levels targeted 10% of cases to highlight for in-depth physics review. A validation subset of 100 cases was scored by three independent observers to characterize AI-model performance.

RESULTS

Following an ROC analysis, a global threshold for model predictions of 0.87 was determined, with a sensitivity of 100% and specificity of 82%. Inspecting the observer scores for the stratified validation dataset showed a statistically significant correlation between observer scores and model predictions.

CONCLUSION

In this work, we describe the implementation of an automated AI-analysis pipeline for daily quantitative analysis of CBCT-guided patient setup registrations. The AI-model was validated against independent expert observers, and appropriate action levels were determined to minimize false positives without sacrificing sensitivity. Case studies demonstrate the potential benefits of such a pipeline to bolster quality and safety programs in radiotherapy. To the authors' knowledge, there are no previous works performing AI-assisted assessment of pre-treatment CBCT-based patient alignment.

A convolutional neural network model for EPID-based non-transit dosimetry

PURPOSE

To develop an alternative computational approach for EPID-based non-transit dosimetry using a convolutional neural network model.

METHOD

A U-net followed by a non-trainable layer named True Dose Modulation recovering the spatialized information was developed. The model was trained on 186 Intensity-Modulated Radiation Therapy Step & Shot beams from 36 treatment plans of different tumor locations to convert grayscale portal images into planar absolute dose distributions. Input data were acquired from an amorphous-Silicon Electronic Portal Image Device and a 6 MV X-ray beam. Ground truths were computed from a conventional kernel-based dose algorithm. The model was trained by a two-step learning process and validated through a five-fold cross-validation procedure with sets of training and validation of 80% and 20%, respectively. A study regarding the dependance of the amount of training data was conducted. The performance of the model was evaluated from a quantitative analysis based the ϒ-index, absolute and relative errors computed between the inferred dose distributions and ground truths for six square and 29 clinical beams from seven treatment plans. These results were also compared to those of an existing portal image-to-dose conversion algorithm.

RESULTS

For the clinical beams, averages of ϒ-index and ϒ-passing rate (2%-2mm > 10% D_max ) of 0.24 (±0.04) and 99.29 (±0.70)% were obtained. For the same metrics and criteria, averages of 0.31 (±0.16) and 98.83 (±2.40)% were obtained with the six square beams. Overall, the developed model performed better than the existing analytical method. The study also showed that sufficient model accuracy can be achieved with the amount of training samples used.

CONCLUSION

A deep learning-based model was developed to convert portal images into absolute dose distributions. The accuracy obtained shows that this method has great potential for EPID-based non-transit dosimetry.

Machine Learning Models Using Routinely Collected Clinical Data Offer Robust and Interpretable Predictions of 90-Day Unplanned Acute Care Use for Cancer Immunotherapy Patients

PURPOSE

Clinical management of patients receiving immune checkpoint inhibitors (ICIs) could be informed using accurate predictive tools to identify patients at risk of short-term acute care utilization (ACU). We used routinely collected data to develop and assess machine learning (ML) algorithms to predict unplanned ACU within 90 days of ICI treatment initiation.

METHODS

We used aggregated electronic health record data from 7,960 patients receiving ICI treatments to train and assess eight ML algorithms. We developed the models using pre-SARS-COV-19 COVID-19 data generated between January 2016 and February 2020. We validated our algorithms using data collected between March 2020 and June 2022 (peri-COVID-19 sample). We assessed performance using area under the receiver operating characteristic curves (AUROC), sensitivity, specificity, and calibration plots. We derived intuitive explanations of predictions using variable importance and Shapley additive explanation analyses. We assessed the marginal performance of ML models compared with that of univariate and multivariate logistic regression (LR) models.

RESULTS

Most algorithms significantly outperformed the univariate and multivariate LR models. The extreme gradient boosting trees (XGBT) algorithm demonstrated the best overall performance (AUROC, 0.70; sensitivity, 0.53; specificity, 0.74) on the peri-COVID-19 sample. The algorithm performance was stable across both pre- and peri-COVID-19 samples, as well as ICI regimen and cancer groups. Type of ICI agents, oxygen saturation, diastolic blood pressure, albumin level, platelet count, immature granulocytes, absolute monocyte, chloride level, red cell distribution width, and alcohol intake were the top 10 key predictors used by the XGBT algorithm.

CONCLUSION

Machine learning algorithms trained using routinely collected data outperformed traditional statistical models when predicting 90-day ACU. The XGBT algorithm has the potential to identify high-ACU risk patients and enable preventive interventions to avoid ACU.

The role of machine learning in HIV risk prediction

Despite advances in reducing HIV-related mortality, persistently high HIV incidence rates are undermining global efforts to end the epidemic by 2030. The UNAIDS Fast-track targets as well as other preventative strategies, such as pre-exposure prophylaxis, have been identified as priority areas to reduce the ongoing transmission threatening to undermine recent progress. Accurate and granular risk prediction is critical for these campaigns but is often lacking in regions where the burden is highest. Owing to their ability to capture complex interactions between data, machine learning and artificial intelligence algorithms have proven effective at predicting the risk of HIV infection in both high resource and low resource settings. However, interpretability of these algorithms presents a challenge to the understanding and adoption of these algorithms. In this perspectives article, we provide an introduction to machine learning and discuss some of the important considerations when choosing the variables used in model development and when evaluating the performance of different machine learning algorithms, as well as the role emerging tools such as Shapely Additive Explanations may play in helping understand and decompose these models in the context of HIV. Finally, we discuss some of the potential public health and clinical use cases for such decomposed risk assessment models in directing testing and preventative interventions including pre-exposure prophylaxis, as well as highlight the potential integration synergies with algorithms that predict the risk of sexually transmitted infections and tuberculosis.

Scalable radiotherapy data curation infrastructure for deep-learning based autosegmentation of organs-at-risk: A case study in head and neck cancer

In this era of patient-centered, outcomes-driven and adaptive radiotherapy, deep learning is now being successfully applied to tackle imaging-related workflow bottlenecks such as autosegmentation and dose planning. These applications typically require supervised learning approaches enabled by relatively large, curated radiotherapy datasets which are highly reflective of the contemporary standard of care. However, little has been previously published describing technical infrastructure, recommendations, methods or standards for radiotherapy dataset curation in a holistic fashion. Our radiation oncology department has recently embarked on a large-scale project in partnership with an external partner to develop deep-learning-based tools to assist with our radiotherapy workflow, beginning with autosegmentation of organs-at-risk. This project will require thousands of carefully curated radiotherapy datasets comprising all body sites we routinely treat with radiotherapy. Given such a large project scope, we have approached the need for dataset curation rigorously, with an aim towards building infrastructure that is compatible with efficiency, automation and scalability. Focusing on our first use-case pertaining to head and neck cancer, we describe our developed infrastructure and novel methods applied to radiotherapy dataset curation, inclusive of personnel and workflow organization, dataset selection, expert organ-at-risk segmentation, quality assurance, patient de-identification, data archival and transfer. Over the course of approximately 13 months, our expert multidisciplinary team generated 490 curated head and neck radiotherapy datasets. This task required approximately 6000 human-expert hours in total (not including planning and infrastructure development time). This infrastructure continues to evolve and will support ongoing and future project efforts.

