On the impact of smoothing and noise on robustness of CT and CBCT radiomics features for patients with head and neck cancers

A comparative analysis of preclinical computed tomography radiomics using cone-beam and micro-computed tomography scanners

Background and purpose

Radiomics analysis extracts quantitative data (features) from medical images. These features could potentially reflect biological characteristics and act as imaging biomarkers within precision medicine. However, there is a lack of cross-comparison and validation of radiomics outputs which is paramount for clinical implementation. In this study, we compared radiomics outputs across two computed tomography (CT)-based preclinical scanners.

Materials and methods

Cone beam CT (CBCT) and µCT scans were acquired using different preclinical CT imaging platforms. The reproducibility of radiomics features on each scanner was assessed using a phantom across imaging energies (40 & 60 kVp) and segmentation volumes (44-238 mm3). Retrospective mouse scans were used to compare feature reliability across varying tissue densities (lung, heart, bone), scanners and after voxel size harmonisation. Reliable features had an intraclass correlation coefficient (ICC) > 0.8.

Results

First order and GLCM features were the most reliable on both scanners across different volumes. There was an inverse relationship between tissue density and feature reliability, with the highest number of features in lung (CBCT=580, µCT=734) and lowest in bone (CBCT=110, µCT=560). Comparable features for lung and heart tissues increased when voxel sizes were harmonised. We have identified tissue-specific preclinical radiomics signatures in mice for the lung (133), heart (35), and bone (15).

Conclusions

Preclinical CBCT and µCT scans can be used for radiomics analysis to support the development of meaningful radiomics signatures. This study demonstrates the importance of standardisation and emphasises the need for multi-centre studies.

Developing and validating multi-omics prediction models for late patient-reported dysphagia in head and neck radiotherapy

Background and purpose. To investigate models developed using radiomic and dosiomic (multi-omics) features from planning and treatment imaging for late patient-reported dysphagia in head and neck radiotherapy.Materials and methods. Training (n = 64) and testing (n = 23) cohorts of head and neck cancer patients treated with curative intent chemo-radiotherapy with a follow-up time greater than 12 months were retrospectively examined. Patients completed the MD Anderson Dysphagia Inventory and a composite score ≤60 was interpreted as patient-reported dysphagia. A chart review collected baseline dysphagia and clinical factors. Multi-omic features were extracted from planning and last synthetic CT images using the pharyngeal constrictor muscle contours as a region of interest. Late patient-reported dysphagia models were developed using a random forest backbone, with feature selection and up-sampling methods to account for the imbalanced data. Models were developed and validated for multi-omic feature combinations for both timepoints.Results. A clinical and radiomic feature model developed using the planning CT achieved good performance (validation: sensitivity = 80 ± 27% / balanced accuracy = 71 ± 23%, testing: sensitivity = 80 ± 10% / balanced accuracy = 73 ± 11%). The synthetic CT models did not show improvement over the plan CT multi-omics models, with poor reliability of the radiomic features on these images. Dosiomic features extracted from the synthetic CT showed promise in predicting late patient-reported dysphagia.Conclusion. Multi-omics models can predict late patient-reported dysphagia in head and neck radiotherapy patients. Synthetic CT dosiomic features show promise in developing successful models to account for changes in delivered dose distribution. Multi-center or prospective studies are required prior to clinical implementation of these models.

Radiomic biomarkers of locoregional recurrence: prognostic insights from oral cavity squamous cell carcinoma preoperative CT scans

Introduction

This study aimed to identify CT-based imaging biomarkers for locoregional recurrence (LR) in Oral Cavity Squamous Cell Carcinoma (OSCC) patients.

Methods

Computed tomography scans were collected from 78 patients with OSCC who underwent surgical treatment at a single medical center. We extracted 1,092 radiomic features from gross tumor volume in each patient's pre-treatment CT. Clinical characteristics were also obtained, including race, sex, age, tobacco and alcohol use, tumor staging, and treatment modality. A feature selection algorithm was used to eliminate the most redundant features, followed by a selection of the best subset of the Logistic regression model (LRM). The best LRM model was determined based on the best prediction accuracy in terms of the area under Receiver operating characteristic curve. Finally, significant radiomic features in the final LRM model were identified as imaging biomarkers.

Results and discussion

Two radiomics biomarkers, Large Dependence Emphasis (LDE) of the Gray Level Dependence Matrix (GLDM) and Long Run Emphasis (LRE) of the Gray Level Run Length Matrix (GLRLM) of the 3D Laplacian of Gaussian (LoG σ=3), have demonstrated the capability to preoperatively distinguish patients with and without LR, exhibiting exceptional testing specificity (1.00) and sensitivity (0.82). The group with LRE > 2.99 showed a 3-year recurrence-free survival rate of 0.81, in contrast to 0.49 for the group with LRE ≤ 2.99. Similarly, the group with LDE > 120 showed a rate of 0.82, compared to 0.49 for the group with LDE ≤ 120. These biomarkers broaden our understanding of using radiomics to predict OSCC progression, enabling personalized treatment plans to enhance patient survival.

Identification of CT radiomic features robust to acquisition and segmentation variations for improved prediction of radiotherapy-treated lung cancer patient recurrence

The primary objective of the present study was to identify a subset of radiomic features extracted from primary tumor imaged by computed tomography of early-stage non-small cell lung cancer patients, which remain unaffected by variations in segmentation quality and in computed tomography image acquisition protocol. The robustness of these features to segmentation variations was assessed by analyzing the correlation of feature values extracted from lesion volumes delineated by two annotators. The robustness to variations in acquisition protocol was evaluated by examining the correlation of features extracted from high-dose and low-dose computed tomography scans, both of which were acquired for each patient as part of the stereotactic body radiotherapy planning process. Among 106 radiomic features considered, 21 were identified as robust. An analysis including univariate and multivariate assessments was subsequently conducted to estimate the predictive performance of these robust features on the outcome of early-stage non-small cell lung cancer patients treated with stereotactic body radiation therapy. The univariate predictive analysis revealed that robust features demonstrated superior predictive potential compared to non-robust features. The multivariate analysis indicated that linear regression models built with robust features displayed greater generalization capabilities by outperforming other models in predicting the outcomes of an external validation dataset.

Robustness of radiomic features in 123I-ioflupane-dopamine transporter single-photon emission computer tomography scan

Radiomic features are usually used to predict target variables such as the absence or presence of a disease, treatment response, or time to symptom progression. One of the potential clinical applications is in patients with Parkinson's disease. Robust radiomic features for this specific imaging method have not yet been identified, which is necessary for proper feature selection. Thus, we are assessing the robustness of radiomic features in dopamine transporter imaging (DaT). For this study, we made an anthropomorphic head phantom with tissue heterogeneity using a personal 3D printer (polylactide 82% infill); the bone was subsequently reproduced with plaster. A surgical cotton ball with radiotracer (123I-ioflupane) was inserted. Scans were performed on the two-detector hybrid camera with acquisition parameters corresponding to international guidelines for DaT single photon emission tomography (SPECT). Reconstruction of SPECT was performed on a clinical workstation with iterative algorithms. Open-source LifeX software was used to extract 134 radiomic features. Statistical analysis was made in RStudio using the intraclass correlation coefficient (ICC) and coefficient of variation (COV). Overall, radiomic features in different reconstruction parameters showed a moderate reproducibility rate (ICC = 0.636, p <0.01). Assessment of ICC and COV within CT attenuation correction (CTAC) and non-attenuation correction (NAC) groups and within particular feature classes showed an excellent reproducibility rate (ICC > 0.9, p < 0.01), except for an intensity-based NAC group, where radiomic features showed a good repeatability rate (ICC = 0.893, p <0.01). By our results, CTAC becomes the main threat to feature stability. However, many radiomic features were sensitive to the selected reconstruction algorithm irrespectively to the attenuation correction. Radiomic features extracted from DaT-SPECT showed moderate to excellent reproducibility rates. These results make them suitable for clinical practice and human studies, but awareness of feature selection should be held, as some radiomic features are more robust than others.

Radiomics-enhanced early regression index for predicting treatment response in rectal cancer: a multi-institutional 0.35 T MRI-guided radiotherapy study

PURPOSE

The accurate prediction of treatment response in locally advanced rectal cancer (LARC) patients undergoing MRI-guided radiotherapy (MRIgRT) is essential for optimising treatment strategies. This multi-institutional study aimed to investigate the potential of radiomics in enhancing the predictive power of a known radiobiological parameter (Early Regression Index, ERITCP) to evaluate treatment response in LARC patients treated with MRIgRT.

METHODS

Patients from three international sites were included and divided into training and validation sets. 0.35 T T2*/T1-weighted MR images were acquired during simulation and at each treatment fraction. The biologically effective dose (BED) conversion was used to account for different radiotherapy schemes: gross tumour volume was delineated on the MR images corresponding to specific BED levels and radiomic features were then extracted. Multiple logistic regression models were calculated, combining ERITCP with other radiomic features. The predictive performance of the different models was evaluated on both training and validation sets by calculating the receiver operating characteristic (ROC) curves.

