Classification and evaluation strategies of auto-segmentation approaches for PET: Report of AAPM task group No. 211

Metrics to evaluate the performance of auto-segmentation for radiation treatment planning: A critical review

Advances in artificial intelligence-based methods have led to the development and publication of numerous systems for auto-segmentation in radiotherapy. These systems have the potential to decrease contour variability, which has been associated with poor clinical outcomes and increased efficiency in the treatment planning workflow. However, there are no uniform standards for evaluating auto-segmentation platforms to assess their efficacy at meeting these goals. Here, we review the most frequently used evaluation techniques which include geometric overlap, dosimetric parameters, time spent contouring, and clinical rating scales. These data suggest that many of the most commonly used geometric indices, such as the Dice Similarity Coefficient, are not well correlated with clinically meaningful endpoints. As such, a multi-domain evaluation, including composite geometric and/or dosimetric metrics with physician-reported assessment, is necessary to gauge the clinical readiness of auto-segmentation for radiation treatment planning.

Motion-compensated FDG PET/CT for oesophageal cancer

PURPOSE

Respiratory-induced motion of oesophageal tumours and lymph nodes can influence positron-emission tomography/computed tomography (PET/CT). The aim was to compare standard three-dimensional (3D) and motion-compensated PET/CT regarding standardized uptake value (SUV), metabolic tumour volume (MTV) and detection of lymph node metastases.

METHODS

This prospective observational study (NCT02424864) included 37 newly diagnosed oesophageal cancer patients. Diagnostic PET/CT was reconstructed in 3D and motion-compensated PET/CT. MTVs of the primary tumour were calculated using an automated region-growing algorithm with SUV thresholds of 2.5 (MTV2.5) and ≥ 50% of SUVmax (MTV50%). Blinded for reconstruction method, a nuclear medicine physician assessed all lymph nodes showing ¹⁸F‑fluorodeoxyglucose uptake for their degree of suspicion.

RESULTS

The mean (95% CI) SUVmax of the primary tumour was 13.1 (10.6-15.5) versus 13.0 (10.4-15.6) for 3D and motion-compensated PET/CT, respectively. MTVs were also similar between the two techniques. Bland-Altman analysis showed mean differences between both measurements (95% limits of agreement) of 0.08 (-3.60-3.75), -0.26 (-2.34-1.82), 4.66 (-29.61-38.92) cm³ and -0.95 (-19.9-18.0) cm³ for tumour SUVmax, lymph node SUVmax, MTV2.5 and MTV50%, respectively. Lymph nodes were classified as highly suspicious (30/34 nodes), suspicious (20/22) and dubious (66/59) for metastases on 3D/motion-compensated PET/CT. No additional lymph node metastases were found on motion-compensated PET/CT. SUVmax of the most intense lymph nodes was similar for both scans: mean (95% CI) 6.6 (4.3-8.8) and 6.8 (4.5-9.1) for 3D and motion-compensated, respectively.

CONCLUSION

SUVmax of the primary oesophageal tumour and lymph nodes was comparable on 3D and motion-compensated PET/CT. The use of motion-compensated PET/CT did not improve lymph node detection.

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.

Challenges in the target volume definition of lung cancer radiotherapy

Radiotherapy, with or without systemic treatment has an important role in the management of lung cancer. In order to deliver the treatment accurately, the clinician must precisely outline the gross tumour volume (GTV), mostly on computed tomography (CT) images. However, due to the limited contrast between tumour and non-malignant changes in the lung tissue, it can be difficult to distinguish the tumour boundaries on CT images leading to large interobserver variation and differences in interpretation. Therefore the definition of the GTV has often been described as the weakest link in radiotherapy with its inaccuracy potentially leading to missing the tumour or unnecessarily irradiating normal tissue. In this article, we review the various techniques that can be used to reduce delineation uncertainties in lung cancer.

EANM/SNMMI practice guideline for [18F]FDG PET/CT external beam radiotherapy treatment planning in uterine cervical cancer v1.0

PURPOSE

The aim of this EANM / SNMMI Practice Guideline with ESTRO endorsement is to provide general information and specific considerations about [18F]FDG PET/CT in advanced uterine cervical cancer for external beam radiotherapy planning with emphasis on staging and target definition, mostly in FIGO stages IB3-IVA and IVB, treated with curative intention.

METHODS

Guidelines from related fields, relevant literature and leading experts have been consulted during the development of this guideline. As this field is rapidly evolving, this guideline cannot be seen as definitive, nor is it a summary of all existing protocols. Local variations should be taken into consideration when applying this guideline.

CONCLUSION

The background, common clinical indications, qualifications and responsibilities of personnel, procedure / specifications of the examination, documentation / reporting and equipment specifications, quality control and radiation safety in imaging is discussed with an emphasis on the multidisciplinary approach.

Convolutional neural networks for PET functional volume fully automatic segmentation: development and validation in a multi-center setting

Purpose

In this work, we addressed fully automatic determination of tumor functional uptake from positron emission tomography (PET) images without relying on other image modalities or additional prior constraints, in the context of multicenter images with heterogeneous characteristics.

Methods

In cervical cancer, an additional challenge is the location of the tumor uptake near or even stuck to the bladder. PET datasets of 232 patients from five institutions were exploited. To avoid unreliable manual delineations, the ground truth was generated with a semi-automated approach: a volume containing the tumor and excluding the bladder was first manually determined, then a well-validated, semi-automated approach relying on the Fuzzy locally Adaptive Bayesian (FLAB) algorithm was applied to generate the ground truth. Our model built on the U-Net architecture incorporates residual blocks with concurrent spatial squeeze and excitation modules, as well as learnable non-linear downsampling and upsampling blocks. Experiments relied on cross-validation (four institutions for training and validation, and the fifth for testing).

Results

The model achieved good Dice similarity coefficient (DSC) with little variability across institutions (0.80 ± 0.03), with higher recall (0.90 ± 0.05) than precision (0.75 ± 0.05) and improved results over the standard U-Net (DSC 0.77 ± 0.05, recall 0.87 ± 0.02, precision 0.74 ± 0.08). Both vastly outperformed a fixed threshold at 40% of SUVmax (DSC 0.33 ± 0.15, recall 0.52 ± 0.17, precision 0.30 ± 0.16). In all cases, the model could determine the tumor uptake without including the bladder. Neither shape priors nor anatomical information was required to achieve efficient training.

Conclusion

The proposed method could facilitate the deployment of a fully automated radiomics pipeline in such a challenging multicenter context.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00259-021-05244-z.

Quantitative PET in the 2020s: a roadmap

Positron emission tomography (PET) plays an increasingly important role in research and clinical applications, catalysed by remarkable technical advances and a growing appreciation of the need for reliable, sensitive biomarkers of human function in health and disease. Over the last 30 years, a large amount of the physics and engineering effort in PET has been motivated by the dominant clinical application during that period, oncology. This has led to important developments such as PET/CT, whole-body PET, 3D PET, accelerated statistical image reconstruction, and time-of-flight PET. Despite impressive improvements in image quality as a result of these advances, the emphasis on static, semi-quantitative 'hot spot' imaging for oncologic applications has meant that the capability of PET to quantify biologically relevant parameters based on tracer kinetics has not been fully exploited. More recent advances, such as PET/MR and total-body PET, have opened up the ability to address a vast range of new research questions, from which a future expansion of applications and radiotracers appears highly likely. Many of these new applications and tracers will, at least initially, require quantitative analyses that more fully exploit the exquisite sensitivity of PET and the tracer principle on which it is based. It is also expected that they will require more sophisticated quantitative analysis methods than those that are currently available. At the same time, artificial intelligence is revolutionizing data analysis and impacting the relationship between the statistical quality of the acquired data and the information we can extract from the data. In this roadmap, leaders of the key sub-disciplines of the field identify the challenges and opportunities to be addressed over the next ten years that will enable PET to realise its full quantitative potential, initially in research laboratories and, ultimately, in clinical practice.