Machine learning models to predict the delivered positions of Elekta multileaf collimator leaves for volumetric modulated arc therapy

PURPOSE

Accurate positioning of multileaf collimator (MLC) leaves during volumetric modulated arc therapy (VMAT) is essential for accurate treatment delivery. We developed a linear regression, support vector machine, random forest, extreme gradient boosting (XGBoost), and an artificial neural network (ANN) for predicting the delivered leaf positions for VMAT plans.

METHODS

For this study, 160 MLC log files from 80 VMAT plans were obtained from a single institution treated on 3 Elekta Versa HD linear accelerators. The gravity vector, X1 and X2 jaw positions, leaf gap, leaf position, leaf velocity, and leaf acceleration were extracted and used as model inputs. The models were trained using 70% of the log files and tested on the remaining 30%. Mean absolute error (MAE), root mean square error (RMSE), the coefficient of determination R² , and fitted line plots showing the relationship between delivered and predicted leaf positions were used to evaluate model performance.

RESULTS

The models achieved the following errors: linear regression (MAE = 0.158 mm, RMSE = 0.225 mm), support vector machine (MAE = 0.141 mm, RMSE = 0.199 mm), random forest (MAE = 0.161 mm, RMSE = 0.229 mm), XGBoost (MAE = 0.185 mm, RMSE = 0.273 mm), and ANN (MAE = 0.361 mm, RMSE = 0.521 mm). A significant correlation between a plan's gamma passing rate (GPR) and the prediction errors of linear regression, support vector machine, and random forest is seen (p < 0.045).

CONCLUSIONS

We examined various models to predict the delivered MLC positions for VMAT plans treated with Elekta linacs. Linear regression, support vector machine, random forest, and XGBoost achieved lower errors than ANN. Models that can accurately predict the individual leaf positions during treatment can help identify leaves that are deviating from the planned position, which can improve a plan's GPR.

Intelligent systems for precision dental diagnosis and treatment planning – A review

Machines have changed the course of mankind. Simple machines were the basis of human civilization. Today with humongous technological development, machines are intelligent enough to carry out very complex nerve-racking tasks. The ability of a machine to learn from algorithms changed eventually into, the machine learning by itself, which constitutes artificial intelligence. Literature has plausible evidence for the use of intelligent systems in medical field. Artificial intelligence has been used in the multiple denominations of dentistry. These machines are used in the precision diagnosis, interpretation of medical images, accumulation of data, classification and compilation of records, determination of treatment and construction of a personalized treatment plan. Artificial intelligence can help in timely diagnosis of complex dental diseases which would ultimately aid in rapid commencement of treatment. Research helps us understand the effectiveness and challenges in the use of this technology. The apt use of intelligent systems could transform the entire medical system for the better.

Recent Applications of Artificial Intelligence in Radiotherapy: Where We Are and Beyond

In recent decades, artificial intelligence (AI) tools have been applied in many medical fields, opening the possibility of finding novel solutions for managing very complex and multifactorial problems, such as those commonly encountered in radiotherapy (RT). We conducted a PubMed and Scopus search to identify the AI application field in RT limited to the last four years. In total, 1824 original papers were identified, and 921 were analyzed by considering the phase of the RT workflow according to the applied AI approaches. AI permits the processing of large quantities of information, data, and images stored in RT oncology information systems, a process that is not manageable for individuals or groups. AI allows the iterative application of complex tasks in large datasets (e.g., delineating normal tissues or finding optimal planning solutions) and might support the entire community working in the various sectors of RT, as summarized in this overview. AI-based tools are now on the roadmap for RT and have been applied to the entire workflow, mainly for segmentation, the generation of synthetic images, and outcome prediction. Several concerns were raised, including the need for harmonization while overcoming ethical, legal, and skill barriers.

Machine Learning-Based Short-Term Mortality Prediction Models for Patients With Cancer Using Electronic Health Record Data: Systematic Review and Critical Appraisal

BACKGROUND

In the United States, national guidelines suggest that aggressive cancer care should be avoided in the final months of life. However, guideline compliance currently requires clinicians to make judgments based on their experience as to when a patient is nearing the end of their life. Machine learning (ML) algorithms may facilitate improved end-of-life care provision for patients with cancer by identifying patients at risk of short-term mortality.

OBJECTIVE

This study aims to summarize the evidence for applying ML in ≤1-year cancer mortality prediction to assist with the transition to end-of-life care for patients with cancer.

METHODS

We searched MEDLINE, Embase, Scopus, Web of Science, and IEEE to identify relevant articles. We included studies describing ML algorithms predicting ≤1-year mortality in patients of oncology. We used the prediction model risk of bias assessment tool to assess the quality of the included studies.

RESULTS

We included 15 articles involving 110,058 patients in the final synthesis. Of the 15 studies, 12 (80%) had a high or unclear risk of bias. The model performance was good: the area under the receiver operating characteristic curve ranged from 0.72 to 0.92. We identified common issues leading to biased models, including using a single performance metric, incomplete reporting of or inappropriate modeling practice, and small sample size.

CONCLUSIONS

We found encouraging signs of ML performance in predicting short-term cancer mortality. Nevertheless, no included ML algorithms are suitable for clinical practice at the current stage because of the high risk of bias and uncertainty regarding real-world performance. Further research is needed to develop ML models using the modern standards of algorithm development and reporting.

Reporting of Model Performance and Statistical Methods in Studies That Use Machine Learning to Develop Clinical Prediction Models: Protocol for a Systematic Review

BACKGROUND

With the growing excitement of the potential benefits of using machine learning and artificial intelligence in medicine, the number of published clinical prediction models that use these approaches has increased. However, there is evidence (albeit limited) that suggests that the reporting of machine learning-specific aspects in these studies is poor. Further, there are no reviews assessing the reporting quality or broadly accepted reporting guidelines for these aspects.

OBJECTIVE

This paper presents the protocol for a systematic review that will assess the reporting quality of machine learning-specific aspects in studies that use machine learning to develop clinical prediction models.