RESULTS

A total of 91 patients was enrolled: 58 were used as training, 33 as validation. Overall, pCR was observed in 25 cases. The model showing the highest performance was obtained combining ERITCP at BED = 26 Gy with a radiomic feature (10th percentile of grey level histogram, 10GLH) calculated at BED = 40 Gy. The area under ROC curve (AUC) of this combined model was 0.98 for training set and 0.92 for validation set, significantly higher (p = 0.04) than the AUC value obtained using ERITCP alone (0.94 in training and 0.89 in validation set).

CONCLUSION

The integration of the radiomic analysis with ERITCP improves the pCR prediction in LARC patients, offering more precise predictive models to further personalise 0.35 T MRIgRT treatments of LARC patients.

Artificial intelligence applied to image-guided radiation therapy (IGRT): a systematic review by the Young Group of the Italian Association of Radiotherapy and Clinical Oncology (yAIRO)

INTRODUCTION

The advent of image-guided radiation therapy (IGRT) has recently changed the workflow of radiation treatments by ensuring highly collimated treatments. Artificial intelligence (AI) and radiomics are tools that have shown promising results for diagnosis, treatment optimization and outcome prediction. This review aims to assess the impact of AI and radiomics on modern IGRT modalities in RT.

METHODS

A PubMed/MEDLINE and Embase systematic review was conducted to investigate the impact of radiomics and AI to modern IGRT modalities. The search strategy was "Radiomics" AND "Cone Beam Computed Tomography"; "Radiomics" AND "Magnetic Resonance guided Radiotherapy"; "Radiomics" AND "on board Magnetic Resonance Radiotherapy"; "Artificial Intelligence" AND "Cone Beam Computed Tomography"; "Artificial Intelligence" AND "Magnetic Resonance guided Radiotherapy"; "Artificial Intelligence" AND "on board Magnetic Resonance Radiotherapy" and only original articles up to 01.11.2022 were considered.

RESULTS

A total of 402 studies were obtained using the previously mentioned search strategy on PubMed and Embase. The analysis was performed on a total of 84 papers obtained following the complete selection process. Radiomics application to IGRT was analyzed in 23 papers, while a total 61 papers were focused on the impact of AI on IGRT techniques.

DISCUSSION

AI and radiomics seem to significantly impact IGRT in all the phases of RT workflow, even if the evidence in the literature is based on retrospective data. Further studies are needed to confirm these tools' potential and provide a stronger correlation with clinical outcomes and gold-standard treatment strategies.

Identification of reproducible radiomic features from on-board volumetric images: A multi-institutional phantom study

BACKGROUND

Radiomics analysis using on-board volumetric images has attracted research attention as a method for predicting prognosis during treatment; however, the lack of standardization is still one of the main concerns.

PURPOSE

This study investigated the factors that influence the reproducibility of radiomic features extracted from on-board volumetric images using an anthropomorphic radiomics phantom. Furthermore, a phantom experiment was conducted with different treatment machines from multiple institutions as external validation to identify reproducible radiomic features.

METHODS

The phantom was designed to be 35 × 20 × 20 cm with eight types of heterogeneous spheres (⌀ = 1, 2, and 3 cm). On-board volumetric images were acquired using 15 treatment machines from eight institutions. Of these, kilovoltage cone-beam computed tomography (kV-CBCT) image data acquired from four treatment machines at one institution were used as an internal evaluation dataset to explore the reproducibility of radiomic features. The remaining image data, including kV-CBCT, megavoltage-CBCT (MV-CBCT), and megavoltage computed tomography (MV-CT) provided by seven different institutions (11 treatment machines), were used as an external validation dataset. A total of 1,302 radiomic features, including 18 first-order, 75 texture, 465 (i.e., 93 × 5) Laplacian of Gaussian (LoG) filter-based, and 744 (i.e., 93 × 8) wavelet filter-based features, were extracted within the spheres. The intraclass correlation coefficient (ICC) was calculated to explore feature repeatability and reproducibility using an internal evaluation dataset. Subsequently, the coefficient of variation (COV) was calculated to validate the feature variability of external institutions. An absolute ICC exceeding 0.85 or COV under 5% was considered indicative of a highly reproducible feature.

RESULTS

For internal evaluation, ICC analysis showed that the median percentage of radiomic features with high repeatability was 95.2%. The ICC analysis indicated that the median percentages of highly reproducible features for inter-tube current, reconstruction algorithm, and treatment machine were decreased by 20.8%, 29.2%, and 33.3%, respectively. For external validation, the COV analysis showed that the median percentage of reproducible features was 31.5%. A total of 16 features, including nine LoG filter-based and seven wavelet filter-based features, were indicated as highly reproducible features. The gray-level run-length matrix (GLRLM) was classified as containing the most frequent features (N = 8), followed by the gray-level dependence matrix (N = 7) and gray-level co-occurrence matrix (N = 1) features.

CONCLUSIONS

We developed the standard phantom for radiomics analysis of kV-CBCT, MV-CBCT, and MV-CT images. With this phantom, we revealed that the differences in the treatment machine and image reconstruction algorithm reduce the reproducibility of radiomic features from on-board volumetric images. Specifically, the most reproducible features for external validation were LoG or wavelet filter-based GLRLM features. However, the acceptability of the identified features should be examined in advance at each institution before applying the findings to prognosis prediction.

Generative adversarial network with radiomic feature reproducibility analysis for computed tomography denoising

BACKGROUND

Most computed tomography (CT) denoising algorithms have been evaluated using image quality analysis (IQA) methods developed for natural image, which do not adequately capture the texture details in medical imaging. Radiomics is an emerging image analysis technique that extracts texture information to provide a more objective basis for medical imaging diagnostics, overcoming the subjective nature of traditional methods. By utilizing the difficulty of reproducing radiomics features under different imaging protocols, we can more accurately evaluate the performance of CT denoising algorithms.

METHOD

We introduced radiomic feature reproducibility analysis as an evaluation metric for a denoising algorithm. Also, we proposed a low-dose CT denoising method based on a generative adversarial network (GAN), which outperformed well-known CT denoising methods.

RESULTS

Although the proposed model produced excellent results visually, the traditional image assessment metrics such as peak signal-to-noise ratio and structural similarity failed to show distinctive performance differences between the proposed method and the conventional ones. However, radiomic feature reproducibility analysis provided a distinctive assessment of the CT denoising performance. Furthermore, radiomic feature reproducibility analysis allowed fine-tuning of the hyper-parameters of the GAN.

CONCLUSION

We demonstrated that the well-tuned GAN architecture outperforms the well-known CT denoising methods. Our study is the first to introduce radiomics reproducibility analysis as an evaluation metric for CT denoising. We look forward that the study may bridge the gap between traditional objective and subjective evaluations in the clinical medical imaging field.

Development and optimisation of a preclinical cone beam computed tomography-based radiomics workflow for radiation oncology research

Background and purpose

Radiomics features derived from medical images have the potential to act as imaging biomarkers to improve diagnosis and predict treatment response in oncology. However, the complex relationships between radiomics features and the biological characteristics of tumours are yet to be fully determined. In this study, we developed a preclinical cone beam computed tomography (CBCT) radiomics workflow with the aim to use in vivo models to further develop radiomics signatures.

Materials and methods

CBCT scans of a mouse phantom were acquired using onboard imaging from a small animal radiotherapy research platform (SARRP, Xstrahl). The repeatability and reproducibility of radiomics outputs were compared across different imaging protocols, segmentation sizes, pre-processing parameters and materials. Robust features were identified and used to compare scans of two xenograft mouse tumour models (A549 and H460).

Results

Changes to the radiomics workflow significantly impact feature robustness. Preclinical CBCT radiomics analysis is feasible with 119 stable features identified from scans imaged at 60 kV, 25 bin width and 0.26 mm slice thickness. Large variation in segmentation volumes reduced the number of reliable radiomics features for analysis. Standardization in imaging and analysis parameters is essential in preclinical radiomics analysis to improve accuracy of outputs, leading to more consistent and reproducible findings.

Conclusions

We present the first optimised workflow for preclinical CBCT radiomics to identify imaging biomarkers. Preclinical radiomics has the potential to maximise the quantity of data captured in in vivo experiments and could provide key information supporting the wider application of radiomics.

Current Applications of Deep Learning and Radiomics on CT and CBCT for Maxillofacial Diseases

The increasing use of computed tomography (CT) and cone beam computed tomography (CBCT) in oral and maxillofacial imaging has driven the development of deep learning and radiomics applications to assist clinicians in early diagnosis, accurate prognosis prediction, and efficient treatment planning of maxillofacial diseases. This narrative review aimed to provide an up-to-date overview of the current applications of deep learning and radiomics on CT and CBCT for the diagnosis and management of maxillofacial diseases. Based on current evidence, a wide range of deep learning models on CT/CBCT images have been developed for automatic diagnosis, segmentation, and classification of jaw cysts and tumors, cervical lymph node metastasis, salivary gland diseases, temporomandibular (TMJ) disorders, maxillary sinus pathologies, mandibular fractures, and dentomaxillofacial deformities, while CT-/CBCT-derived radiomics applications mainly focused on occult lymph node metastasis in patients with oral cancer, malignant salivary gland tumors, and TMJ osteoarthritis. Most of these models showed high performance, and some of them even outperformed human experts. The models with performance on par with human experts have the potential to serve as clinically practicable tools to achieve the earliest possible diagnosis and treatment, leading to a more precise and personalized approach for the management of maxillofacial diseases. Challenges and issues, including the lack of the generalizability and explainability of deep learning models and the uncertainty in the reproducibility and stability of radiomic features, should be overcome to gain the trust of patients, providers, and healthcare organizers for daily clinical use of these models.