A comparison of methods for fully automatic segmentation of tumors and involved nodes in PET/CT of head and neck cancers

Target volume delineation is a vital but time-consuming and challenging part of radiotherapy, where the goal is to deliver sufficient dose to the target while reducing risks of side effects. For head and neck cancer (HNC) this is complicated by the complex anatomy of the head and neck region and the proximity of target volumes to organs at risk. The purpose of this study was to compare and evaluate conventional PET thresholding methods, six classical machine learning algorithms and a 2D U-Net convolutional neural network (CNN) for automatic gross tumor volume (GTV) segmentation of HNC in PET/CT images. For the latter two approaches the impact of single versus multimodality input on segmentation quality was also assessed. 197 patients were included in the study. The cohort was split into training and test sets (157 and 40 patients, respectively). Five-fold cross-validation was used on the training set for model comparison and selection. Manual GTV delineations represented the ground truth. Tresholding, classical machine learning and CNN segmentation models were ranked separately according to the cross-validation Sørensen-Dice similarity coefficient (Dice). PET thresholding gave a maximum mean Dice of 0.62, whereas classical machine learning resulted in maximum mean Dice scores of 0.24 (CT) and 0.66 (PET; PET/CT). CNN models obtained maximum mean Dice scores of 0.66 (CT), 0.68 (PET) and 0.74 (PET/CT). The difference in cross-validation Dice between multimodality PET/CT and single modality CNN models was significant (p ≤ 0.0001). The top-ranked PET/CT-based CNN model outperformed the best-performing thresholding and classical machine learning models, giving significantly better segmentations in terms of cross-validation and test set Dice, true positive rate, positive predictive value and surface distance-based metrics (p ≤ 0.0001). Thus, deep learning based on multimodality PET/CT input resulted in superior target coverage and less inclusion of surrounding normal tissue.

Positron emission tomography for radiotherapy planning in head and neck cancer: What impact?

PET-computed tomography (CT) plays a growing role to guide target volume delineation for head and neck cancer in radiation oncology. Pretherapeutic [18F]FDG PET-CT adds information to morphological imaging. First, as a whole-body imaging modality, it reveals regional or distant metastases that induce major therapeutic changes in more than 10% of the cases. Moreover, it allows better pathological lymph node selection which improves overall regional control and overall survival. Second, locally, it allows us to define the metabolic tumoral volume, which is a reliable prognostic feature for survival outcome. [18F]FDG PET-CT-based gross tumor volume (GTV) is on average significantly smaller than GTV based on CT. Nevertheless, the overlap is incomplete and more evaluation of composite GTV based on PET and GTV based on CT are needed. However, in clinical practice, the study showed that using GTV PET alone for treatment planning was similar to using GTVCT for local control and dose distribution was better as a dose to organs at risk significantly decreased. In addition to FDG, pretherapeutic PET could give access to different biological tumoral volumes - thanks to different tracers - guiding heterogeneous dose delivery (dose painting concept) to resistant subvolumes. During radiotherapy treatment, follow-up [18F]FDG PET-CT revealed an earlier and more important diminution of GTV than other imaging modality. It may be a valuable support for adaptative radiotherapy as a new treatment plan with a significant impact on dose distribution became possible. Finally, additional studies are required to prospectively validate long-term outcomes and lower toxicity resulting from the use of PET-CT in treatment planning.

Multiple U-Net-Based Automatic Segmentations and Radiomics Feature Stability on Ultrasound Images for Patients With Ovarian Cancer

Few studies have reported the reproducibility and stability of ultrasound (US) images based radiomics features obtained from automatic segmentation in oncology. The purpose of this study is to study the accuracy of automatic segmentation algorithms based on multiple U-net models and their effects on radiomics features from US images for patients with ovarian cancer. A total of 469 US images from 127 patients were collected and randomly divided into three groups: training sets (353 images), validation sets (23 images), and test sets (93 images) for automatic segmentation models building. Manual segmentation of target volumes was delineated as ground truth. Automatic segmentations were conducted with U-net, U-net++, U-net with Resnet as the backbone (U-net with Resnet), and CE-Net. A python 3.7.0 and package Pyradiomics 2.2.0 were used to extract radiomic features from the segmented target volumes. The accuracy of automatic segmentations was evaluated by Jaccard similarity coefficient (JSC), dice similarity coefficient (DSC), and average surface distance (ASD). The reliability of radiomics features were evaluated by Pearson correlation and intraclass correlation coefficients (ICC). CE-Net and U-net with Resnet outperformed U-net and U-net++ in accuracy performance by achieving a DSC, JSC, and ASD of 0.87, 0.79, 8.54, and 0.86, 0.78, 10.00, respectively. A total of 97 features were extracted from the delineated target volumes. The average Pearson correlation was 0.86 (95% CI, 0.83–0.89), 0.87 (95% CI, 0.84–0.90), 0.88 (95% CI, 0.86–0.91), and 0.90 (95% CI, 0.88–0.92) for U-net++, U-net, U-net with Resnet, and CE-Net, respectively. The average ICC was 0.84 (95% CI, 0.81–0.87), 0.85 (95% CI, 0.82–0.88), 0.88 (95% CI, 0.85–0.90), and 0.89 (95% CI, 0.86–0.91) for U-net++, U-net, U-net with Resnet, and CE-Net, respectively. CE-Net based segmentation achieved the best radiomics reliability. In conclusion, U-net based automatic segmentation was accurate enough to delineate the target volumes on US images for patients with ovarian cancer. Radiomics features extracted from automatic segmented targets showed good reproducibility and for reliability further radiomics investigations.

Repeatability of two semi-automatic artificial intelligence approaches for tumor segmentation in PET

BACKGROUND

Positron emission tomography (PET) is routinely used for cancer staging and treatment follow-up. Metabolic active tumor volume (MATV) as well as total MATV (TMATV-including primary tumor, lymph nodes and metastasis) and/or total lesion glycolysis derived from PET images have been identified as prognostic factor or for the evaluation of treatment efficacy in cancer patients. To this end, a segmentation approach with high precision and repeatability is important. However, the implementation of a repeatable and accurate segmentation algorithm remains an ongoing challenge.

METHODS

In this study, we compare two semi-automatic artificial intelligence (AI)-based segmentation methods with conventional semi-automatic segmentation approaches in terms of repeatability. One segmentation approach is based on a textural feature (TF) segmentation approach designed for accurate and repeatable segmentation of primary tumors and metastasis. Moreover, a convolutional neural network (CNN) is trained. The algorithms are trained, validated and tested using a lung cancer PET dataset. The segmentation accuracy of both segmentation approaches is compared using the Jaccard coefficient (JC). Additionally, the approaches are externally tested on a fully independent test-retest dataset. The repeatability of the methods is compared with those of two majority vote (MV2, MV3) approaches, 41%SUVMAX, and a SUV > 4 segmentation (SUV4). Repeatability is assessed with test-retest coefficients (TRT%) and intraclass correlation coefficient (ICC). An ICC > 0.9 was regarded as representing excellent repeatability.

RESULTS

The accuracy of the segmentations with the reference segmentation was good (JC median TF: 0.7, CNN: 0.73). Both segmentation approaches outperformed most other conventional segmentation methods in terms of test-retest coefficient (TRT% mean: TF: 13.0%, CNN: 13.9%, MV2: 14.1%, MV3: 28.1%, 41%SUVMAX: 28.1%, SUV4: 18.1%) and ICC (TF: 0.98, MV2: 0.97, CNN: 0.99, MV3: 0.73, SUV4: 0.81, and 41%SUVMAX: 0.68).