METHODS

We will include studies that use a supervised machine learning algorithm to develop a prediction model for use in clinical practice (ie, for diagnosis or prognosis of a condition or identification of candidates for health care interventions). We will search MEDLINE for studies published in 2019, pseudorandomly sort the records, and screen until we obtain 100 studies that meet our inclusion criteria. We will assess reporting quality with a novel checklist developed in parallel with this review, which includes content derived from existing reporting guidelines, textbooks, and consultations with experts. The checklist will cover 4 key areas where the reporting of machine learning studies is unique: modelling steps (order and data used for each step), model performance (eg, reporting the performance of each model compared), statistical methods (eg, describing the tuning approach), and presentation of models (eg, specifying the predictors that contributed to the final model).

RESULTS

We completed data analysis in August 2021 and are writing the manuscript. We expect to submit the results to a peer-reviewed journal in early 2022.

CONCLUSIONS

This review will contribute to more standardized and complete reporting in the field by identifying areas where reporting is poor and can be improved.

TRIAL REGISTRATION

PROSPERO International Prospective Register of Systematic Reviews CRD42020206167; https://www.crd.york.ac.uk/PROSPERO/display_record.php?RecordID=206167.

INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID)

RR1-10.2196/30956.

Development of AI-driven prediction models to realize real-time tumor tracking during radiotherapy

BACKGROUND

In infrared reflective (IR) marker-based hybrid real-time tumor tracking (RTTT), the internal target position is predicted with the positions of IR markers attached on the patient's body surface using a prediction model. In this work, we developed two artificial intelligence (AI)-driven prediction models to improve RTTT radiotherapy, namely, a convolutional neural network (CNN) and an adaptive neuro-fuzzy inference system (ANFIS) model. The models aim to improve the accuracy in predicting three-dimensional tumor motion.

METHODS

From patients whose respiration-induced motion of the tumor, indicated by the fiducial markers, exceeded 8 mm, 1079 logfiles of IR marker-based hybrid RTTT (IR Tracking) with the gimbal-head radiotherapy system were acquired and randomly divided into two datasets. All the included patients were breathing freely with more than four external IR markers. The historical dataset for the CNN model contained 1003 logfiles, while the remaining 76 logfiles complemented the evaluation dataset. The logfiles recorded the external IR marker positions at a frequency of 60 Hz and fiducial markers as surrogates for the detected target positions every 80-640 ms for 20-40 s. For each logfile in the evaluation dataset, the prediction models were trained based on the data in the first three quarters of the recording period. In the last quarter, the performance of the patient-specific prediction models was tested and evaluated. The overall performance of the AI-driven prediction models was ranked by the percentage of predicted target position within 2 mm of the detected target position. Moreover, the performance of the AI-driven models was compared to a regression prediction model currently implemented in gimbal-head radiotherapy systems.

RESULTS

The percentage of the predicted target position within 2 mm of the detected target position was 95.1%, 92.6% and 85.6% for the CNN, ANFIS, and regression model, respectively. In the evaluation dataset, the CNN, ANFIS, and regression model performed best in 43, 28 and 5 logfiles, respectively.

CONCLUSIONS

The proposed AI-driven prediction models outperformed the regression prediction model, and the overall performance of the CNN model was slightly better than that of the ANFIS model on the evaluation dataset.

Regression Analysis of Rectal Cancer and Possible Application of Artificial Intelligence (AI) Utilization in Radiotherapy

Artificial Intelligence (AI) has been widely employed in the medical field in recent years in such areas as image segmentation, medical image registration, and computer-aided detection. This study explores one application of using AI in adaptive radiation therapy treatment planning by predicting the tumor volume reduction rate (TVRR). Cone beam computed tomography (CBCT) scans of twenty rectal cancer patients were collected to observe the change in tumor volume over the course of a standard five-week radiotherapy treatment. In addition to treatment volume, patient data including patient age, gender, weight, number of treatment fractions, and dose per fraction were also collected. Application of a stepwise regression model showed that age, dose per fraction and weight were the best predictors for tumor volume reduction rate.

Emerging methods for prostate cancer imaging: evaluating cancer structure and metabolic alterations more clearly

Imaging plays a fundamental role in all aspects of the cancer management pathway. However, conventional imaging techniques are largely reliant on morphological and size descriptors that have well-known limitations, particularly when considering targeted-therapy response monitoring. Thus, new imaging methods have been developed to characterise cancer and are now routinely implemented, such as diffusion-weighted imaging, dynamic contrast enhancement, positron emission technology (PET) and magnetic resonance spectroscopy. However, despite the improvement these techniques have enabled, limitations still remain. Novel imaging methods are now emerging, intent on further interrogating cancers. These techniques are at different stages of maturity along the biomarker pathway and aim to further evaluate the cancer microstructure (vascular, extracellular and restricted diffusion for cytometry in tumours) magnetic resonance imaging (MRI), luminal water fraction imaging] as well as the metabolic alterations associated with cancers (novel PET tracers, hyperpolarised MRI). Finally, the use of machine learning has shown powerful potential applications. By using prostate cancer as an exemplar, this Review aims to showcase these potentially potent imaging techniques and what stage we are at in their application to conventional clinical practice.

Combining computed tomography and biologically effective dose in radiomics and deep learning improves prediction of tumor response to robotic lung stereotactic body radiation therapy

PURPOSE

The aim of this study is to improve the performance of machine learning (ML) models in predicting response of non-small cell lung cancer (NSCLC) to stereotactic body radiation therapy (SBRT) by integrating image features from pre-treatment computed tomography (CT) with features from the biologically effective dose (BED) distribution.

MATERIALS AND METHODS

Image features, consisting of crafted radiomic features or machine-learned features extracted using a convolutional neural network, were calculated from pre-treatment CT data and from dose distributions converted into BED for 80 NSCLC lesions over 76 patients treated with robotic guided SBRT. ML models using different combinations of features were trained to predict complete or partial response according to response criteria in solid tumors, including radiomics CT (Rad_CT ), radiomics CT and BED (Rad_CT,BED ), deep learning (DL) CT (DL_CT ), and DL CT and BED (DL_CT,BED ). Training of ML included feature selection by neighborhood component analysis followed by ensemble ML using robust boosting. A model was considered as acceptable when the sum of average sensitivity and specificity on test data in repeated cross validations was at least 1.5.