Cone-beam CT delta-radiomics to predict genitourinary toxicities and international prostate symptom of prostate cancer patients: a pilot study

For prostate cancer (PCa) patients treated with definitive radiotherapy (RT), acute and late RT-related genitourinary (GU) toxicities adversely impact disease-specific quality of life. Early warning of potential RT toxicities can prompt interventions that may prevent or mitigate future adverse events. During intensity modulated RT (IMRT) of PCa, daily cone-beam computed tomography (CBCT) images are used to improve treatment accuracy through image guidance. This work investigated the performance of CBCT-based delta-radiomic features (DRF) models to predict acute and sub-acute International Prostate Symptom Scores (IPSS) and Common Terminology Criteria for Adverse Events (CTCAE) version 5 GU toxicity grades for 50 PCa patients treated with definitive RT. Delta-radiomics models were built using logistic regression, random forest for feature selection, and a 1000 iteration bootstrapping leave one analysis for cross validation. To our knowledge, no prior studies of PCa have used DRF models based on daily CBCT images. AUC of 0.83 for IPSS and greater than 0.7 for CTCAE grades were achieved as early as week 1 of treatment. DRF extracted from CBCT images showed promise for the development of models predictive of RT outcomes. Future studies will include using artificial intelligence and machine learning to expand CBCT sample sizes available for radiomics analysis.

Impact of image quality on radiomics applications

Radiomics features extracted from medical images have been widely reported to be useful in the patient specific outcome modeling for variety of assessment and prediction purposes. Successful application of radiomics features as imaging biomarkers, however, is dependent on the robustness of the approach to the variation in each step of the modeling workflow. Variation in the input image quality is one of the main sources that impacts the reproducibility of radiomics analysis when a model is applied to broader range of medical imaging data. The quality of medical image is generally affected by both the scanner related factors such as image acquisition/reconstruction settings and the patient related factors such as patient motion. This article aimed to review the published literatures in this field that reported the impact of various imaging factors on the radiomics features through the change in image quality. The literatures were categorized by different imaging modalities and also tabulated based on the imaging parameters and the class of radiomics features included in the study. Strategies for image quality standardization were discussed based on the relevant literatures and recommendations for reducing the impact of image quality variation on the radiomics in multi-institutional clinical trial were summarized at the end of this article.

Lung cancer diagnosis using deep attention-based multiple instance learning and radiomics

BACKGROUND

Early diagnosis of lung cancer is a key intervention for the treatment of lung cancer in which computer-aided diagnosis (CAD) can play a crucial role. Most published CAD methods perform lung cancer diagnosis by classifying each lung nodule in isolation. However, this does not reflect clinical practice, where clinicians diagnose a patient based on a set of images of nodules, instead of looking at one nodule at a time. Besides, the low interpretability of the output provided by these methods presents an important barrier for their adoption.

METHOD

In this article, we treat lung cancer diagnosis as a multiple instance learning (MIL) problem, which better reflects the diagnosis process in the clinical setting and provides higher interpretability of the output. We selected radiomics as the source of input features and deep attention-based MIL as the classification algorithm. The attention mechanism provides higher interpretability by estimating the importance of each instance in the set for the final diagnosis. To improve the model's performance in a small imbalanced dataset, we propose a new bag simulation method for MIL.

RESULTS AND CONCLUSION

The results show that our method can achieve a mean accuracy of 0.807 with a standard error of the mean (SEM) of 0.069, a recall of 0.870 (SEM 0.061), a positive predictive value of 0.928 (SEM 0.078), a negative predictive value of0.591 $0.591$ (SEM 0.155), and an area under the curve (AUC) of 0.842 (SEM 0.074), outperforming other MIL methods. Additional experiments show that the proposed oversampling strategy significantly improves the model's performance. In addition, experiments show that our method provides a good indication of the importance of each nodule in determining the diagnosis, which combined with the well-defined radiomic features, to make the results more interpretable and acceptable for doctors and patients.

Generative models improve radiomics performance in different tasks and different datasets: An experimental study

PURPOSE

Radiomics is an active area of research focusing on high throughput feature extraction from medical images with a wide array of applications in clinical practice, such as clinical decision support in oncology. However, noise in low dose computed tomography (CT) scans can impair the accurate extraction of radiomic features. In this article, we investigate the possibility of using deep learning generative models to improve the performance of radiomics from low dose CTs.

METHODS

We used two datasets of low dose CT scans - NSCLC Radiogenomics and LIDC-IDRI - as test datasets for two tasks - pre-treatment survival prediction and lung cancer diagnosis. We used encoder-decoder networks and conditional generative adversarial networks (CGANs) trained in a previous study as generative models to transform low dose CT images into full dose CT images. Radiomic features extracted from the original and improved CT scans were used to build two classifiers - a support vector machine (SVM) and a deep attention based multiple instance learning model - for survival prediction and lung cancer diagnosis respectively. Finally, we compared the performance of the models derived from the original and improved CT scans.

RESULTS

Denoising with the encoder-decoder network and the CGAN improved the area under the curve (AUC) of survival prediction from 0.52 to 0.57 (p-value < 0.01). On the other hand, the encoder-decoder network and the CGAN improved the AUC of lung cancer diagnosis from 0.84 to 0.88 and 0.89 respectively (p-value < 0.01). Finally, there are no statistically significant improvements in AUC using encoder-decoder networks and CGAN (p-value = 0.34) when networks trained at 75 and 100 epochs.

CONCLUSION

Generative models can improve the performance of low dose CT-based radiomics in different tasks. Hence, denoising using generative models seems to be a necessary pre-processing step for calculating radiomic features from low dose CTs.

Predicting survival in patients with glioblastoma using MRI radiomic features extracted from radiation planning volumes

BACKGROUND

Quantitative image analysis using pre-operative magnetic resonance imaging (MRI) has been able to predict survival in patients with glioblastoma (GBM). The study explored the role of postoperative radiation (RT) planning MRI-based radiomics to predict the outcomes, with features extracted from the gross tumor volume (GTV) and clinical target volume (CTV).

METHODS

Patients with IDH-wildtype GBM treated with adjuvant RT having MRI as a part of RT planning process were included in the study. 546 features were extracted from each GTV and CTV. A LASSO Cox model was applied, and internal validation was performed using leave-one-out cross-validation with overall survival as endpoint. Cross-validated time-dependent area under curve (AUC) was constructed to test the efficacy of the radiomics model, and clinical features were used to generate a combined model. Analysis was done for the entire group and in individual surgical groups-gross total excision (GTR), subtotal resection (STR), and biopsy.

RESULTS

235 patients were included in the study with 57, 118, and 60 in the GTR, STR, and biopsy subgroup, respectively. Using the radiomics model, binary risk groups were feasible in the entire cohort (p < 0.01) and biopsy group (p = 0.04), but not in the other two surgical groups individually. The integrated AUC (iAUC) was 0.613 for radiomics-based classification in the biopsy subgroup, which improved to 0.632 with the inclusion of clinical features.

CONCLUSION

Imaging features extracted from the GTV and CTV regions can lead to risk-stratification of GBM undergoing biopsy, while the utility in other individual subgroups needs to be further explored.

Reproducibility and Repeatability of CBCT-Derived Radiomics Features

PURPOSE

This study was conducted in order to determine the reproducibility and repeatability of cone-beam computed tomography (CBCT) radiomics features.

METHODS

The first-, second-, and fifth-day CBCT images from 10 head and neck (H&N) cancer patients and 10 pelvic cancer patients were retrospectively collected for this study. Eighteen common radiomics features were extracted from the longitudinal CBCT images using two radiomics packages. The reproducibility of CBCT-derived radiomics features was assessed using the first-day image as input and compared across the two software packages. The site-specific intraclass correlation coefficient (ICC) was used to quantitatively assess the agreement between packages. The repeatability of CBCT-based radiomics features was evaluated by comparing the following days of CBCT to the first-day image and quantified using site-specific concordance correlation coefficient (CCC). Furthermore, the correlation with volume for all the features was assessed with linear regression and R 2 as correlation parameters.

RESULTS

The first-order histogram-based features such as skewness and entropy showed good agreement computed in either software package (ICCs ≥ 0.80), while the kurtosis measurements were consistent in H&N patients between the two software tools but not in pelvic cases. The ICCs for GLCM-based features showed good agreement (ICCs ≥ 0.80) between packages in both H&N and pelvic groups except for the GLCM-correction. The GLRLM-based texture features were overall less consistent as calculated by the two different software packages compared with the GLCM-based features. The CCC values of all first-order and second-order GLCM features (except GLCM-energy) were all above 0.80 from the 2-day part test-retest set, while the CCC values all dropped below the cutoff after 5-day treatment scans. All first-order histogram-based and GLCM-texture-based features were not highly correlated with volume, while two GLRLM features, in both H&N and pelvic cohorts, showed R 2 ≥0.8, meaning a high correlation with volume.