CONCLUSION

The semi-automatic AI-based segmentation approaches used in this study provided better repeatability than conventional segmentation approaches. Moreover, both algorithms lead to accurate segmentations for both primary tumors as well as metastasis and are therefore good candidates for PET tumor segmentation.

Prediction of lymph node metastases using pre-treatment PET radiomics of the primary tumour in esophageal adenocarcinoma: an external validation study

OBJECTIVES

To improve clinical lymph node staging (cN-stage) in oesophageal adenocarcinoma by developing and externally validating three prediction models; one with clinical variables only, one with positron emission tomography (PET) radiomics only, and a combined clinical and radiomics model.

METHODS

Consecutive patients with fluorodeoxyglucose (FDG) avid tumours treated with neoadjuvant therapy between 2010 and 2016 in two international centres (n = 130 and n = 60, respectively) were included. Four clinical variables (age, gender, clinical T-stage and tumour regression grade) and PET radiomics from the primary tumour were used for model development. Diagnostic accuracy, area under curve (AUC), discrimination and calibration were calculated for each model. The prognostic significance was also assessed.

RESULTS

The incidence of lymph node metastases was 58% in both cohorts. The areas under the curve of the clinical, radiomics and combined models were 0.79, 0.69 and 0.82 in the developmental cohort, and 0.65, 0.63 and 0.69 in the external validation cohort, with good calibration demonstrated. The area under the curve of current cN-stage in development and validation cohorts was 0.60 and 0.66, respectively. For overall survival, the combined clinical and radiomics model achieved the best discrimination performance in the external validation cohort (X² = 6.08, df = 1, p = 0.01).

CONCLUSION

Accurate diagnosis of lymph node metastases is crucial for prognosis and guiding treatment decisions. Despite finding improved predictive performance in the development cohort, the models using PET radiomics derived from the primary tumour were not fully replicated in an external validation cohort.

ADVANCES IN KNOWLEDGE

This international study attempted to externally validate a new prediction model for lymph node metastases using PET radiomics. A model combining clinical variables and PET radiomics improved discrimination of lymph node metastases, but these results were not externally replicated.

Optimal method for metabolic tumour volume assessment of cervical cancers with inter-observer agreement on [18F]-fluoro-deoxy-glucose positron emission tomography with computed tomography

PURPOSE

Cervical cancer metabolic tumour volume (MTV) derived from [18F]-FDG PET/CT has a role in prognostication and therapy planning. There is no standard method of outlining MTV on [18F]-FDG PET/CT. The aim of this study was to assess the optimal method to outline primary cervical tumours on [18F]-FDG PET/CT using MRI-derived tumour volumes as the reference standard.

METHODS

81 consecutive cervical cancer patients with pre-treatment staging MRI and [18F]-FDG PET/CT imaging were included. MRI volumes were compared with different PET segmentation methods. Method 1 measured MTVs at different SUV_max thresholds ranging from 20 to 60% (MTV₂₀-MTV₆₀) with bladder masking and manual adjustment when required. Method 2 created an isocontour around the tumour prior to different SUV_max thresholds being applied. Method 3 used an automated gradient method. Inter-observer agreement of MTV, following manual adjustment when required, was recorded.

RESULTS

For method 1, the MTV₂₅ and MTV₃₀ were closest to the MRI volumes for both readers (mean percentage change from MRI volume of 2.9% and 13.4% for MTV₂₅ and - 13.1% and - 2.0% for MTV₃₀ for readers 1 and 2). 70% of lesions required manual adjustment at MTV₂₅ compared with 45% at MTV₃₀. There was excellent inter-observer agreement between MTV₃₀ to MTV₆₀ (ICC ranged from 0.898-0.976 with narrow 95% confidence intervals (CIs)) and moderate agreement at lower thresholds (ICC estimates of 0.534 and 0.617, respectively for the MTV₂₀ and MTV₂₅ with wide 95% CIs). Bladder masking was performed in 86% of cases overall. For method 2, excellent correlation was demonstrated at MTV₂₅ and MTV₃₀ (mean % change from MRI volume of -3.9% and - 8.6% for MTV₂₅ and - 16.9% and 19% for MTV₃₀ for readers 1 and 2, respectively). This method also demonstrated excellent ICC across all thresholds with no manual adjustment. Method 3 demonstrated excellent ICC of 0.96 (95% CI 0.94-0.97) but had a mean percentage difference from the MRI volume of - 19.1 and - 18.2% for readers 1 and 2, respectively. 21% required manual adjustment for both readers.

CONCLUSION

MTV₃₀ provides the optimal correlation with MRI volume taking into consideration the excellent inter-reader agreement and less requirement for manual adjustment.

Artificial intelligence applications for oncological positron emission tomography imaging

Positron emission tomography (PET), a functional and dynamic molecular imaging technique, is generally used to reveal tumors' biological behavior. Radiomics allows a high-throughput extraction of multiple features from images with artificial intelligence (AI) approaches and develops rapidly worldwide. Quantitative and objective features of medical images have been explored to recognize reliable biomarkers, with the development of PET radiomics. This paper will review the current clinical exploration of PET-based classical machine learning and deep learning methods, including disease diagnosis, the prediction of histological subtype, gene mutation status, tumor metastasis, tumor relapse, therapeutic side effects, therapeutic intervention and evaluation of prognosis. The applications of AI in oncology will be mainly discussed. The image-guided biopsy or surgery assisted by PET-based AI will be introduced as well. This paper aims to present the applications and methods of AI for PET imaging, which may offer important details for further clinical studies. Relevant precautions are put forward and future research directions are suggested.

Comparison of 11 automated PET segmentation methods in lymphoma

Segmentation of lymphoma lesions in FDG PET/CT images is critical in both assessing individual lesions and quantifying patient disease burden. Simple thresholding methods remain common despite the large heterogeneity in lymphoma lesion location, size, and contrast. Here, we assess 11 automated PET segmentation methods for their use in two scenarios: individual lesion segmentation and patient-level disease quantification in lymphoma. Lesions on ¹⁸F-FDG PET/CT scans of 90 lymphoma patients were contoured by a nuclear medicine physician. Thresholding, active contours, clustering, adaptive region-growing, and convolutional neural network (CNN) methods were implemented on all physician-identified lesions. Lesion-level segmentation was evaluated using multiple segmentation performance metrics (Dice, Hausdorff Distance). Patient-level quantification of total disease burden (SUV_total) and metabolic tumor volume (MTV) was assessed using Spearman's correlation coefficients between the segmentation output and physician contours. Lesion segmentation and patient quantification performance was compared to inter-physician agreement in a subset of 20 patients segmented by a second nuclear medicine physician. In total, 1223 lesions with median tumor-to-background ratio of 4.0 and volume of 1.8 cm³, were evaluated. When assessed for lesion segmentation, a 3D CNN, DeepMedic, achieved the highest performance across all evaluation metrics. DeepMedic, clustering methods, and an iterative threshold method had lesion-level segmentation performance comparable to the degree of inter-physician agreement. For patient-level SUV_total and MTV quantification, all methods except 40% and 50% SUV_max and adaptive region-growing achieved a performance that was similar the agreement of the two physicians. Multiple methods, including a 3D CNN, clustering, and an iterative threshold method, achieved both good lesion-level segmentation and patient-level quantification performance in a population of 90 lymphoma patients. These methods are thus recommended over thresholding methods such as 40% and 50% SUV_max, which were consistently found to be significantly outside the limits defined by inter-physician agreement.