RESULTS

Complete or partial response occurred in 58 out of 80 lesions. The best models to predict the tumor response were those using BED variables, achieving significantly better area under curve (AUC) and accuracy than those using only features from CT, including a Rad_CT,BED model using three radiomic features from BED, which scored an accuracy of 0.799 (95% confidence intervals (0.75-0.85)) and AUC of 0.773 (0.688-0.846), and a DL_CT,BED model also using three variables with an accuracy of 0.798 (0.649-0.829) and AUC of 0.812 (0.755-0.867).

CONCLUSION

According to our results, the inclusion of BED features improves the response prediction of ML models for lung cancer patients undergoing SBRT, regardless of the use of radiomic or DL features.

Non-small cell lung carcinoma histopathological subtype phenotyping using high-dimensional multinomial multiclass CT radiomics signature

OBJECTIVE

The aim of this study was to identify the most important features and assess their discriminative power in the classification of the subtypes of NSCLC.

METHODS

This study involved 354 pathologically proven NSCLC patients including 134 squamous cell carcinoma (SCC), 110 large cell carcinoma (LCC), 62 not other specified (NOS), and 48 adenocarcinoma (ADC). In total, 1433 radiomics features were extracted from 3D volumes of interest drawn on the malignant lesion identified on CT images. Wrapper algorithm and multivariate adaptive regression splines were implemented to identify the most relevant/discriminative features. A multivariable multinomial logistic regression was employed with 1000 bootstrapping samples based on the selected features to classify four main subtypes of NSCLC.

RESULTS

The results revealed that the texture features, specifically gray level size zone matrix features (GLSZM), were the significant indicators of NSCLC subtypes. The optimized classifier achieved an average precision, recall, F1-score, and accuracy of 0.710, 0.703, 0.706, and 0.865, respectively, based on the selected features by the wrapper algorithm.

CONCLUSIONS

Our CT radiomics approach demonstrated impressive potential for the classification of the four main histological subtypes of NSCLC, It is anticipated that CT radiomics could be useful in treatment planning and precision medicine.

Detecting Spurious Correlations With Sanity Tests for Artificial Intelligence Guided Radiology Systems

Artificial intelligence (AI) has been successful at solving numerous problems in machine perception. In radiology, AI systems are rapidly evolving and show progress in guiding treatment decisions, diagnosing, localizing disease on medical images, and improving radiologists' efficiency. A critical component to deploying AI in radiology is to gain confidence in a developed system's efficacy and safety. The current gold standard approach is to conduct an analytical validation of performance on a generalization dataset from one or more institutions, followed by a clinical validation study of the system's efficacy during deployment. Clinical validation studies are time-consuming, and best practices dictate limited re-use of analytical validation data, so it is ideal to know ahead of time if a system is likely to fail analytical or clinical validation. In this paper, we describe a series of sanity tests to identify when a system performs well on development data for the wrong reasons. We illustrate the sanity tests' value by designing a deep learning system to classify pancreatic cancer seen in computed tomography scans.

Integrating Multiomics Information in Deep Learning Architectures for Joint Actuarial Outcome Prediction in Non-Small Cell Lung Cancer Patients After Radiation Therapy

PURPOSE

Novel actuarial deep learning neural network (ADNN) architectures are proposed for joint prediction of radiation therapy outcomes-radiation pneumonitis (RP) and local control (LC)-in stage III non-small cell lung cancer (NSCLC) patients. Unlike normal tissue complication probability/tumor control probability models that use dosimetric information solely, our proposed models consider complex interactions among multiomics information including positron emission tomography (PET) radiomics, cytokines, and miRNAs. Additional time-to-event information is also used in the actuarial prediction.

METHODS AND MATERIALS

Three architectures were investigated: ADNN-DVH considered dosimetric information only; ADNN-com integrated multiomics information; and ADNN-com-joint combined RP2 (RP grade ≥2) and LC prediction. In these architectures, differential dose-volume histograms (DVHs) were fed into 1D convolutional neural networks (CNN) for extracting reduced representations. Variational encoders were used to learn representations of imaging and biological data. Reduced representations were fed into Surv-Nets to predict time-to-event probabilities for RP2 and LC independently and jointly by incorporating time information into designated loss functions.

RESULTS

Models were evaluated on 117 retrospective patients and were independently tested on 25 newly accrued patients prospectively. A multi-institutional RTOG0617 data set of 327 patients was used for external validation. ADNN-DVH yielded cross-validated c-indexes (95% confidence intervals) of 0.660 (0.630-0.690) for RP2 prediction and 0.727 (0.700-0.753) for LC prediction, outperforming a generalized Lyman model for RP2 (0.613 [0.583-0.643]) and a generalized log-logistic model for LC (0.569 [0.545-0.594]). The independent internal test and external validation yielded similar results. ADNN-com achieved an even better performance than ADNN-DVH on both cross-validation and independent internal test. Furthermore, ADNN-com-joint, which yielded performance similar to ADNN-com, realized joint prediction with c-indexes of 0.705 (0.676-0.734) for RP2 and 0.740 (0.714-0.765) for LC and achieved an area under a free-response receiving operator characteristic curve (AU-FROC) of 0.729 (0.697-0.773) for the joint prediction of RP2 and LC.

CONCLUSION

Novel deep learning architectures that integrate multiomics information outperformed traditional normal tissue complication probability/tumor control probability models in actuarial prediction of RP2 and LC.

Performance stability evaluation of atlas-based machine learning radiation therapy treatment planning in prostate cancer

Atlas-based machine learning (ML) for radiation therapy (RT) treatment planning is effective at tailoring dose distributions to account for unique patient anatomies by selecting the most appropriate patients from the training database (atlases) to inform dose prediction for new patients. However, variations in clinical practice between the training dataset and a new patient to be planned may impact ML performance by confounding atlas selection. In this study, we simulated various contouring practices in prostate cancer RT to investigate the impact of changing input data on atlas-based ML treatment planning. We generated 225 ML plans for nine bespoke contouring protocol scenarios (reduced target margins, modified organ-at-risk (OAR) definitions, and inclusion of optional OARs less represented in the training database) on 25 patient datasets by applying a single, previously trained and validated ML model for prostate cancer followed by dose mimicking to create a final deliverable plan. ML treatment plans for each scenario were compared to base ML treatment plans that followed a contouring protocol consistent with the model training data. ML performance was evaluated based on atlas distance metrics that are calculated during ML dose prediction. There were significant changes between atlases selected for the base ML treatment plans and treatment plans when planning target volume margins were reduced and/or optional OARs were included. The deliverability of ML predicted dose distributions based on gamma analysis between predicted and mimicked final deliverable dose showed significant differences for seven out of eight scenarios compared with the base ML treatment plans. Overall, there were small but statistically significant dosimetric changes in predicted and mimicked dose with addition of optional OAR contours. This work presents a framework for benchmarking and performance monitoring of ML treatment planning algorithms in the context of evolving clinical practices.