CONCLUSION

The reproducibility and repeatability of CBCT-based radiomics features were assessed and compared for the first time on both H&N and pelvic sites. There were overlaps of stable features in both disease sites, yet the overall stability of radiomics features may be disease-/protocol-specific and a function of time between scans.

Generative models improve radiomics reproducibility in low dose CTs: a simulation study

Radiomics is an active area of research in medical image analysis, however poor reproducibility of radiomics has hampered its application in clinical practice. This issue is especially prominent when radiomic features are calculated from noisy images, such as low dose computed tomography (CT) scans. In this article, we investigate the possibility of improving the reproducibility of radiomic features calculated on noisy CTs by using generative models for denoising. Our work concerns two types of generative models-encoder-decoder network (EDN) and conditional generative adversarial network (CGAN). We then compared their performance against a more traditional 'non-local means' denoising algorithm. We added noise to sinograms of full dose CTs to mimic low dose CTs with two levels of noise: low-noise CT and high-noise CT. Models were trained on high-noise CTs and used to denoise low-noise CTs without re-training. We tested the performance of our model in real data, using a dataset of same-day repeated low dose CTs in order to assess the reproducibility of radiomic features in denoised images. EDN and the CGAN achieved similar improvements on the concordance correlation coefficients (CCC) of radiomic features for low-noise images from 0.87 [95%CI, (0.833, 0.901)] to 0.92 [95%CI, (0.909, 0.935)] and for high-noise images from 0.68 [95%CI, (0.617, 0.745)] to 0.92 [95%CI, (0.909, 0.936)], respectively. The EDN and the CGAN improved the test-retest reliability of radiomic features (mean CCC increased from 0.89 [95%CI, (0.881, 0.914)] to 0.94 [95%CI, (0.927, 0.951)]) based on real low dose CTs. These results show that denoising using EDN and CGANs could be used to improve the reproducibility of radiomic features calculated from noisy CTs. Moreover, images at different noise levels can be denoised to improve the reproducibility using the above models without need for re-training, provided the noise intensity is not excessively greater that of the high-noise CTs. To the authors' knowledge, this is the first effort to improve the reproducibility of radiomic features calculated on low dose CT scans by applying generative models.

Statistical harmonization can improve the development of a multicenter CT-based radiomic model predictive of nonresponse to induction chemotherapy in laryngeal cancers

PURPOSE

To develop a radiomic model predicting nonresponse to induction chemotherapy in laryngeal cancers, from multicenter pretherapeutic contrast-enhanced computed tomography (CE-CT) and evaluate the benefit of feature harmonization in such a context.

METHODS

Patients (n = 104) eligible for laryngeal preservation chemotherapy were included in five centers. Primary tumor was manually delineated on the CE-CT images. The following radiomic features were extracted with an in-house software (MIRAS v1.1, LaTIM UMR 1101): intensity, shape, and textural features derived from Gray-Level Co-occurrence Matrix (GLCM), Neighborhood Gray Tone Difference Matrix (NGTDM), Gray-Level Run Length Matrix (GLRLM), and Gray-Level Size Zone Matrix (GLSZM). Harmonization was performed using ComBat after unsupervised hierarchical clustering, used to determine labels automatically, given the high heterogeneity of imaging characteristics across and within centers. Patients with similar feature distributions were grouped with unsupervised clustering into an optimal number of clusters (2) determined with "silhouette scoring." Statistical harmonization was then carried out with ComBat on these 2 identified clusters. The cohort was split into training/validation (n = 66) and testing (n = 32) sets. Area under the receiver operating characteristics curves (AUC) were used to evaluate the ability of radiomic features (before and after harmonization) to predict nonresponse to chemotherapy, and specificity (Sp) and sensitivity (Se) were used to quantify their performance in the testing set.

RESULTS

Without harmonization, none of the features identified as predictive in the training set remained significant in the testing set. After ComBat, one textural feature identified in the training set keeps a predictive trend in the testing set-Zone Percentage, derived from the GLSZM, was predictive of nonresponse in the training set (AUC = 0.62, Se = 70%, Sp = 64%, P = 0.04) and obtained a satisfactory performance in the testing set (Se = 80%, Sp = 67%, P = 0.03), although significance was limited by the size of the testing set. These results are consistent with previously published findings in head and neck cancers.

CONCLUSIONS

Radiomic features from CE-CT could help in the selection of patients for induction chemotherapy in laryngeal cancers, with relatively good sensitivity and specificity in predicting lack of response. Statistical harmonization with ComBat and unsupervised clustering seems to improve the predictive value of features extracted in such a heterogeneous multicenter setting.

Delta radiomics for rectal cancer response prediction using low field magnetic resonance guided radiotherapy: an external validation

INTRODUCTION

A recent study performed on 16 locally advanced rectal cancer (LARC) patients treated using magnetic resonance guided radiotherapy (MRgRT) has identified two delta radiomics features as predictors of clinical complete response (cCR) after neoadjuvant radio-chemotherapy (nCRT). This study aims to validate these features (ΔL_least and Δglnu) on an external larger dataset, expanding the analysis also for pathological complete response (pCR) prediction.

METHODS

A total of 43 LARC patients were enrolled: Gross Tumour Volume (GTV) was delineated on T2/T1* MR images acquired during MRgRT and the two delta features were calculated. Receiver Operating Characteristic (ROC) curve analysis was performed on the 16 cases of the original study and the best cut-off value was identified. The performance of ΔL_least and Δglnu was evaluated at the best cut-off value.

RESULTS

On the original dataset of 16 patients, ΔL_least reported an AUC of 0.81 for cCR and 0.93 for pCR, while Δglnu 0.72 and 0.54 respectively. The best cut-off values of ΔL_least was 0.73 for both outcomes, while Δglnu reported 0.54 for cCR and 0.93 for pCR. At the external validation, ΔL_least showed an accuracy of 81% for cCR and 79% for pCR, while Δglnu reported 63% for cCR and 40% for pCR.

CONCLUSION

The accuracy of ΔL_least in predicting cCR and pCR is significantly higher than those obtained considering Δglnu, but inferior if compared with other image-based biomarker, such as the early-regression index. Studies with larger cohorts of patients are recommended to further investigate the role of delta radiomic features in MRgRT.

Diagnostic Accuracy and Failure Mode Analysis of a Deep Learning Algorithm for the Detection of Intracranial Hemorrhage

OBJECTIVE

To determine the institutional diagnostic accuracy of an artificial intelligence (AI) decision support systems (DSS), Aidoc, in diagnosing intracranial hemorrhage (ICH) on noncontrast head CTs and to assess the potential generalizability of an AI DSS.

METHODS

This retrospective study included 3,605 consecutive, emergent, adult noncontrast head CT scans performed between July 1, 2019, and December 30, 2019, at our institution (51% female subjects, mean age of 61 ± 21 years). Each scan was evaluated for ICH by both a certificate of added qualification certified neuroradiologist and Aidoc. We determined the diagnostic accuracy of the AI model and performed a failure mode analysis with quantitative CT radiomic image characterization.

RESULTS

Of the 3,605 scans, 349 cases of ICH (9.7% of studies) were identified. The neuroradiologist and Aidoc interpretations were concordant in 96.9% of cases and the overall sensitivity, specificity, positive predictive value, and negative predictive value were 92.3%, 97.7%, 81.3%, and 99.2%, respectively, with positive predictive values unexpectedly lower than in previously reported studies. Prior neurosurgery, type of ICH, and number of ICHs were significantly associated with decreased model performance. Quantitative image characterization with CT radiomics failed to reveal significant differences between concordant and discordant studies.

DISCUSSION

This study revealed decreased diagnostic accuracy of an AI DSS at our institution. Despite extensive evaluation, we were unable to identify the source of this discrepancy, raising concerns about the generalizability of these tools with indeterminate failure modes. These results further highlight the need for standardized study design to allow for rigorous and reproducible site-to-site comparison of emerging deep learning technologies.

Repeatability of CBCT radiomic features and their correlation with CT radiomic features for prostate cancer

PURPOSE

Radiomic features of cone-beam CT (CBCT) images have potential as biomarkers to predict treatment response and prognosis for patients of prostate cancer. Previous studies of radiomic feature analysis for prostate cancer were assessed in a variety of imaging modalities, including MRI, PET, and CT, but usually limited to a pretreatment setting. However, CBCT images may provide an opportunity to capture early morphological changes to the tumor during treatment that could lead to timely treatment adaptation. This work investigated the quality of CBCT-based radiomic features and their relationship with reconstruction methods applied to the CBCT projections and the preprocessing methods used in feature extraction. Moreover, CBCT features were correlated with planning CT (pCT) features to further assess the viability of CBCT radiomic features.

METHODS

The quality of 42 CBCT-based radiomic features was assessed according to their repeatability and reproducibility. Repeatability was quantified by correlating radiomic features between 20 CBCT scans that also had repeated scans within 15 minutes. Reproducibility was quantified by correlating radiomic features between the planning CT (pCT) and the first fraction CBCT for 20 patients. Concordance correlation coefficients (CCC) of radiomic features were used to estimate the repeatability and reproducibility of radiomic features. The same patient dataset was assessed using different reconstruction methods applied to the CBCT projections. CBCT images were generated using 18 reconstruction methods using iterative (iCBCT) and standard (sCBCT) reconstructions, three convolution filters, and five noise suppression filters. Eighteen preprocessing settings were also considered.