Radiomics in PET/CT: Current Status and Future AI-Based Evolutions

This short review aims at providing the readers with an update on the current status, as well as future perspectives in the quickly evolving field of radiomics applied to the field of PET/CT imaging. Numerous pitfalls have been identified in study design, data acquisition, segmentation, features calculation and modeling by the radiomics community, and these are often the same issues across all image modalities and clinical applications, however some of these are specific to PET/CT (and SPECT/CT) imaging and therefore the present paper focuses on those. In most cases, recommendations and potential methodological solutions do exist and should therefore be followed to improve the overall quality and reproducibility of published studies. In terms of future evolutions, the techniques from the larger field of artificial intelligence (AI), including those relying on deep neural networks (also known as deep learning) have already shown impressive potential to provide solutions, especially in terms of automation, but also to maybe fully replace the tools the radiomics community has been using until now in order to build the usual radiomics workflow. Some important challenges remain to be addressed before the full impact of AI may be realized but overall the field has made striking advances over the last few years and it is expected advances will continue at a rapid pace.

Fully automatic segmentation of diffuse large B cell lymphoma lesions on 3D FDG-PET/CT for total metabolic tumour volume prediction using a convolutional neural network

PURPOSE

Lymphoma lesion detection and segmentation on whole-body FDG-PET/CT are a challenging task because of the diversity of involved nodes, organs or physiological uptakes. We sought to investigate the performances of a three-dimensional (3D) convolutional neural network (CNN) to automatically segment total metabolic tumour volume (TMTV) in large datasets of patients with diffuse large B cell lymphoma (DLBCL).

METHODS

The dataset contained pre-therapy FDG-PET/CT from 733 DLBCL patients of 2 prospective LYmphoma Study Association (LYSA) trials. The first cohort (n = 639) was used for training using a 5-fold cross validation scheme. The second cohort (n = 94) was used for external validation of TMTV predictions. Ground truth masks were manually obtained after a 41% SUVmax adaptive thresholding of lymphoma lesions. A 3D U-net architecture with 2 input channels for PET and CT was trained on patches randomly sampled within PET/CTs with a summed cross entropy and Dice similarity coefficient (DSC) loss. Segmentation performance was assessed by the DSC and Jaccard coefficients. Finally, TMTV predictions were validated on the second independent cohort.

RESULTS

Mean DSC and Jaccard coefficients (± standard deviation) in the validations set were 0.73 ± 0.20 and 0.68 ± 0.21, respectively. An underestimation of mean TMTV by - 12 mL (2.8%) ± 263 was found in the validation sets of the first cohort (P = 0.27). In the second cohort, an underestimation of mean TMTV by - 116 mL (20.8%) ± 425 was statistically significant (P = 0.01).

CONCLUSION

Our CNN is a promising tool for automatic detection and segmentation of lymphoma lesions, despite slight underestimation of TMTV. The fully automatic and open-source features of this CNN will allow to increase both dissemination in routine practice and reproducibility of TMTV assessment in lymphoma patients.

Fully 3D Active Surface with Machine Learning for PET Image Segmentation

In order to tackle three-dimensional tumor volume reconstruction from Positron Emission Tomography (PET) images, most of the existing algorithms rely on the segmentation of independent PET slices. To exploit cross-slice information, typically overlooked in these 2D implementations, I present an algorithm capable of achieving the volume reconstruction directly in 3D, by leveraging an active surface algorithm. The evolution of such surface performs the segmentation of the whole stack of slices simultaneously and can handle changes in topology. Furthermore, no artificial stop condition is required, as the active surface will naturally converge to a stable topology. In addition, I include a machine learning component to enhance the accuracy of the segmentation process. The latter consists of a forcing term based on classification results from a discriminant analysis algorithm, which is included directly in the mathematical formulation of the energy function driving surface evolution. It is worth noting that the training of such a component requires minimal data compared to more involved deep learning methods. Only eight patients (i.e., two lung, four head and neck, and two brain cancers) were used for training and testing the machine learning component, while fifty patients (i.e., 10 lung, 25 head and neck, and 15 brain cancers) were used to test the full 3D reconstruction algorithm. Performance evaluation is based on the same dataset of patients discussed in my previous work, where the segmentation was performed using the 2D active contour. The results confirm that the active surface algorithm is superior to the active contour algorithm, outperforming the earlier approach on all the investigated anatomical districts with a dice similarity coefficient of 90.47 ± 2.36% for lung cancer, 88.30 ± 2.89% for head and neck cancer, and 90.29 ± 2.52% for brain cancer. Based on the reported results, it can be claimed that the migration into a 3D system yielded a practical benefit justifying the effort to rewrite an existing 2D system for PET imaging segmentation.

Texture Feature Comparison Between Step-and-Shoot and Continuous-Bed-Motion 18F-FDG PET

Our objective was to investigate the differences in texture features between step-and-shoot (SS) and continuous-bed-motion (CBM) imaging in phantom and clinical studies. Methods: A National Electrical Manufacturers Association body phantom was filled with ¹⁸F-FDG solution at a sphere-to-background ratio of 4:1. SS and CBM were performed using the same acquisition duration, and the data were reconstructed using 3-dimensional ordered-subset expectation maximization with time-of-flight algorithms. Texture features were extracted using the software LIFEx. A volume of interest was delineated on the 22-, 28-, and 37-mm spheres with a threshold of 42% of the maximum SUV. The voxel intensities were discretized using 2 resampling methods, namely a fixed bin size and a fixed bin number discretization. The discrete resampling values were set to 64 and 128. In total, 31 texture features were calculated with gray-level cooccurrence matrix (GLCM), gray-level run length matrix, neighborhood gray-level different matrix, and gray-level zone length matrix. The texture features of the SS and CBM images were compared for all settings using the paired t test and the coefficient of variation. In a clinical study, 27 lesions from 20 patients were examined using the same acquisition and image processing as were used during the phantom study. The percentage difference (%Diff) and correlation between the texture features from SS and CBM images were calculated to evaluate agreement between the 2 scanning techniques. Results: In the phantom study, the 11 features exhibited no significant difference between SS and CBM images, and the coefficient of variation was no more than 10%, depending on resampling conditions, whereas entropy and dissimilarity from GLCM fulfilled the criteria for all settings. In the clinical study, the entropy and dissimilarity from GLCM exhibited a low %Diff and excellent correlation in all resampling conditions. The %Diff of entropy was lower than that of dissimilarity. Conclusion: Differences between the texture features of SS and CBM images varied depending on the type of feature. Because entropy for GLCM exhibits minimal differences between SS and CBM images irrespective of resampling conditions, entropy may be the optimal feature to reduce the differences between the 2 scanning techniques.

Imaging for Target Delineation in Head and Neck Cancer Radiotherapy

The definition of tumor involved volumes in patients with head and neck cancer poses great challenges with the increasing use of highly conformal radiotherapy techniques eg, volumetric modulated arc therapy and intensity modulated proton therapy. The risk of underdosing the tumor might increase unless great care is taken in the process. The information gained from imaging is increasing with both PET and MRI becoming readily available for the definition of targets. The information gained from these techniques is indeed multidimensional as one often acquire data on eg, metabolism, diffusion, and hypoxia together with anatomical and structural information. Nevertheless, much work remains to fully exploit the available information on a patient-specific level. Multimodality target definition in radiotherapy is a chain of processes that must be individually scrutinized, optimized and quality assured. Any uncertainties or errors in image acquisition, reconstruction, interpretation, and delineation are systematic errors and hence will potentially have a detrimental effect on the entire radiotherapy treatment and hence; the chance of cure or the risk of unnecessary side effects. Common guidelines and procedures create a common minimum standard and ground for evaluation and development. In Denmark, the treatment of head and neck cancer is organized within the multidisciplinary Danish Head and Neck Cancer Group (DAHANCA). The radiotherapy quality assurance group of DAHANCA organized a workshop in January 2020 with participants from oncology, radiology, and nuclear medicine from all centers in Denmark, treating patients with head and neck cancer. The participants agreed on a national guideline on imaging for target delineation in head and neck cancer radiotherapy, which has been approved by the DAHANCA group. The guidelines are available in the Supplementary. The use of multimodality imaging is being recommended for the planning of all radical treatments with a macroscopic tumor. 2-[¹⁸F]FDG-PET/CT should be available, preferable in the treatment position. The recommended MRI sequences are T1, T2 with and without fat suppression, and T1 with contrast enhancement, preferable in the treatment position. The interpretation of clinical information, including thorough physical examination as well as imaging, should be done in a multidisciplinary setting with an oncologist, radiologist, and nuclear medicine specialist.