Future of biomarker evaluation in the realm of artificial intelligence algorithms: application in improved therapeutic stratification of patients with breast and prostate cancer

Clinical workflows in oncology depend on predictive and prognostic biomarkers. However, the growing number of complex biomarkers contributes to costly and delayed decision-making in routine oncology care and treatment. As cancer is expected to rank as the leading cause of death and the single most important barrier to increasing life expectancy in the 21st century, there is a major emphasis on precision medicine, particularly individualisation of treatment through better prediction of patient outcome. Over the past few years, both surgical and pathology specialties have suffered cutbacks and a low uptake of pathology specialists means a solution is required to enable high-throughput screening and personalised treatment in this area to alleviate bottlenecks. Digital imaging in pathology has undergone an exponential period of growth. Deep-learning (DL) platforms for hematoxylin and eosin (H&E) image analysis, with preliminary artificial intelligence (AI)-based grading capabilities of specimens, can evaluate image characteristics which may not be visually apparent to a pathologist and offer new possibilities for better modelling of disease appearance and possibly improve the prediction of disease stage and patient outcome. Although digital pathology and AI are still emerging areas, they are the critical components for advancing personalised medicine. Integration of transcriptomic analysis, clinical information and AI-based image analysis is yet an uncultivated field by which healthcare professionals can make improved treatment decisions in cancer. This short review describes the potential application of integrative AI in offering better detection, quantification, classification, prognosis and prediction of breast and prostate cancer and also highlights the utilisation of machine learning systems in biomarker evaluation.

Training and Validation of Deep Learning-Based Auto-Segmentation Models for Lung Stereotactic Ablative Radiotherapy Using Retrospective Radiotherapy Planning Contours

PURPOSE

Deep learning-based auto-segmented contour (DC) models require high quality data for their development, and previous studies have typically used prospectively produced contours, which can be resource intensive and time consuming to obtain. The aim of this study was to investigate the feasibility of using retrospective peer-reviewed radiotherapy planning contours in the training and evaluation of DC models for lung stereotactic ablative radiotherapy (SABR).

METHODS

Using commercial deep learning-based auto-segmentation software, DC models for lung SABR organs at risk (OAR) and gross tumor volume (GTV) were trained using a deep convolutional neural network and a median of 105 contours per structure model obtained from 160 publicly available CT scans and 50 peer-reviewed SABR planning 4D-CT scans from center A. DCs were generated for 50 additional planning CT scans from center A and 50 from center B, and compared with the clinical contours (CC) using the Dice Similarity Coefficient (DSC) and 95% Hausdorff distance (HD).

RESULTS

Comparing DCs to CCs, the mean DSC and 95% HD were 0.93 and 2.85mm for aorta, 0.81 and 3.32mm for esophagus, 0.95 and 5.09mm for heart, 0.98 and 2.99mm for bilateral lung, 0.52 and 7.08mm for bilateral brachial plexus, 0.82 and 4.23mm for proximal bronchial tree, 0.90 and 1.62mm for spinal cord, 0.91 and 2.27mm for trachea, and 0.71 and 5.23mm for GTV. DC to CC comparisons of center A and center B were similar for all OAR structures.

CONCLUSIONS

The DCs developed with retrospective peer-reviewed treatment contours approximated CCs for the majority of OARs, including on an external dataset. DCs for structures with more variability tended to be less accurate and likely require using a larger number of training cases or novel training approaches to improve performance. Developing DC models from existing radiotherapy planning contours appears feasible and warrants further clinical workflow testing.

Quantitative Molecular Positron Emission Tomography Imaging Using Advanced Deep Learning Techniques

The widespread availability of high-performance computing and the popularity of artificial intelligence (AI) with machine learning and deep learning (ML/DL) algorithms at the helm have stimulated the development of many applications involving the use of AI-based techniques in molecular imaging research. Applications reported in the literature encompass various areas, including innovative design concepts in positron emission tomography (PET) instrumentation, quantitative image reconstruction and analysis techniques, computer-aided detection and diagnosis, as well as modeling and prediction of outcomes. This review reflects the tremendous interest in quantitative molecular imaging using ML/DL techniques during the past decade, ranging from the basic principles of ML/DL techniques to the various steps required for obtaining quantitatively accurate PET data, including algorithms used to denoise or correct for physical degrading factors as well as to quantify tracer uptake and metabolic tumor volume for treatment monitoring or radiation therapy treatment planning and response prediction.This review also addresses future opportunities and current challenges facing the adoption of ML/DL approaches and their role in multimodality imaging.

Multi-institutional dose-segmented dosiomic analysis for predicting radiation pneumonitis after lung stereotactic body radiation therapy

PURPOSE

To predict radiation pneumonitis (RP) grade 2 or worse after lung stereotactic body radiation therapy (SBRT) using dose-based radiomic (dosiomic) features.

METHODS

This multi-institutional study included 247 early-stage nonsmall cell lung cancer patients who underwent SBRT with a prescribed dose of 48-70 Gy at an isocenter between June 2009 and March 2016. Ten dose-volume indices (DVIs) were used, including the mean lung dose, internal target volume size, and percentage of entire lung excluding the internal target volume receiving greater than x Gy (x = 5, 10, 15, 20, 25, 30, 35, and 40). A total of 6,808 dose-segmented dosiomic features, such as shape, first order, and texture features, were extracted from the dose distribution. Patients were randomly partitioned into two groups: model training (70%) and test datasets (30%) over 100 times. Dosiomic features were converted to z-scores (standardized values) with a mean of zero and a standard deviation (SD) of one to put different variables on the same scale. The feature dimension was reduced using the following methods: interfeature correlation based on Spearman's correlation coefficients and feature importance based on a light gradient boosting machine (LightGBM) feature selection function. Three different models were developed using LightGBM as follows: (a) a model with ten DVIs (DVI model), (b) a model with the selected dosiomic features (dosiomic model), and (c) a model with ten DVIs and selected dosiomic features (hybrid model). Suitable hyperparameters were determined by searching the largest average area under the curve (AUC) value in the receiver operating characteristic curve (ROC-AUC) via stratified fivefold cross-validation. Each of the final three models with the closest the ROC-AUC value to the average ROC-AUC value was applied to the test datasets. The classification performance was evaluated by calculating the ROC-AUC, AUC in the precision-recall curve (PR-AUC), accuracy, precision, recall, and f1-score. The entire process was repeated 100 times with randomization, and 100 individual models were developed for each of the three models. Then the mean value and SD for the 100 random iterations were calculated for each performance metric.