RESULTS

Overall, CBCT radiomic features were more repeatable than reproducible. Five radiomic features are repeatable in > 97% of the reconstruction and preprocessing methods, and come from the gray-level size zone matrix (GLSZM), neighborhood gray-tone difference matrix (NGTDM), and gray-level-run length matrix (GLRLM) radiomic feature classes. These radiomic features were reproducible in > 9.8% of the reconstruction and preprocessing methods. Noise suppression and convolution filter smoothing increased radiomic features repeatability, but decreased reproducibility. The top-repeatable iCBCT method (iCBCT-Sharp-VeryHigh) is more repeatable than the top-repeatable sCBCT method (sCBCT-Smooth) in 64% of the radiomic features.

CONCLUSION

Methods for reconstruction and preprocessing that improve CBCT radiomic feature repeatability often decrease reproducibility. The best approach may be to use methods that strike a balance repeatability and reproducibility such as iCBCT-Sharp-VeryLow-1-Lloyd-256 that has 17 repeatable and eight reproducible radiomic features. Previous radiomic studies that only used pCT radiomic features have generated prognostic models of prostate cancer outcome. Since our study indicates that CBCT radiomic features correlated well with a subset of pCT radiomic features, one may expect CBCT radiomics to also generate prognostic models for prostate cancer.

Influence of inter-observer delineation variability on radiomic features of the parotid gland

PURPOSE

This study aimed to quantify the variability in the values of radiomic features extracted from a right parotid gland (RPG) delineated by a series of independent observers.

METHODS

This was a secondary analysis of anonymous data from a delineation workshop. Inter-observer variability of the RPG from 40 participants was quantified using DICE similarity coefficient (DSC) and Hausdorff distance (HD). An additional contour was generated using Varian SmartSegmentation. Radiomic features extracted include four shape features, six histogram features, and 32 texture features. The absolute mean paired percentage difference (PPD) in feature values from the expert and participants were ranked . Feature robustness was classified using pre- determined thresholds.

RESULTS

63% of participants achieved a DSC > 0.7, the auto- segmentation DSC was 0.76. The average HD for the participants was 16.16 mm ± 0.66 mm, and 15.16 mm for the auto-segmentation. 48% (n = 20) and 33% (n = 14) of features were deemed to be robust with a mean absolute PPD < 5%, for the auto-segmentation and manual delineations respectively; the majority of which were from the grey-run length matrix family. 7% (n = 3) of features from the auto- segmentation and 10% (n = 4) from the manual contours were deemed to be unstable with a mean absolute PPD > 50%. The value of the most robust feature was not related to DSC and HD.

CONCLUSION

Inter-observer delineation variability affects the value of the radiomic features extracted from the RPG. This study identifies the radiomic features least sensitive to these uncertainties. Further investigation of the clinical relevance of these features in prediction of xerostomia is warranted.

Machine and deep learning methods for radiomics

Radiomics is an emerging area in quantitative image analysis that aims to relate large-scale extracted imaging information to clinical and biological endpoints. The development of quantitative imaging methods along with machine learning has enabled the opportunity to move data science research towards translation for more personalized cancer treatments. Accumulating evidence has indeed demonstrated that noninvasive advanced imaging analytics, that is, radiomics, can reveal key components of tumor phenotype for multiple three-dimensional lesions at multiple time points over and beyond the course of treatment. These developments in the use of CT, PET, US, and MR imaging could augment patient stratification and prognostication buttressing emerging targeted therapeutic approaches. In recent years, deep learning architectures have demonstrated their tremendous potential for image segmentation, reconstruction, recognition, and classification. Many powerful open-source and commercial platforms are currently available to embark in new research areas of radiomics. Quantitative imaging research, however, is complex and key statistical principles should be followed to realize its full potential. The field of radiomics, in particular, requires a renewed focus on optimal study design/reporting practices and standardization of image acquisition, feature calculation, and rigorous statistical analysis for the field to move forward. In this article, the role of machine and deep learning as a major computational vehicle for advanced model building of radiomics-based signatures or classifiers, and diverse clinical applications, working principles, research opportunities, and available computational platforms for radiomics will be reviewed with examples drawn primarily from oncology. We also address issues related to common applications in medical physics, such as standardization, feature extraction, model building, and validation.

Technical Note: ROdiomiX: A validated software for radiomics analysis of medical images in radiation oncology

PURPOSE

This study introduces an in-house-designed software platform (ROdiomiX) for the radiomics analysis of medical images in radiation oncology. ROdiomiX is a MATLAB-based framework for the computation of radiomic features and feature aggregation techniques in compliance with the Image-Biomarker-Standardization-Initiative (IBSI) guidelines, which includes preprocessing protocols and quantitative benchmark results for analysis of computational phantom images.

METHODS AND MATERIALS

The ROdiomiX software system consists of a series of computation cores implemented on the basis of the guidelines proposed by the IBSI. It is capable of quantitative computation of the following 11 different feature categories: Local-Intensity, Intensity-Histogram, Intensity-Based-Statistical, Intensity-Volume-Histogram, Gray-Level-Co-occurrence, Gray-Level-Run-Length, Gray-Level-Size-Zone, Gray-Level-Distance-Zone, Neighborhood-Grey-Tone-Difference, Neighboring-Grey-Level-Dependence, and Morphological feature. ROdiomiX was validated against benchmark values for the IBSI 3D digital phantom, as well as one designed in-house (HFH). The intraclass correlation coefficient (ICC) for estimating the degree of absolute agreement between ROdiomiX computation and benchmark values for different features at the 95% confidence level (CL) was used for comparison.

RESULTS

Among the 11 feature categories with 149 total features including 10 different feature aggregation methods (following the IBSI guidelines), the percent difference between absolute feature values computed by the ROdiomiX software and benchmark values reported for IBSI and HFH digital phantoms were 0.14% + 0.43% and 0.11% + 0.27%, respectively. The ICC values were >0.997 for all ten feature categories for both the IBSI and HFH digital phantoms.

CONCLUSION

The authors successfully developed a platform for computation of quantitative radiomic features. The image preprocessing and computational software cores were designed following the procedures specified by the IBSI. Benchmarking testing was in excellent agreement against the IBSI- and HFH-designed computational phantoms.

The role of radiomics in prostate cancer radiotherapy

"Radiomics," as it refers to the extraction and analysis of a large number of advanced quantitative radiological features from medical images using high-throughput methods, is perfectly suited as an engine for effectively sifting through the multiple series of prostate images from before, during, and after radiotherapy (RT). Multiparametric (mp)MRI, planning CT, and cone beam CT (CBCT) routinely acquired throughout RT and the radiomics pipeline are developed for extraction of thousands of variables. Radiomics data are in a format that is appropriate for building descriptive and predictive models relating image features to diagnostic, prognostic, or predictive information. Prediction of Gleason score, the histopathologic cancer grade, has been the mainstay of the radiomic efforts in prostate cancer. While Gleason score (GS) is still the best predictor of treatment outcome, there are other novel applications of quantitative imaging that are tailored to RT. In this review, we summarize the radiomics efforts and discuss several promising concepts such as delta-radiomics and radiogenomics for utilizing image features for assessment of the aggressiveness of prostate cancer and its outcome. We also discuss opportunities for quantitative imaging with the advance of instrumentation in MRI-guided therapies.

Optimal co-clinical radiomics: Sensitivity of radiomic features to tumour volume, image noise and resolution in co-clinical T1-weighted and T2-weighted magnetic resonance imaging

Background

Radiomics analyses has been proposed to interrogate the biology of tumour as well as to predict/assess response to therapy in vivo. The objective of this work was to assess the sensitivity of radiomics features to noise, resolution, and tumour volume in the context of a co-clinical trial.

Methods

Triple negative breast cancer (TNBC) patients were recruited into an ongoing co-clinical imaging trial. Sub-typed matched TNBC patient-derived tumour xenografts (PDX) were generated to investigate optimal co-clinical MR radiomic features. The MR imaging protocol included T1-weighed and T2-weighted imaging. To test the sensitivity of radiomics to resolution, PDX were imaged at three different resolutions. Multiple sets of images with varying signal-to-noise ratio (SNR) were generated, and an image independent patch-based method was implemented to measure the noise levels. Forty-eight radiomic features were extracted from manually segmented 2D and 3D segmented tumours and normal tissues of T1- and T2- weighted co-clinical MR images.

Findings

Sixteen radiomics features were identified as volume dependent and corrected for volume-dependency following normalization. Features from grey-level run-length matrix (GLRLM), grey-level size zone matrix (GLSZM) were identified as most sensitive to noise. Radiomic features Kurtosis and Run-length variance (RLV) from GLSZM were most sensitive to changes in resolution in both T1w and T2w MRI. In general, 3D radiomic features were more robust compared to 2D (single slice) measures, although the former exhibited higher variability between subjects.

Interpretation

Tumour volume, noise characteristics, and image resolution significantly impact radiomic analysis in co-clinical studies.