Metabolic radiogenomics in lung cancer: associations between FDG PET image features and oncogenic signaling pathway alterations

This study investigated the associations between image features extracted from tumor ¹⁸F-fluorodeoxyglucose (FDG) uptake and genetic alterations in patients with lung cancer. A total of 137 patients (age, 62.7 ± 10.2 years) who underwent FDG positron emission tomography/computed tomography (PET/CT) and targeted deep sequencing analysis for a tumor lesion, comprising 61 adenocarcinoma (ADC), 31 squamous cell carcinoma (SQCC), and 45 small cell lung cancer (SCLC) patients, were enrolled in this study. From the tumor lesions, 86 image features were extracted, and 381 genes were assessed. PET features were associated with genetic mutations: 41 genes with 24 features in ADC; 35 genes with 22 features in SQCC; and 43 genes with 25 features in SCLC (FDR < 0.05). Clusters based on PET features showed an association with alterations in oncogenic signaling pathways: Cell cycle and WNT signaling pathways in ADC (p = 0.023, p = 0.035, respectively); Cell cycle, p53, and WNT in SQCC (p = 0.045, 0.009, and 0.029, respectively); and TGFβ in SCLC (p = 0.030). In addition, SUV_peak and SUV_max were associated with a mutation of the TGFβ signaling pathway in ADC (FDR = 0.001, < 0.001). In this study, PET image features had significant associations with alterations in genes and oncogenic signaling pathways in patients with lung cancer.

Reinventing radiation therapy with machine learning and imaging bio-markers (radiomics): State-of-the-art, challenges and perspectives

Radiation therapy is a pivotal cancer treatment that has significantly progressed over the last decade due to numerous technological breakthroughs. Imaging is now playing a critical role on deployment of the clinical workflow, both for treatment planning and treatment delivery. Machine-learning analysis of predefined features extracted from medical images, i.e. radiomics, has emerged as a promising clinical tool for a wide range of clinical problems addressing drug development, clinical diagnosis, treatment selection and implementation as well as prognosis. Radiomics denotes a paradigm shift redefining medical images as a quantitative asset for data-driven precision medicine. The adoption of machine-learning in a clinical setting and in particular of radiomics features requires the selection of robust, representative and clinically interpretable biomarkers that are properly evaluated on a representative clinical data set. To be clinically relevant, radiomics must not only improve patients' management with great accuracy but also be reproducible and generalizable. Hence, this review explores the existing literature and exposes its potential technical caveats, such as the lack of quality control, standardization, sufficient sample size, type of data collection, and external validation. Based upon the analysis of 165 original research studies based on PET, CT-scan, and MRI, this review provides an overview of new concepts, and hypotheses generating findings that should be validated. In particular, it describes evolving research trends to enhance several clinical tasks such as prognostication, treatment planning, response assessment, prediction of recurrence/relapse, and prediction of toxicity. Perspectives regarding the implementation of an AI-based radiotherapy workflow are presented.

Impact of suboptimal dosimetric coverage of pretherapeutic 18F-FDG PET/CT hotspots on outcome in patients with locally advanced cervical cancer treated with chemoradiotherapy followed by brachytherapy

INTRODUCTION

Areas of high uptake on pre-treatment 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography/computed tomography (PET/CT), denoted as "hotspots", have been identified as preferential sites of local relapse in locally advanced cervical cancer (LACC). The purpose of this study was to analyze the dosimetric coverage of these hotspots with high dose-rate brachytherapy (BT).

METHODS

For each patient, a rigid registration of the CT from the pre-treatment PET/CT with the radiotherapy planning CT was performed using 3D SlicerTM, followed by a manual volume correction by translation and deformation if necessary. The fuzzy locally adaptive Bayesian (FLAB) algorithm was applied to PET images to simultaneously define an overall tumour volume and the high-uptake sub-volume V1. The inclusion of V1 in the high-risk clinical target volume (CTV HR) and its dosimetric coverage were evaluated using 3D SlicerTM. The average of the 3-4 BT sessions was reported.

RESULTS

Forty-two patients with recurrence after chemoradiotherapy (CRT) for LACC were matched to 42 patients without recurrence. Mean ± standard deviation follow-up was 26 ± 11 months. In the recurrence group, V1 was not included in the CTV HR and not covered by the 85 Gy isodose in 17/42 patients (41%) (1/20 with pelvic recurrence and 16/22 with distant recurrence) and not by the 80 Gy isodose in 7/42 patients (17%) (all with distant recurrence). In the non-recurrence group, V1 was not included in CTV HR and not covered by the 85 Gy isodose in 3 patients only (7%). The hotspots coverage by the 85 Gy isodose was significantly better in patients who did not recur, but only when compared to patients with distant relapse (p < 0.0001).

CONCLUSION

Suboptimal dosimetric coverage of high FDG uptakes on pretherapeutic PET could be associated with an increased risk of recurrence.

Image guidance in radiation therapy for better cure of cancer

The key goal and main challenge of radiation therapy is the elimination of tumors without any concurring damages of the surrounding healthy tissues and organs. Radiation doses required to achieve sufficient cancer-cell kill exceed in most clinical situations the dose that can be tolerated by the healthy tissues, especially when large parts of the affected organ are irradiated. High-precision radiation oncology aims at optimizing tumor coverage, while sparing normal tissues. Medical imaging during the preparation phase, as well as in the treatment room for localization of the tumor and directing the beam, referred to as image-guided radiotherapy (IGRT), is the cornerstone of precision radiation oncology. Sophisticated high-resolution real-time IGRT using X-rays, computer tomography, magnetic resonance imaging, or ultrasound, enables delivery of high radiation doses to tumors without significant damage of healthy organs. IGRT is the most convincing success story of radiation oncology over the last decades, and it remains a major driving force of innovation, contributing to the development of personalized oncology, for example, through the use of real-time imaging biomarkers for individualized dose delivery.

Individualized discrimination of tumor recurrence from radiation necrosis in glioma patients using an integrated radiomics-based model

PURPOSE

To develop and validate an integrated model for discriminating tumor recurrence from radiation necrosis in glioma patients.

METHODS

Data from 160 pathologically confirmed glioma patients were analyzed. The diagnostic model was developed in a primary cohort (n = 112). Textural features were extracted from postoperative ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) positron emission tomography (PET), ¹¹C-methionine (¹¹C-MET) PET, and magnetic resonance images. The least absolute shrinkage and selection operator regression model was used for feature selection and radiomics signature building. Multivariable logistic regression analysis was used to develop a model for predicting tumor recurrence. The radiomics signature, quantitative PET parameters, and clinical risk factors were incorporated in the model. The clinical value of the model was then assessed in an independent validation cohort using the remaining 48 glioma patients.

RESULTS

The integrated model consisting of 15 selected features was significantly associated with postoperative tumor recurrence (p < 0.001 for both primary and validation cohorts). Predictors contained in the individualized diagnosis model included the radiomics signature, the mean of tumor-background ratio (TBR) of ¹⁸F-FDG, maximum of TBR of ¹¹C-MET PET, and patient age. The integrated model demonstrated good discrimination, with an area under the curve (AUC) of 0.988, with a 95% confidence interval (CI) of 0.975-1.000. Application in the validation cohort showed good differentiation (AUC of 0.914 and 95% CI of 0.881-0.945). Decision curve analysis showed that the integrated diagnosis model was clinically useful.