RESULTS

Thirty-seven (15.0%) patients developed RP after SBRT. The ROC-AUC and PR-AUC values in the DVI, dosiomic, and hybrid models were 0.660 ± 0.054 and 0.272 ± 0.052, 0.837 ± 0.054 and 0.510 ± 0.115, and 0.846 ± 0.049 and 0.531 ± 0.116, respectively. For each performance metric, the dosiomic and hybrid models outperformed the DVI models (P < 0.05). Texture-based dosiomic feature was confirmed as an effective indicator for predicting RP.

CONCLUSIONS

Our dose-segmented dosiomic approach improved the prediction of the incidence of RP after SBRT.

Requirements and reliability of AI in the medical context

The digital information age has been a catalyst in creating a renewed interest in Artificial Intelligence (AI) approaches, especially the subclass of computer algorithms that are popularly grouped into Machine Learning (ML). These methods have allowed one to go beyond limited human cognitive ability into understanding the complexity in the high dimensional data. Medical sciences have seen a steady use of these methods but have been slow in adoption to improve patient care. There are some significant impediments that have diluted this effort, which include availability of curated diverse data sets for model building, reliable human-level interpretation of these models, and reliable reproducibility of these methods for routine clinical use. Each of these aspects has several limiting conditions that need to be balanced out, considering the data/model building efforts, clinical implementation, integration cost to translational effort with minimal patient level harm, which may directly impact future clinical adoption. In this review paper, we will assess each aspect of the problem in the context of reliable use of the ML methods in oncology, as a representative study case, with the goal to safeguard utility and improve patient care in medicine in general.

Computer Vision in the Operating Room: Opportunities and Caveats

Effectiveness of computer vision techniques has been demonstrated through a number of applications, both within and outside healthcare. The operating room environment specifically is a setting with rich data sources compatible with computational approaches and high potential for direct patient benefit. The aim of this review is to summarize major topics in computer vision for surgical domains. The major capabilities of computer vision are described as an aid to surgical teams to improve performance and contribute to enhanced patient safety. Literature was identified through leading experts in the fields of surgery, computational analysis and modeling in medicine, and computer vision in healthcare. The literature supports the application of computer vision principles to surgery. Potential applications within surgery include operating room vigilance, endoscopic vigilance, and individual and team-wide behavioral analysis. To advance the field, we recommend collecting and publishing carefully annotated datasets. Doing so will enable the surgery community to collectively define well-specified common objectives for automated systems, spur academic research, mobilize industry, and provide benchmarks with which we can track progress. Leveraging computer vision approaches through interdisciplinary collaboration and advanced approaches to data acquisition, modeling, interpretation, and integration promises a powerful impact on patient safety, public health, and financial costs.

Promises and perils of artificial intelligence in dentistry

Artificial intelligence (AI) is a subdiscipline of computer science that has made substantial progress in medicine and there is a growing body of AI research in dentistry. Dentists should have an understanding of the foundational concepts and the ability to critically evaluate dental research in AI. Machine learning (ML) is a subfield of AI that most dental AI research is dedicated to. The most prolific area of ML research is automated interpretation of dental imaging. Other areas include providing treatment recommendations, predicting future disease and treatment outcomes. The research impact is limited by small datasets that do not harness the positive correlation between very large datasets and ML performance. There is also a need to standardize research methodologies and utilize performance metrics that are appropriate for the clinical context. In addition to research challenges, this article discusses the ethical, legal and logistical considerations associated with implementation in clinical practice. This includes explainable AI, model bias, data privacy and security. The future implications of AI in dentistry involve a promise for a novel form of practicing dentistry however, the effect of AI on patient outcomes is yet to be determined.

Prediction of multi-criteria optimization (MCO) parameter efficiency in volumetric modulated arc therapy (VMAT) treatment planning using machine learning (ML)

PURPOSE

To predict the impact of optimization parameter changes on dosimetric plan quality criteria in multi-criteria optimized volumetric-modulated-arc therapy (VMAT) planning prior to optimization using machine learning (ML).

METHODS

A data base comprising a total of 21,266 VMAT treatment plans for 44 cranial and 18 spinal patient geometries was generated. The underlying optimization algorithm is governed by three highly composite parameters which model a combination of important aspects of the solution. Patient geometries were parametrized via volume- and shape properties of the voxel objects and overlap-volume histograms (OVH) of the planning-target-volume (PTV) and a relevant organ-at-risk (OAR). The impact of changes in one of the three optimization parameters on the maximally achievable value range of five dosimetric properties of the resulting dose distributions was studied. To predict the extent of this impact based on patient geometry, treatment site, and current parameter settings prior to optimization, three different ML-models were trained and tested. Precision-recall curves, as well as the area-under-curve (AUC) of the resulting receiver-operator-characteristic (ROC) curves were analyzed for model assessment.

RESULTS

Successful identification of parameter regions resulting in a high variability of dosimetric plan properties depended on the choice of geometry features, the treatment indication and the plan property under investigation. AUC values between 0.82 and 0.99 could be achieved. The best average-precision (AP) values obtained from the corresponding precision/recall curves ranged from 0.71 to 0.99.

CONCLUSIONS

Machine learning models trained on a database of pre-optimized treatment plans can help finding relevant optimization parameter ranges prior to optimization.

Lessons learned in transitioning to AI in the medical imaging of COVID-19

The coronavirus disease 2019 (COVID-19) pandemic has wreaked havoc across the world. It also created a need for the urgent development of efficacious predictive diagnostics, specifically, artificial intelligence (AI) methods applied to medical imaging. This has led to the convergence of experts from multiple disciplines to solve this global pandemic including clinicians, medical physicists, imaging scientists, computer scientists, and informatics experts to bring to bear the best of these fields for solving the challenges of the COVID-19 pandemic. However, such a convergence over a very brief period of time has had unintended consequences and created its own challenges. As part of Medical Imaging Data and Resource Center initiative, we discuss the lessons learned from career transitions across the three involved disciplines (radiology, medical imaging physics, and computer science) and draw recommendations based on these experiences by analyzing the challenges associated with each of the three associated transition types: (1) AI of non-imaging data to AI of medical imaging data, (2) medical imaging clinician to AI of medical imaging, and (3) AI of medical imaging to AI of COVID-19 imaging. The lessons learned from these career transitions and the diffusion of knowledge among them could be accomplished more effectively by recognizing their associated intricacies. These lessons learned in the transitioning to AI in the medical imaging of COVID-19 can inform and enhance future AI applications, making the whole of the transitions more than the sum of each discipline, for confronting an emergency like the COVID-19 pandemic or solving emerging problems in biomedicine.