Radiomics in medical imaging-"how-to" guide and critical reflection

Radiomics is a quantitative approach to medical imaging, which aims at enhancing the existing data available to clinicians by means of advanced mathematical analysis. Through mathematical extraction of the spatial distribution of signal intensities and pixel interrelationships, radiomics quantifies textural information by using analysis methods from the field of artificial intelligence. Various studies from different fields in imaging have been published so far, highlighting the potential of radiomics to enhance clinical decision-making. However, the field faces several important challenges, which are mainly caused by the various technical factors influencing the extracted radiomic features.The aim of the present review is twofold: first, we present the typical workflow of a radiomics analysis and deliver a practical "how-to" guide for a typical radiomics analysis. Second, we discuss the current limitations of radiomics, suggest potential improvements, and summarize relevant literature on the subject.

Radiomics Approach Outperforms Diameter Criteria for Predicting Pathological Lateral Lymph Node Metastasis After Neoadjuvant (Chemo)Radiotherapy in Advanced Low Rectal Cancer

BACKGROUND

Advanced low rectal cancer has a non-negligible risk of lateral pelvic lymph node (LPLN) metastasis (LPLNM) and lateral local recurrence (LR) after neoadjuvant (chemo)radiotherapy and total mesorectal excision. LPLN dissection (LPLND) reduces LR but increases postoperative complications and sexual/urinary dysfunction.

OBJECTIVE

The aim of this study was to develop a new radiomics-based prediction model for LPLNM in patients with rectal cancer.

METHODS

A total of 247 patients with rectal cancer and enlarged LPLNs treated by (chemo)radiotherapy and LPLND were enrolled in this retrospective, multicenter study. LPLN radiomic features were extracted from pretreatment portal venous-phase computed tomography images. A radiomics score of LPLN was constructed based on the least absolute shrinkage and selection operator regression in a primary cohort of 175 patients. Model performance was assessed in terms of discrimination, calibration, and decision curve analysis, and was externally validated in 72 patients.

RESULTS

The radiomics score showed significantly better discrimination compared with pretreatment short-axis diameter measurements in both the primary (area under the curve [AUC] 0.91 vs. 0.83, p = 0.0015) and validation (AUC 0.90 vs. 0.80, p = 0.0298) cohorts. Decision curve analysis also indicated the superiority of the radiomics score. In a subanalysis of patients with a short-axis diameter ≥ 7 mm, the radiomics nomogram, incorporating the radiomics score and LPLN shrinkage to ≤ 4 mm, had better discrimination compared with a model incorporating only LPLN shrinkage in both cohorts.

CONCLUSIONS

Radiomics-based prediction modeling provides individualized risk estimation of LPLNM in rectal cancer patients treated with (chemo)radiotherapy, and outperforms measurements of pretreatment LPLN diameter.

External Validation of Early Regression Index (ERITCP) as Predictor of Pathologic Complete Response in Rectal Cancer Using Magnetic Resonance-Guided Radiation Therapy

PURPOSE

Tumor control probability (TCP)-based early regression index (ERI_TCP) is a radiobiological parameter that showed promising results in predicting pathologic complete response (pCR) on T2-weighted 1.5 T magnetic resonance (MR) images of patients with locally advanced rectal cancer. This study aims to validate the ERI_TCP in the context of low-tesla MR-guided radiation therapy, using images acquired with different magnetic field strength (0.35 T) and image contrast (T2/T1). Furthermore, the optimal timing for pCR prediction was estimated, calculating the ERI index at different biologically effective dose (BED) levels.

METHODS AND MATERIALS

Fifty-two patients with locally advanced rectal cancer treated with neoadjuvant chemoradiation therapy were enrolled in this multi-institutional retrospective study. For each patient, a 0.35 T T2/T1-weighted MR image was acquired during simulation and on each treatment day. Gross tumor volume was contoured according to International Commission on Radiation Units Report 83 guidelines. According to the original definition, ERI_TCP was calculated considering the residual tumor volume at BED = 25 Gy. ERI was also calculated in correspondence with several BED levels: 13, 21, 32, 40, 46, 54, 59, and 67. The predictive performance of the different ERI indices were evaluated in terms of receiver operating characteristic curve. The robustness of ERI_TCP with respect to the interobserver variability was also evaluated considering 2 operators and calculating the intraclass correlation index.

RESULTS

Fourteen patients showed pCR. ERI_TCP correctly 47 of 52 cases (accuracy = 90%), showing good results in terms of sensitivity (86%), specificity (92%), negative predictive value (95%), and positive predictive value (80%). The analysis at different BED levels shows that the best predictive performance is obtained when this parameter is calculated at BED = 25 Gy (area under the curve = 0.93). ERI_TCP results are robust with respect to interobserver variability (intraclass correlation index = 0.99).

CONCLUSIONS

This study confirmed the validity and the robustness of ERI_TCP as a pCR predictor in the context of low-tesla MR-guided radiation therapy and indicate 25 Gy as the best BED level to perform predictions.

A Literature Review of the Potential Diagnostic Biomarkers of Head and Neck Neoplasms

Head and neck neoplasms have a poor prognosis because of their late diagnosis. Finding a biomarker to detect these tumors in an early phase could improve the prognosis and survival rate. This literature review provides an overview of biomarkers, covering the different -omics fields to diagnose head and neck neoplasms in the early phase. To date, not a single biomarker, nor a panel of biomarkers for the detection of head and neck tumors has been detected with clinical applicability. Limitations for the clinical implementation of the investigated biomarkers are mainly the heterogeneity of the study groups (e.g., small population in which the biomarker was tested, and/or only including high-risk populations) and a low sensitivity and/or specificity of the biomarkers under study. Further research on biomarkers to diagnose head and neck neoplasms in an early stage, is therefore needed.

Applications of radiomics in precision diagnosis, prognostication and treatment planning of head and neck squamous cell carcinomas

Recent advancements in computational power, machine learning, and artificial intelligence technology have enabled automated evaluation of medical images to generate quantitative diagnostic and prognostic biomarkers. Such objective biomarkers are readily available and have the potential to improve personalized treatment, precision medicine, and patient selection for clinical trials. In this article, we explore the merits of the most recent addition to the “-omics” concept for the broader field of head and neck cancer – “Radiomics”. This review discusses radiomics studies focused on (molecular) characterization, classification, prognostication and treatment guidance for head and neck squamous cell carcinomas (HNSCC). We review the underlying hypothesis, general concept and typical workflow of radiomic analysis, and elaborate on current and future challenges to be addressed before routine clinical application.

Reliability of CT radiomic features reflecting tumour heterogeneity according to image quality and image processing parameters

The reliability of radiomics features (RFs) is crucial for quantifying tumour heterogeneity. We assessed the influence of imaging, segmentation, and processing conditions (quantization range, bin number, signal-to-noise ratio [SNR], and unintended outliers) on RF measurement. Low SNR and unintended outliers increased the standard deviation and mean values of histograms to calculate the first-order RFs. Variations in imaging processing conditions significantly altered the shape of the probability distribution (centre of distribution, extent of dispersion, and segmentation of probability clusters) in second-order RF matrices (i.e. grey-level co-occurrence and grey-level run length), thereby eventually causing fluctuations in RF estimation. Inconsistent imaging and processing conditions decreased the number of reliably measured RFs in terms of individual RF values (intraclass correlation coefficient ≥0.75) and inter-lesion RF ratios (coefficient of variation <15%). No RF could be reliably estimated under inconsistent SNR and inclusion of outlier conditions. By contrast, with high SNR and no outliers, all first-order RFs, 11 (42%) grey-level co-occurrence RFs and five (42%) grey-level run length RFs showed acceptable reliability. Our study suggests that optimization of SNR, exclusion of outliers, and application of relevant quantization range and bin number should be performed to ensure the robustness of radiomics studies assessing tumor heterogeneity.

Clustering of Largely Right-Censored Oropharyngeal Head and Neck Cancer Patients for Discriminative Groupings to Improve Outcome Prediction

Clustering is the task of identifying groups of similar subjects according to certain criteria. The AJCC staging system can be thought as a clustering mechanism that groups patients based on their disease stage. This grouping drives prognosis and influences treatment. The goal of this work is to evaluate the efficacy of machine learning algorithms to cluster the patients into discriminative groups to improve prognosis for overall survival (OS) and relapse free survival (RFS) outcomes. We apply clustering over a retrospectively collected data from 644 head and neck cancer patients including both clinical and radiomic features. In order to incorporate outcome information into the clustering process and deal with the large proportion of censored samples, the feature space was scaled using the regression coefficients fitted using a proxy dependent variable, martingale residuals, instead of follow-up time. Two clusters were identified and evaluated using cross validation. The Kaplan Meier (KM) curves between the two clusters differ significantly for OS and RFS (p-value < 0.0001). Moreover, there was a relative predictive improvement when using the cluster label in addition to the clinical features compared to using only clinical features where AUC increased by 5.7% and 13.0% for OS and RFS, respectively.

Differential Diagnosis of Nonhypervascular Pancreatic Neuroendocrine Neoplasms From Pancreatic Ductal Adenocarcinomas, Based on Computed Tomography Radiological Features and Texture Analysis

RATIONALE AND OBJECTIVES

To determine computed tomography (CT) radiological features and texture features that are rewarding in differentiating nonhypervascular pancreatic neuroendocrine neoplasms (PNENs) from pancreatic ductal adenocarcinomas (PDACs).