CONCLUSIONS

Our developed model could be used to assist the postoperative individualized diagnosis of tumor recurrence in patients with gliomas.

Impact of Peer Review in Reducing Uncertainty in the Definition of the Lung Target Volume Among Trainee Oncologists

AIMS

To evaluate the impact of peer review and contouring workshops on reducing uncertainty in target volume delineation for lung cancer radiotherapy.

MATERIALS AND METHODS

Data from two lung cancer target volume delineation courses were analysed. In total, 22 trainees in clinical oncology working across different UK centres attended these courses with priori experience in lung cancer radiotherapy. The courses were made up of short presentations and contouring practice sessions. The participants were divided into two groups and asked to first individually delineate (IND) and then individually peer review (IPR) the contours of another participant. The contours were discussed with an expert panel consisting of two consultant clinical oncologists and a consultant radiologist. Contours were analysed quantitatively by measuring the volume and local distance standard deviation (localSD) from the reference expert consensus contour and qualitatively through visual analysis. Feedback from the participants was obtained using a questionnaire.

RESULTS

All participants applied minor editing to the contours during IPR, leading to a non-statistically significant reduction in the mean delineated volume (IND = 140.92 cm3, IPR = 125.26 cm3, P = 0.211). The overall interobserver variation was similar, with a localSD of 0.33 cm and 0.38 cm for the IND and IPR, respectively (P = 0.848). Six participants (29%) carried out correct major changes by either including tumour or excluding healthy tissue. One participant (5%) carried out an incorrect edit by excluding parts of the tumour, while another observer failed to identify a major contour error. The participants' level of confidence in target volume delineation increased following the course and identified the discussions with the radiologist and colleagues as the most important highlights of the course.

CONCLUSION

IPR could improve target volume delineation quality among trainee oncologists by identifying most major contour errors. However, errors were also introduced after IPR, suggesting the need to further discuss major changes with a multidisciplinary team.

2-[18F]FDG PET/CT radiomics in lung cancer: An overview of the technical aspect and its emerging role in management of the disease

Lung cancer is the most common cancer, worldwide, and a major health issue with a remarkable mortality rate. 2-[¹⁸F]fluoro-2-deoxy-D-glucose positron emission tomography/computed tomography (2-[¹⁸F]FDG PET/CT) plays an indispensable role in the management of lung cancer patients. Long-established quantitative parameters such as size, density, and metabolic activity have been and are being employed in the current practice to enhance interpretation and improve diagnostic and prognostic value. The introduction of radiomics analysis revolutionized the quantitative evaluation of medical imaging, revealing data within images beyond visual interpretation. The "big data" are extracted from high-quality images and are converted into information that correlates to relevant genetic, pathologic, clinical, or prognostic features. Technically advanced, diverse methods have been implemented in different studies. The standardization of image acquisition, segmentation and features analysis is still a debated issue. Importantly, a body of features has been extracted and employed for diagnosis, staging, risk stratification, prognostication, and therapeutic response. 2-[¹⁸F]FDG PET/CT-derived features show promising value in non-invasively diagnosing the malignant nature of pulmonary nodules, differentiating lung cancer subtypes, and predicting response to different therapies as well as survival. In this review article, we aimed to provide an overview of the technical aspects used in radiomics analysis in non-small cell lung cancer (NSCLC) and elucidate the role of 2-[¹⁸F]FDG PET/CT-derived radiomics in the diagnosis, prognostication, and therapeutic response.

Radiogenomics Based on PET Imaging

Radiogenomics or imaging genomics is a novel omics strategy of associating imaging data with genetic information, which has the potential to advance personalized medicine. Imaging features extracted from PET or PET/CT enable assessment of in vivo functional and physiological activity and provide comprehensive tumor information non-invasively. However, PET features are considered secondary to features on conventional imaging, and there has not yet been a review of the radiogenomic approach using PET features. This review article summarizes the current state of PET-based radiogenomic research for cancer, which discusses some of its limitations and directions for future study.

Use of Baseline 18 F-FDG PET/CT to Identify Initial Sub-Volumes Associated With Local Failure After Concomitant Chemoradiotherapy in Locally Advanced Cervical Cancer

Introduction: Locally advanced cervical cancer (CC) patients treated by chemoradiotherapy (CRT) have a significant local recurrence rate. The objective of this work was to assess the overlap between the initial high-uptake sub-volume (V1) on baseline 18F-fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) scans and the metabolic relapse (V2) sites after CRT in locally advanced CC. Methods: PET/CT performed before treatment and at relapse in 21 patients diagnosed with LACC and treated with CRT were retrospectively analyzed. CT images at the time of recurrence were registered to baseline CT using the 3D Slicer TM Expert Automated Registration module. The corresponding PET images were then registered using the corresponding transform. The fuzzy locally adaptive Bayesian (FLAB) algorithm was implemented using 3 classes (one for the background and the other two for tumor) in PET1 to simultaneously define an overall tumor volume and the sub-volume V1. In PET2, FLAB was implemented using 2 classes (one for background, one for tumor), in order to define V2. Four indices were used to determine the overlap between V1 and V2 (Dice coefficients, overlap fraction, X = (V1nV2)/V1 and Y = (V1nV2)/V2). Results: The mean (±standard deviation) follow-up was 26 ± 11 months. The measured overlaps between V1 and V2 were moderate to good according to the four metrics, with 0.62-0.81 (0.72 ± 0.05), 0.72-1.00 (0.85 ± 0.10), 0.55-1.00 (0.73 ± 0.16) and 0.50-1.00 (0.76 ± 0.12) for Dice, overlap fraction, X and Y, respectively. Conclusion: In our study, the overlaps between the initial high-uptake sub-volume and the recurrent metabolic volume showed moderate to good concordance. These results now need to be confirmed in a larger cohort using a more standardized patient repositioning procedure for sequential PET/CT imaging, as there is potential for RT dose escalation exploiting the pre-treatment PET high-uptake sub-volume.

Development of a new fully three-dimensional methodology for tumours delineation in functional images

Delineation of tumours in Positron Emission Tomography (PET) plays a crucial role in accurate diagnosis and radiotherapy treatment planning. In this context, it is of outmost importance to devise efficient and operator-independent segmentation algorithms capable of reconstructing the tumour three-dimensional (3D) shape. In previous work, we proposed a system for 3D tumour delineation on PET data (expressed in terms of Standardized Uptake Value - SUV), based on a two-step approach. Step 1 identified the slice enclosing the maximum SUV and generated a rough contour surrounding it. Such contour was then used to initialize step 2, where the 3D shape of the tumour was obtained by separately segmenting 2D PET slices, leveraging the slice-by-slice marching approach. Additionally, we combined active contours and machine learning components to improve performance. Despite its success, the slice marching approach poses unnecessary limitations that are naturally removed by performing the segmentation directly in 3D. In this paper, we migrate our system into 3D. In particular, the segmentation in step 2 is now performed by evolving an active surface directly in the 3D space. The key points of such an advancement are that it performs the shape reconstruction on the whole stack of slices simultaneously, naturally leveraging cross-slice information that could not be exploited before. Additionally, it does not require any specific stopping condition, as the active surface naturally reaches a stable topology once convergence is achieved. Performance of this fully 3D approach is evaluated on the same dataset discussed in our previous work, which comprises fifty PET scans of lung, head and neck, and brain tumours. The results have confirmed that a benefit is indeed achieved in practice for all investigated anatomical districts, both quantitatively, through a set of commonly used quality indicators (dice similarity coefficient >87.66%, Hausdorff distance < 1.48 voxel and Mahalanobis distance < 0.82 voxel), and qualitatively in terms of Likert score (>3 in 54% of the tumours).