A Deep Learning-Based Automated CT Segmentation of Prostate Cancer Anatomy for Radiation Therapy Planning-A Retrospective Multicenter Study

A commercial deep learning (DL)-based automated segmentation tool (AST) for computed tomography (CT) is evaluated for accuracy and efficiency gain within prostate cancer patients. Thirty patients from six clinics were reviewed with manual- (MC), automated- (AC) and automated and edited (AEC) contouring methods. In the AEC group, created contours (prostate, seminal vesicles, bladder, rectum, femoral heads and penile bulb) were edited, whereas the MC group included empty datasets for MC. In one clinic, lymph node CTV delineations were evaluated for interobserver variability. Compared to MC, the mean time saved using the AST was 12 min for the whole data set (46%) and 12 min for the lymph node CTV (60%), respectively. The delineation consistency between MC and AEC groups according to the Dice similarity coefficient (DSC) improved from 0.78 to 0.94 for the whole data set and from 0.76 to 0.91 for the lymph nodes. The mean DSCs between MC and AC for all six clinics were 0.82 for prostate, 0.72 for seminal vesicles, 0.93 for bladder, 0.84 for rectum, 0.69 for femoral heads and 0.51 for penile bulb. This study proves that using a general DL-based AST for CT images saves time and improves consistency.

From multisource data to clinical decision aids in radiation oncology: The need for a clinical data science community

Big data are no longer an obstacle; now, by using artificial intelligence (AI), previously undiscovered knowledge can be found in massive data collections. The radiation oncology clinic daily produces a large amount of multisource data and metadata during its routine clinical and research activities. These data involve multiple stakeholders and users. Because of a lack of interoperability, most of these data remain unused, and powerful insights that could improve patient care are lost. Changing the paradigm by introducing powerful AI analytics and a common vision for empowering big data in radiation oncology is imperative. However, this can only be achieved by creating a clinical data science community in radiation oncology. In this work, we present why such a community is needed to translate multisource data into clinical decision aids.

Machine learning applications in radiation oncology: Current use and needs to support clinical implementation

BACKGROUND AND PURPOSE

The use of artificial intelligence (AI)/ machine learning (ML) applications in radiation oncology is emerging, however no clear guidelines on commissioning of ML-based applications exist. The purpose of this study was therefore to investigate the current use and needs to support implementation of ML-based applications in routine clinical practice.

MATERIALS AND METHODS

A survey was conducted among medical physicists in radiation oncology, consisting of four parts: clinical applications (1), model training, acceptance and commissioning (2), quality assurance (QA) in clinical practice and General Data Protection Regulation (GDPR) (3), and need for education and guidelines (4). Survey answers of medical physicists of the same radiation oncology centre were treated as a separate unique responder in case reporting on different AI applications.

RESULTS

In total, 213 medical physicists from 202 radiation oncology centres were included in the analysis. Sixty-nine percent (1 4 7) was using (37%) or preparing (32%) to use ML in clinic, mostly for contouring and treatment planning. In 86%, human observers were still involved in daily clinical use for quality check of the output of the ML algorithm. Knowledge on ethics, legislation and data sharing was limited and scattered among responders. Besides the need for (implementation) guidelines, training of medical physicists and larger databases containing multicentre data was found to be the top priority to accommodate the further introduction of ML in clinical practice.

CONCLUSION

The results of this survey indicated the need for education and guidelines on the implementation and quality assurance of ML-based applications to benefit clinical introduction.

Overview of artificial intelligence-based applications in radiotherapy: Recommendations for implementation and quality assurance

Artificial Intelligence (AI) is currently being introduced into different domains, including medicine. Specifically in radiation oncology, machine learning models allow automation and optimization of the workflow. A lack of knowledge and interpretation of these AI models can hold back wide-spread and full deployment into clinical practice. To facilitate the integration of AI models in the radiotherapy workflow, generally applicable recommendations on implementation and quality assurance (QA) of AI models are presented. For commonly used applications in radiotherapy such as auto-segmentation, automated treatment planning and synthetic computed tomography (sCT) the basic concepts are discussed in depth. Emphasis is put on the commissioning, implementation and case-specific and routine QA of AI models needed for a methodical introduction in clinical practice.

Artificial Intelligence in Dentistry: Chances and Challenges

The term "artificial intelligence" (AI) refers to the idea of machines being capable of performing human tasks. A subdomain of AI is machine learning (ML), which "learns" intrinsic statistical patterns in data to eventually cast predictions on unseen data. Deep learning is a ML technique using multi-layer mathematical operations for learning and inferring on complex data like imagery. This succinct narrative review describes the application, limitations and possible future of AI-based dental diagnostics, treatment planning, and conduct, for example, image analysis, prediction making, record keeping, as well as dental research and discovery. AI-based applications will streamline care, relieving the dental workforce from laborious routine tasks, increasing health at lower costs for a broader population, and eventually facilitate personalized, predictive, preventive, and participatory dentistry. However, AI solutions have not by large entered routine dental practice, mainly due to 1) limited data availability, accessibility, structure, and comprehensiveness, 2) lacking methodological rigor and standards in their development, 3) and practical questions around the value and usefulness of these solutions, but also ethics and responsibility. Any AI application in dentistry should demonstrate tangible value by, for example, improving access to and quality of care, increasing efficiency and safety of services, empowering and enabling patients, supporting medical research, or increasing sustainability. Individual privacy, rights, and autonomy need to be put front and center; a shift from centralized to distributed/federated learning may address this while improving scalability and robustness. Lastly, trustworthiness into, and generalizability of, dental AI solutions need to be guaranteed; the implementation of continuous human oversight and standards grounded in evidence-based dentistry should be expected. Methods to visualize, interpret, and explain the logic behind AI solutions will contribute ("explainable AI"). Dental education will need to accompany the introduction of clinical AI solutions by fostering digital literacy in the future dental workforce.

Ontologies in radiation oncology

Ontologies are a formal, computer-compatible method for representing scientific knowledge about a given domain. They provide a standardized vocabulary, taxonomy and set of relations between concepts. When formatted in a standard way, they can be read and reasoned upon by computers as well as by humans. At the 2019 International Conference on the Use of Computers in Radiation Therapy, there was a session devoted to ontologies in radiation therapy. This paper is a compilation of the material presented, and is meant as an introduction to the subject. This is done by means of a didactic introduction to the topic followed by a series of applications in radiation therapy. The goal of this article is to provide the medical physicist and related professionals with sufficient background that they can understand their construction as well as their practical uses.