MATERIALS AND METHODS

We compared patients to pathologically proven nonhypervascular PNENs and age-matched controls with pathologically proven PDACs in a 1:2 ratio. Preoperative CT images in the arterial phase (AP) and portal vein phase (PVP) were obtained. Two radiologists independently reviewed the morphological characteristics of each tumor. Three-dimensional regions of interest (ROIs), drawn using ITK-SNAP software, were input into AK software (Artificial Intelligent Kit, GE) to extract texture features from AP and PVP images. Differences between PNENs and PDACs were analyzed with the chi-squared test, least absolute shrinkage and selection operator, kappa statistics, and uni- and multivariate logistic regression analyses.

RESULTS

In total, 40 nonhypervascular PNENs and 80 PDACs were evaluated. Maximum diameter on axial section, margin, calcification, vascularity in the tumor, and tumor heterogeneity were significantly different between PDACs and nonhypervascular PNENs. Multivariate analysis showed well-defined tumor margin (odds ratio: 21.0) and presence of calcification (odds ratio: 4.4) were significant predictors of nonhypervascular PNENs. The area under the receiver operating characteristic curve of the radiological feature model, AP texture model, and PVP texture model were 0.780, 0.855, and 0.929, respectively, based on logistic regression.

CONCLUSION

A well-defined margin and calcification in the tumor were helpful in discriminating nonhypervascular PNENs from PDACs. Texture analysis of contrast-enhanced CT images could be beneficial in differentially diagnosing nonhypervascular PNENs and PDACs.

Radiomics: from qualitative to quantitative imaging

Historically, medical imaging has been a qualitative or semi-quantitative modality. It is difficult to quantify what can be seen in an image, and to turn it into valuable predictive outcomes. As a result of advances in both computational hardware and machine learning algorithms, computers are making great strides in obtaining quantitative information from imaging and correlating it with outcomes. Radiomics, in its two forms "handcrafted and deep," is an emerging field that translates medical images into quantitative data to yield biological information and enable radiologic phenotypic profiling for diagnosis, theragnosis, decision support, and monitoring. Handcrafted radiomics is a multistage process in which features based on shape, pixel intensities, and texture are extracted from radiographs. Within this review, we describe the steps: starting with quantitative imaging data, how it can be extracted, how to correlate it with clinical and biological outcomes, resulting in models that can be used to make predictions, such as survival, or for detection and classification used in diagnostics. The application of deep learning, the second arm of radiomics, and its place in the radiomics workflow is discussed, along with its advantages and disadvantages. To better illustrate the technologies being used, we provide real-world clinical applications of radiomics in oncology, showcasing research on the applications of radiomics, as well as covering its limitations and its future direction.

Application of radiomics for the prediction of HPV status for patients with head and neck cancers

PURPOSE

To perform radiomic analysis of primary tumors extracted from pretreatment contrast-enhanced computed tomography (CE-CT) images for patients with oropharyngeal cancers to identify discriminant features and construct an optimal classifier for the characterization and prediction of human papilloma virus (HPV) status.

MATERIALS AND METHODS

One hundred and eighty seven patients with oropharyngeal cancers with known HPV status (confirmed by immunohistochemistry-p16 protein testing) were retrospectively studied as follows: Group A: 95 patients (19HPV- and 76HPV+) from the MICAII grand challenge. Group B: 92 patients (52HPV- and 40HPV+) from our institution. Radiomic features (172) were extracted from pretreatment diagnostic CE-CT images of the gross tumor volume (GTV). Levene and Kolmogorov-Smirnov's tests with absolute biserial correlation (>0.48) were used to identify the discriminant features between the HPV+ and HPV- groups. The discriminant features were used to train and test eight different classifiers. Area under receiver operating characteristic (AUC), positive predictive and negative predictive values (PPV and NPV, respectively) were used to evaluate the performance of the classifiers. Principal component analysis (PCA) was applied on the discriminant feature set and seven PCs were used to train and test a generalized linear model (GLM) classifier.

RESULTS

Among 172 radiomic features only 12 radiomic features (from 3 categories) were significantly different (P < 0.05, |BSC| > 0.48) between the HPV+ and HPV- groups. Among the eight classifiers trained and applied for prediction of HPV status, the GLM showed the highest performance for each discriminant feature and the combined 12 features: AUC/PPV/NPV = 0.878/0.834/0.811. The GLM high prediction power was AUC/PPV/NPV = 0.849/0.731/0.788 and AUC/PPV/NPV = 0.869/0.807/0.870 for unseen test datasets for groups A and B, respectively. After eliminating the correlation among discriminant features by applying PCA analysis, the performance of the GLM was improved by 3.3%, 2.2%, and 1.8% for AUC, PPV, and NPV, respectively.

CONCLUSIONS

Results imply that GTV's for HPV+ patients exhibit higher intensities, smaller lesion size, greater sphericity/roundness, and higher spatial intensity-variation/heterogeneity. Results are suggestive that radiomic features primarily associated with the spatial arrangement and morphological appearance of the tumor on contrast-enhanced diagnostic CT datasets may be potentially used for classification of HPV status.

Impact of image preprocessing methods on reproducibility of radiomic features in multimodal magnetic resonance imaging in glioblastoma

To investigate the effect of image preprocessing, in respect to intensity inhomogeneity correction and noise filtering, on the robustness and reproducibility of the radiomics features extracted from the Glioblastoma (GBM) tumor in multimodal MR images (mMRI). In this study, for each patient 1461 radiomics features were extracted from GBM subregions (i.e., edema, necrosis, enhancement, and tumor) of mMRI (i.e., FLAIR, T1, T1C, and T2) volumes for five preprocessing combinations (in total 116 880 radiomics features). The robustness and reproducibility of the radiomics features were assessed under four comparisons: (a) Baseline versus modified bias field; (b) Baseline versus modified bias field followed by noise filtering; (c) Baseline versus modified noise, and (d) Baseline versus modified noise followed bias field correction. The concordance correlation coefficient (CCC), dynamic range (DR), and interclass correlation coefficient (ICC) were used as metrics. Shape features and subsequently, local binary pattern (LBP) filtered images were highly stable and reproducible against bias field correction and noise filtering in all measurements. In all MRI modalities, necrosis regions (NC: n ® ~449/1461, 30%) had the highest number of highly robust features, with CCC and DR >= 0.9, in comparison with edema (ED: n ® ~296/1461, 20%), enhanced (EN: n ® ~ 281/1461, 19%) and active‐tumor regions (TM: n ® ~254/1461, 17%). The necrosis regions (NC: n¯ ~ 449/1461, 30%) had a higher number of highly robust features (CCC and DR >= 0.9) than edema (ED: n¯ ~ 296/1461, 20%), enhanced (EN: n¯ ~ 281/1461, 19%) and active‐tumor (TM: n¯ ~ 254/1461, 17%) regions across all modalities. Furthermore, our results identified that the percentage of high reproducible features with ICC >= 0.9 after bias field correction (23.2%), and bias field correction followed by noise filtering (22.4%) were higher in contrast with noise smoothing and also noise smoothing follow by bias correction. These preliminary findings imply that preprocessing sequences can also have a significant impact on the robustness and reproducibility of mMRI‐based radiomics features and identification of generalizable and consistent preprocessing algorithms is a pivotal step before imposing radiomics biomarkers into the clinic for GBM patients.

The effects of volume of interest delineation on MRI-based radiomics analysis: evaluation with two disease groups

BACKGROUND

Manual delineation of volume of interest (VOI) is widely used in current radiomics analysis, suffering from high variability. The tolerance of delineation differences and possible influence on each step of radiomics analysis are not clear, requiring quantitative assessment. The purpose of our study was to investigate the effects of delineation of VOIs on radiomics analysis for the preoperative prediction of metastasis in nasopharyngeal carcinoma (NPC) and sentinel lymph node (SLN) metastasis in breast cancer.

METHODS

This study retrospectively enrolled two datasets (NPC group: 238 cases; SLN group: 146 cases). Three operations, namely, erosion, smoothing, and dilation, were implemented on the VOIs accurately delineated by radiologists to generate diverse VOI variations. Then, we extracted 2068 radiomics features and evaluated the effects of VOI differences on feature values by the intra-class correlation coefficient (ICC). Feature selection was conducted by Maximum Relevance Minimum Redundancy combined with 0.632+ bootstrap algorithms. The prediction performance of radiomics models with random forest classifier were tested on an independent validation cohort by the area under the receive operating characteristic curve (AUC).

RESULTS

The larger the VOIs changed, the fewer features with high ICCs. Under any variation, SLN group showed fewer features with ICC ≥ 0.9 compared with NPC group. Not more than 15% top-predictive features identical to the accurate VOIs were observed across feature selection. The differences of AUCs of models derived from VOIs across smoothing or dilation with 3 pixels were not statistically significant compared with the accurate VOIs (p > 0.05) except for T2-weighted fat suppression images (smoothing: 0.845 vs. 0.725, p = 0.001; dilation: 0.800 vs. 0.725, p = 0.042). Dilation with 5 and 7 pixels contributed to remarkable AUCs in SLN group but the opposite in NPC group. The radiomics models did not perform well when tested by data from other delineations.

CONCLUSIONS

Differences in delineation of VOIs affected radiomics analysis, related to specific disease and MRI sequences. Differences from smooth delineation or expansion with 3 pixels width around the tumors or lesions were acceptable. The delineation for radiomics analysis should follow a predefined and unified standard.