Gradient-based generation of intermediate images for heterogeneous tumor segmentation within hybrid PET/MRI scans

Segmentation of tumors from hybrid PET/MRI scans plays an essential role in accurate diagnosis and treatment planning. However, when treating tumors, several challenges, notably heterogeneity and the problem of leaking into surrounding tissues with similar high uptake, have to be considered. To address these issues, we propose an automated method for accurate delineation of tumors in hybrid PET/MRI scans. The method is mainly based on creating intermediate images. In fact, an automatic detection technique that determines a preliminary Interesting Uptake Region (IUR) is firstly performed. To overcome the leakage problem, a separation technique is adopted to generate the final IUR. Then, smart seeds are provided for the Graph Cut (GC) technique to obtain the tumor map. To create intermediate images that tend to reduce heterogeneity faced on the original images, the tumor map gradient is combined with the gradient image. Lastly, segmentation based on the GC_sum^max technique is applied to the generated images. The proposed method has been validated on PET phantoms as well as on real-world PET/MRI scans of prostate, liver and pancreatic tumors. Experimental comparison revealed the superiority of the proposed method over state-of-the-art methods. This confirms the crucial role of automatically creating intermediate images in addressing the problem of wrongly estimating arc weights for heterogeneous targets.

PET segmentation of bulky tumors: Strategies and workflows to improve inter-observer variability

Background

PET-based tumor delineation is an error prone and labor intensive part of image analysis. Especially for patients with advanced disease showing bulky tumor FDG load, segmentations are challenging. Reducing the amount of user-interaction in the segmentation might help to facilitate segmentation tasks especially when labeling bulky and complex tumors. Therefore, this study reports on segmentation workflows/strategies that may reduce the inter-observer variability for large tumors with complex shapes with different levels of user-interaction.

Methods

Twenty PET images of bulky tumors were delineated independently by six observers using four strategies: (I) manual, (II) interactive threshold-based, (III) interactive threshold-based segmentation with the additional presentation of the PET-gradient image and (IV) the selection of the most reasonable result out of four established semi-automatic segmentation algorithms (Select-the-best approach). The segmentations were compared using Jaccard coefficients (JC) and percentage volume differences. To obtain a reference standard, a majority vote (MV) segmentation was calculated including all segmentations of experienced observers. Performed and MV segmentations were compared regarding positive predictive value (PPV), sensitivity (SE), and percentage volume differences.

Results

The results show that with decreasing user-interaction the inter-observer variability decreases. JC values and percentage volume differences of Select-the-best and a workflow including gradient information were significantly better than the measurements of the other segmentation strategies (p-value<0.01). Interactive threshold-based and manual segmentations also result in significant lower and more variable PPV/SE values when compared with the MV segmentation.

Conclusions

FDG PET segmentations of bulky tumors using strategies with lower user-interaction showed less inter-observer variability. None of the methods led to good results in all cases, but use of either the gradient or the Select-the-best workflow did outperform the other strategies tested and may be a good candidate for fast and reliable labeling of bulky and heterogeneous tumors.

Impact of contouring variability on oncological PET radiomics features in the lung

Radiomics features extracted from oncological PET images are currently under intense scrutiny within the context of risk stratification for a variety of cancers. However, the lack of robustness assessment poses problems for their application across institutions and for broader patient populations. The objective of the current study was to examine the extent to which radiomics parameters from oncological PET vary in response to manual contouring variability in lung cancer. Imaging data employed in the study consisted of 26 PET scans with lesions in the lung being created through the use of an anthropomorphic phantom in conjunction with Monte Carlo simulations. From each of the simulated lesions, 25 radiomics features related to the gray-level co-occurrence matrices (GLCOM), gray-level size zone matrices (GLSZM), and gray-level neighborhood difference matrices (GLNDM) were extracted from ground truth contour and from manual contours provided by 10 raters in regard to four intensity discretization schemes with number of gray levels of 32, 64, 128, and 256, respectively. The impact of interrater variability in tumor delineation upon the agreement between raters on radiomics features was examined via interclass correlation and leave-p-out assessment. Only weak and moderate correlations were found between segmentation accuracy as measured by the Dice coefficient and percent feature error from ground truth for the vast majority of the features being examined. GLNDM-based texture parameters emerged as the top performing category of radiomcs features in terms of robustness against contouring variability for discretization schemes engaging number of gray levels of 32, 64, and 128 while GLCOM-based parameters stood out for discretization scheme engaging 256 gray levels. How and to what extent interrater reliability of radiomics features vary in response to the number of raters were largely feature-dependent. It was concluded that impact of contouring variability on PET-based radiomics features is present to varying degrees and could be experienced as a barrier to convey PET-based radiomics research to clinical relevance.

Radiotherapy Planning and Molecular Imaging in Lung Cancer

INTRODUCTION

In patients suitable for radical chemoradiotherapy for lung cancer, 18F-FDGPET/ CT is a proposed management to improve the accuracy of high dose radiotherapy. However, there is a high rate of locoregional failure in patients with locally advanced non-small cell lung cancer (NSCLC), probably due to the fact that standard dosing may not be effective in all patients. The aim of the present review was to address some criticisms associated with the radiotherapy image-guided in NSCLC.

MATERIALS AND METHODS

A systematic literature search was conducted. Only published articles that met the following criteria were included: articles, only original papers, radiopharmaceutical ([18F]FDG and any tracer other than [18F]FDG), target, only specific for lung cancer radiotherapy planning, and experimental design (eventually "in vitro" studies were excluded). Peer-reviewed indexed journals, regardless of publication status (published, ahead of print, in press, etc.) were included. Reviews, case reports, abstracts, editorials, poster presentations, and publications in languages other than English were excluded. The decision to include or exclude an article was made by consensus and any disagreement was resolved through discussion.

RESULTS

Hundred eligible full-text articles were assessed. Diverse information is now available in the literature about the role of FDG and new alternative radiopharmaceuticals for the planning of radiotherapy in NSCLC. In particular, the role of alternative technologies for the segmentation of FDG uptake is essential, although indeterminate for RT planning. The pros and cons of the available techniques have been extensively reported.

CONCLUSION

PET/CT has a central place in the planning of radiotherapy for lung cancer and, in particular, for NSCLC assuming a substantial role in the delineation of tumor volume. The development of new radiopharmaceuticals can help overcome the problems related to the disadvantage of FDG to accumulate also in activated inflammatory cells, thus improving tumor characterization and providing new prognostic biomarkers.

Guidance on the use of PET for treatment planning in radiotherapy clinical trials

The aim of this article is to propose meaningful guidance covering the practical and technical issues involved when planning or conducting clinical trials involving positron emission tomography (PET)-guided radiotherapy. The complexity of imaging requirements will depend on the study aims, design and PET methods used. Where PET is used to adapt radiotherapy, a high level of accuracy and reproducibility is required to ensure effective and safe treatment delivery. The guidance in this document is intended to assist researchers designing clinical trials involving PET-guided radiotherapy to provide sufficient information about the appropriate methods to complete PET-CT imaging to a consistent standard at participating centres. The guidance is divided into six categories: application of PET in radiotherapy, resource requirements, quality assurance, imaging protocol design, data management and image processing. Each section provides an overview of the recent literature to support the specific recommendations. This guidance builds on previous recommendations from the National Cancer Research Institute PET Research Network and has been produced in collaboration with the National Radiotherapy Trials Quality Assurance Group.