Artificial Intelligence: reshaping the practice of radiological sciences in the 21st century

Advances in computing hardware and software platforms have led to the recent resurgence in artificial intelligence (AI) touching almost every aspect of our daily lives by its capability for automating complex tasks or providing superior predictive analytics. AI applications are currently spanning many diverse fields from economics to entertainment, to manufacturing, as well as medicine. Since modern AI's inception decades ago, practitioners in radiological sciences have been pioneering its development and implementation in medicine, particularly in areas related to diagnostic imaging and therapy. In this anniversary article, we embark on a journey to reflect on the learned lessons from past AI's chequered history. We further summarize the current status of AI in radiological sciences, highlighting, with examples, its impressive achievements and effect on re-shaping the practice of medical imaging and radiotherapy in the areas of computer-aided detection, diagnosis, prognosis, and decision support. Moving beyond the commercial hype of AI into reality, we discuss the current challenges to overcome, for AI to achieve its promised hope of providing better precision healthcare for each patient while reducing cost burden on their families and the society at large.

Construct validation of machine learning in the prediction of short-term postoperative complications following total shoulder arthroplasty

BACKGROUND

We aimed to demonstrate that supervised machine learning (ML) models can better predict postoperative complications after total shoulder arthroplasty (TSA) than comorbidity indices.

METHODS

The American College of Surgeons-National Surgical Quality Improvement Program database was queried from 2005-2017 for TSA cases. Training and validation sets were created by randomly assigning 80% and 20% of the data set. Included variables were age, body mass index (BMI), operative time, smoking status, comorbidities, diagnosis, and preoperative hematocrit and albumin. Complications included any adverse event, transfusion, extended length of stay (>3 days), surgical site infection, return to the operating room, deep vein thrombosis or pulmonary embolism, and readmission. Each SML algorithm was compared with one another and to a baseline model using American Society of Anesthesiologists (ASA) classification. Model strength was evaluated by calculating the area under the receiver operating characteristic curve (AUC) and the positive predictive value (PPV) of complications.

RESULTS

We identified a total of 17,119 TSA cases. Mean age, BMI, and length of stay were 69.5 ± 9.6 years, 31.1 ± 6.8, and 2.0 ± 2.2 days. Percentage hematocrit, BMI, and operative time were of highest importance in outcome prediction. SML algorithms outperformed ASA classification models for predicting any adverse event (71.0% vs. 63.0%), transfusion (77.0% vs. 64.0%), extended length of stay (68.0% vs. 60.0%), surgical site infection (65.0% vs. 58.0%), return to the operating room (59.0% vs. 54.0%), and readmission (64.0% vs. 58.0%). SML algorithms demonstrated the greatest PPV for any adverse event (62.5%), extended length of stay (61.4%), transfusion (52.2%), and readmission (10.1%). ASA classification had a 0.0% PPV for complications.

CONCLUSION

With continued validation, intelligent models could calculate patient-specific risk for complications to adjust perioperative care and site of surgery.

RapidPlan knowledge based planning: iterative learning process and model ability to steer planning strategies

Purpose

To determine if the performance of a knowledge based RapidPlan (RP) planning model could be improved with an iterative learning process, i.e. if plans generated by an RP model could be used as new input to re-train the model and achieve better performance.

Methods

Clinical VMAT plans from 83 patients presenting with head and neck cancer were selected to train an RP model, CL-1. With this model, new plans on the same patients were generated, and subsequently used as input to train a novel model, CL-2. Both models were validated on a cohort of 20 patients and dosimetric results compared. Another set of 83 plans was realised on the same patients with different planning criteria, by using a simple template with no attempt to manually improve the plan quality. Those plans were employed to train another model, TP-1. The differences between the plans generated by CL-1 and TP-1 for the validation cohort of patients were compared with respect to the differences between the original plans used to build the two models.

Results

The CL-2 model presented an improvement relative to CL-1, with higher R² values and better regression plots. The mean doses to parallel organs decreased with CL-2, while D_1% to serial organs increased (but not significantly). The different models CL-1 and TP-1 were able to yield plans according to each original strategy.

Conclusion

A refined RP model allowed the generation of plans with improved quality, mostly for parallel organs at risk and, possibly, also the intrinsic model quality.

Discrimination of DNA Methylation Signal from Background Variation for Clinical Diagnostics

Advances in the study of human DNA methylation variation offer a new avenue for the translation of epigenetic research results to clinical applications. Although current approaches to methylome analysis have been helpful in revealing an epigenetic influence in major human diseases, this type of analysis has proven inadequate for the translation of these advances to clinical diagnostics. As in any clinical test, the use of a methylation signal for diagnostic purposes requires the estimation of an optimal cutoff value for the signal, which is necessary to discriminate a signal induced by a disease state from natural background variation. To address this issue, we propose the application of a fundamental signal detection theory and machine learning approaches. Simulation studies and tests of two available methylome datasets from autism and leukemia patients demonstrate the feasibility of this approach in clinical diagnostics, providing high discriminatory power for the methylation signal induced by disease, as well as high classification performance. Specifically, the analysis of whole biomarker genomic regions could suffice for a diagnostic, markedly decreasing its cost.

Using a neural network to predict deviations in mean heart dose during the treatment of left-sided deep inspiration breath hold patients

PURPOSE

We investigated if a neural network could be used to predict the change in mean heart dose when a patient's heart deviates from its planned position during radiotherapy treatment.

METHODS

Predictions were made based on parameters available at the time of treatment planning. The dose prescription, deep inspiration breath-hold (DIBH) amplitude, heart volume, lung volume, V90% and mean heart dose were used to predict the increase in dose to the heart when a shift towards the treatment field was undertaken. The network was trained using 3 mm, 5 mm and 7 mm shifts in heart positions for 50 patients' giving 150 data points in total. The neural network architecture was also varied to find the most optimal network design. The final neural network was then tested using cross-validation to evaluate the model's ability to generalise to new data.

RESULTS

The optimal neural network found was comprised of a single hidden layer of 30 neurons. Based on twenty train/test splits, 94% of all prediction errors were below 0.2 Gy, 97.3% were below 0.3 Gy and 100% were below 0.5 Gy. The average RMSE and maximum prediction error over all train/test splits were 0.13 Gy and 0.5 Gy respectively.

CONCLUSIONS

Our approach using a neural network provides a clinically acceptable estimate of the increase in Mean Heart Dose (MHD), without the need for further imaging, contouring or evaluation. The trained neural network gives clinicians the information and tools required to evaluate what shift in heart position would be acceptable and which scenarios require immediate action before treatment continues.