Detection of Dominant Intra-prostatic Lesions in Patients With Prostate Cancer Using an Artificial Neural Network and MR Multi-modal Radiomics Analysis

Purpose: The aim of this study was to identify and rank discriminant radiomics features extracted from MR multi-modal images to construct an adaptive model for characterization of Dominant Intra-prostatic Lesions (DILs) from normal prostatic gland tissues (NT).

Methods and Materials: Two cohorts were retrospectively studied: Group A consisted of 98 patients and Group B 19 patients. Two image modalities were acquired using a 3.0T MR scanner: Axial T2 Weighted (T2W) and axial diffusion weighted (DW) imaging. A linear regression method was used to construct apparent diffusion coefficient (ADC) maps from DW images. DILs and the NT in the mirrored location were drawn on each modality. One hundred and sixty-eight radiomics features were extracted from DILs and NT. A Partial-Least-Squares-Correlation (PLSC) with one-way ANOVA along with bootstrapping ratio techniques were recruited to identify and rank the most discriminant latent variables. An artificial neural network (ANN) was constructed based on the optimal latent variable feature to classify the DILs and NTs. Nineteen patients were randomly chosen to test the contour variability effect on the radiomics analysis and the performance of the ANN. Finally, the trained ANN and a two dimension (2D) convolutional sampling method were combined and used to estimate DIL-NT probability map for two test cases.

Results: Among 168 radiomics-based latent variables, only the first four variables of each modality in the PLSC space were found to be significantly different between the DILs and NTs. Area Under Receiver Operating Characteristic (AUROC), Positive Predictive and Negative Predictive values (PPV and NPV) for the conventional method were 94%, 0.95, and 0.92, respectively. When the feature vector was randomly permuted 10,000 times, a very strong permutation-invariant efficiency (p < 0.0001) was achieved. The radiomic-based latent variables of the NTs and DILs showed no statistically significant differences (F_statistic < Fc = 4.11 with Confidence Level of 95% for all 8 variables) against contour variability. Dice coefficients between DIL-NT probability map and physician contours for the two test cases were 0.82 and 0.71, respectively.

Conclusion: This study demonstrates the high performance of combining radiomics information extracted from multimodal MR information such as T2WI and ADC maps, and adaptive models to detect DILs in patients with PCa.

Predicting survival and local control after radiochemotherapy in locally advanced head and neck cancer by means of computed tomography based radiomics

PURPOSE

To appraise the ability of a radiomics signature to predict clinical outcome after definitive radiochemotherapy (RCT) of stage III-IV head and neck cancer.

METHODS

A cohort of 110 patients was included in a retrospective analysis. Radiomics texture features were extracted from the gross tumor volumes contoured on planning computed tomography (CT) images. The cohort of patients was randomly divided into a training (70 patients) and a validation (40 patients) cohorts. Textural features were correlated to survival and control data to build predictive models. All the significant predictors of the univariate analysis were included in a multivariate model. The quality of the models was appraised by means of the concordance index (CI).

RESULTS

A signature with 3 features was identified as predictive of overall survival (OS) with CI = 0.88 and 0.90 for the training and validation cohorts, respectively. A signature with 2 features was identified for progression-free survival (PFS; CI = 0.72 and 0.80); 2 features also characterized the signature for local control (LC; CI = 0.72 and 0.82). In all cases, the stratification in high- and low-risk groups for the training and validation cohorts led to significant differences in the actuarial curves. In the validation cohort the mean OS times (in months) were 78.9 ± 2.1 vs 67.4 ± 6.0 in the low- and high-risk groups, respectively, the PFS was 73.1 ± 3.7 and 50.7 ± 7.2, while the LC was 78.7 ± 2.1 and 63.9 ± 6.5.

CONCLUSION

CT-based radiomic signatures that correlate with survival and control after RCT were identified and allow low- and high-risk groups of patients to be identified.

NCTN Assessment on Current Applications of Radiomics in Oncology

Radiomics is a fast-growing research area based on converting standard-of-care imaging into quantitative minable data and building subsequent predictive models to personalize treatment. Radiomics has been proposed as a study objective in clinical trial concepts and a potential biomarker for stratifying patients across interventional treatment arms. In recognizing the growing importance of radiomics in oncology, a group of medical physicists and clinicians from NRG Oncology reviewed the current status of the field and identified critical issues, providing a general assessment and early recommendations for incorporation in oncology studies.

Practical guidelines for handling head and neck computed tomography artifacts for quantitative image analysis

Radiomics studies have demonstrated the potential use of quantitative image features to improve prognostic stratification of patients with head and neck cancer. Imaging protocol parameters that can affect radiomics feature values have been investigated, but the effects of artifacts caused by intrinsic patient factors have not. Two such artifacts that are common in patients with head and neck cancer are streak artifacts caused by dental fillings and beam-hardening artifacts caused by bone. The purpose of this study was to test the impact of these artifacts and if needed, methods for compensating for these artifacts in head and neck radiomics studies. The robustness of feature values was tested by removing slices of the gross tumor volume (GTV) on computed tomography images from 30 patients with head and neck cancer; these images did not have streak artifacts or had artifacts far from the GTV. The range of each feature value over a percentage of the GTV was compared to the inter-patient variability at full volume. To determine the effects of beam-hardening artifacts, we scanned a phantom with 5 cartridges of different materials encased in polystyrene buildup. A cylindrical hole through the cartridges contained either a rod of polylactic acid to simulate water or a rod of polyvinyl chloride to simulate bone. A region of interest was drawn in each cartridge flush with the rod. Most features were robust with up to 50% of the original GTV removed. Most feature values did not significantly differ when measured with the polylactic acid rod or the polyvinyl chloride rod. Of those that did, the size of the difference did not exceed the inter-patient standard deviation in most cases. We conclude that simply removing slices affected by streak artifacts can enable these scans to be included in radiomics studies and that contours of structures can abut bone without being affected by beam hardening if needed.

Repeatability and Reproducibility of Radiomic Features: A Systematic Review

Purpose:

An ever-growing number of predictive models used to inform clinical decision making have included quantitative, computer-extracted imaging biomarkers, or “radiomic features.” Broadly generalizable validity of radiomics-assisted models may be impeded by concerns about reproducibility. We offer a qualitative synthesis of 41 studies that specifically investigated the repeatability and reproducibility of radiomic features, derived from a systematic review of published peer-reviewed literature.

Methods and Materials:

The PubMed electronic database was searched using combinations of the broad Haynes and Ingui filters along with a set of text words specific to cancer, radiomics (including texture analyses), reproducibility, and repeatability. This review has been reported in compliance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. From each full-text article, information was extracted regarding cancer type, class of radiomic feature examined, reporting quality of key processing steps, and statistical metric used to segregate stable features.

Results:

Among 624 unique records, 41 full-text articles were subjected to review. The studies primarily addressed non-small cell lung cancer and oropharyngeal cancer. Only 7 studies addressed in detail every methodologic aspect related to image acquisition, preprocessing, and feature extraction. The repeatability and reproducibility of radiomic features are sensitive at various degrees to processing details such as image acquisition settings, image reconstruction algorithm, digital image preprocessing, and software used to extract radiomic features. First-order features were overall more reproducible than shape metrics and textural features. Entropy was consistently reported as one of the most stable first-order features. There was no emergent consensus regarding either shape metrics or textural features; however, coarseness and contrast appeared among the least reproducible.

Conclusions:

Investigations of feature repeatability and reproducibility are currently limited to a small number of cancer types. Reporting quality could be improved regarding details of feature extraction software, digital image manipulation (preprocessing), and the cutoff value used to distinguish stable features.

We offer a qualitative synthesis of 41 studies that specifically investigated the repeatability and reproducibility of radiomic features. The repeatability and reproducibility of radiomic features are sensitive at various degrees to image quality and to software used to extract radiomic features. Investigations of feature repeatability and reproducibility are currently limited to a small number of cancer types. No consensus was found regarding the most repeatable and reproducible features with respect to different settings.

Accounting for reconstruction kernel-induced variability in CT radiomic features using noise power spectra

Large variability in computed tomography (CT) radiomics feature values due to CT imaging parameters can have subsequent implications on the prognostic or predictive significance of these features. Here, we investigated the impact of pitch, dose, and reconstruction kernel on CT radiomic features. Moreover, we introduced correction factors to reduce feature variability introduced by reconstruction kernels. The credence cartridge radiomics and American College of Radiology (ACR) phantoms were scanned on five different scanners. ACR phantom was used for 3-D noise power spectrum (NPS) measurements to quantify correlated noise. The coefficient of variation (COV) was used as the variability assessment metric. The variability in texture features due to different kernels was reduced by applying the NPS peak frequency and region of interest (ROI) maximum intensity as correction factors. Most texture features were dose independent but were strongly kernel dependent, which is demonstrated by a significant shift in NPS peak frequency among kernels. Percentage improvement in robustness was calculated for each feature from original and corrected %COV values. Percentage improvements in robustness of 19 features were in the range of 30% to 78% after corrections. We show that NPS peak frequency and ROI maximum intensity can be used as correction factors to reduce variability in CT texture feature values due to reconstruction kernels.