Revisiting the identification of tumor sub-volumes predictive of residual uptake after (chemo)radiotherapy: influence of segmentation methods on 18F-FDG PET/CT images

Our aim was to evaluate the impact of the accuracy of image segmentation techniques on establishing an overlap between pre-treatment and post-treatment functional tumour volumes in 18FDG-PET/CT imaging. Simulated images and a clinical cohort were considered. Three different configurations (large, small or non-existent overlap) of a single simulated example was used to elucidate the behaviour of each approach. Fifty-four oesophageal and head and neck (H&N) cancer patients treated with radiochemotherapy with both pre- and post-treatment PET/CT scans were retrospectively analysed. Images were registered and volumes were determined using combinations of thresholds and the fuzzy locally adaptive Bayesian (FLAB) algorithm. Four overlap metrics were calculated. The simulations showed that thresholds lead to biased overlap estimation and that accurate metrics are obtained despite spatially inaccurate volumes. In the clinical dataset, only 17 patients exhibited residual uptake smaller than the pre-treatment volume. Overlaps obtained with FLAB were consistently moderate for esophageal and low for H&N cases across all metrics. Overlaps obtained using threshold combinations varied greatly depending on thresholds and metrics. In both cases overlaps were variable across patients. Our findings do not support optimisation of radiotherapy planning based on pre-treatment 18FDG-PET/CT image definition of high-uptake sub-volumes. Combinations of thresholds may have led to overestimation of overlaps in previous studies.

Quantitative imaging: Erring patterns in manual delineation of PET-imaged lung lesions

BACKGROUND AND PURPOSE

Uncertainty and variability in manual contouring of PET-imaged tumor targets are well recognized; however, the error patterns associated with it were little known and rarely investigated. The present study is aimed to quantitatively assess the erring patterns inherent to manual delineation of PET-imaged lung lesions in a setting with complete ground truth.

MATERIALS AND METHODS

Images being used for assessment consisted of 26 synthetic PET datasets created by using the anthropomorphic Zubal phantom in conjunction with the Monte Carlo based SimSET computational package. Each dataset included one PET-positive lesion differing in shape, dimension, uptake heterogeneity, and anatomical location inside the lung. Target contours were provided by 10 raters and the contour accuracy was evaluated using 12 metrics from five categories - spatial overlap, pair counting, information theory, distance, and volume.

RESULTS

In terms of spatial overlap, manual contouring results intersect substantially with the ground truth whereas tend to oversegment the lesions. Shapes of the segmented tumor volumes are in general geometrically consistent with the ground truth but lack sensitivity in characterizing topographical details. No complete consensus could be achieved between manual contours and the ground truth for any of the given lesions being examined when assessing using pair counting- and informatics-based metrics thus indicating an intrinsic stochastic component of manual contouring. Evaluation based on metrics related to distance and volume demonstrated that it is at the borderline areas between the lesions and the normal tissues where the majority part of manual delineation errors occurred and the extent of volume being identified false positively as cancerous by the raters is appreciable.

CONCLUSION

Quantification of segmentation errors associated with expert manual contouring of PET positive lesion in the lung reveals general patterns in what otherwise might be thought of as randomness. Findings from the current study may allow for the formation of new hypotheses towards improving the accuracy and precision of manual delineation of PET positive tumor targets in the lung.

Introducing FDG PET/CT-guided chemoradiotherapy for stage III NSCLC in low- and middle-income countries: preliminary results from the IAEA PERTAIN trial

Purpose

Patients with stage III non-small-cell lung cancer (NSCLC) treated with chemoradiotherapy (CRT) in low- and middle-income countries (LMIC) continue to have a poor prognosis. It is known that FDG PET/CT improves staging, treatment selection and target volume delineation (TVD), and although its use has grown rapidly, it is still not widely available in LMIC. CRT is often used as sequential treatment, but is known to be more effective when given concurrently. The aim of the PERTAIN study was to assess the impact of introducing FDG PET/CT-guided concurrent CRT, supported by training and quality control (QC), on the overall survival (OS) and progression-free survival (PFS) of patients with stage III NSCLC.

Methods

The study included patients with stage III NSCLC from nine medical centres in seven countries. A retrospective cohort was managed according to local practices between January 2010 and July 2014, which involved only optional diagnostic FDG PET/CT for staging (not for TVD), followed by sequential or concurrent CRT. A prospective cohort between August 2015 and October 2018 was treated according to the study protocol including FDG PET/CT in treatment position for staging and multimodal TVD followed by concurrent CRT by specialists trained in protocol-specific TVD and with TVD QC. Kaplan–Meier analysis was used to assess OS and PFS in the retrospective and prospective cohorts.

Results

Guidelines for FDG PET/CT image acquisition and TVD were developed and published. All specialists involved in the PERTAIN study received training between June 2014 and May 2016. The PET/CT scanners used received EARL accreditation. In November 2018 a planned interim analysis was performed including 230 patients in the retrospective cohort with a median follow-up of 14 months and 128 patients in the prospective cohort, of whom 69 had a follow-up of at least 1 year. Using the Kaplan–Meier method, OS was significantly longer in the prospective cohort than in the retrospective cohort (23 vs. 14 months, p = 0.012). In addition, median PFS was significantly longer in the prospective cohort than in the retrospective cohort (17 vs. 11 months, p = 0.012).

Conclusion

In the PERTAIN study, the preliminary results indicate that introducing FDG PET/CT-guided concurrent CRT for patients with stage III NSCLC in LMIC resulted in a significant improvement in OS and PFS. The final study results based on complete data are expected in 2020.

Electronic supplementary material

The online version of this article (10.1007/s00259-019-04421-5) contains supplementary material, which is available to authorized users.

A phantom study to assess the reproducibility, robustness and accuracy of PET image segmentation methods against statistical fluctuations

Background

Automatic and semi-automatic segmentation methods for PET serve as alternatives to manual delineation and eliminate observer variability. The robustness of these segmentation methods against statistical fluctuations arising from variable size, contrast and noise are vital for providing reliable clinical outcomes for diagnosis and treatment response assessment. In this study, the performances of several segmentation methods have been investigated using the torso NEMA phantom against statistical fluctuations.

Methods

The six hot spheres (0.5-27ml) and the background of the phantom were filled with different activities of ¹⁸F to yield 2:1 and 4:1 contrast ratios. The phantom was scanned on a TrueV PET-CT scanner for 120 minutes. The images were reconstructed using OSEM (4iterations-21subsets) for different durations (15, 20, 34 and 67 minutes) to represent different noise levels and smoothed with a 4-mm Gaussian filter. Each sphere with different settings was delineated using a fixed 40% threshold (40T), fuzzy clustering mean (FCM), adaptive threshold and region based variational (C-V) segmentation methods and compared with the gold standard volume, which was estimated from the known diameter and position of each sphere.

Results

The smallest three spheres at the 2:1 contrast level are not evaluable for the 40T method. For the other spheres, the 40T method grossly overestimates the volumes and the segmented volumes are highly dependent on the statistical variations. These volumes are the least reproducible (80%) with a mean Dice Similarity Coefficient (DSC) of 0.67 and 90% classification error (CE). The other three methods reduce the dependency on noise and contrast in a similar manner by providing low bias (<10%) and CE (<25%) as well as a high DSC (0.88) and reproducibility (30%) for objects >17mm in diameter. However, for the smallest three spheres at a 2:1 contrast level, the performances of all three methods were significantly low, with the adaptive method being superior to the FCM and C-V (mean bias 168% and 350%, mean DSC 0.65 and 0.50, mean CE 227% and 454% for the adaptive and other two methods (approximately similar for FCM and C-V), respectively).

Conclusions

The segmentation accuracy of the fixed threshold-based method depends on size, contrast and noise. The intensity thresholds determined by the adaptive threshold methods are less sensitive to noise and therefore, the segmented volumes are more reproducible across different acquisition durations. A similar performance can be achieved with the FCM and C-V methods. Though, for small lesions (< 2cm diameter) with low counts and contrast, the adaptive threshold-based method outperforms the FCM and C-V methods, and the performance of neither of these methods is optimal for volumes <2cm in diameter. These three methods can only reliably be used to delineate tumours for diagnostic and monitoring purposes provided that the contrast between the tumour and background is not below a 2:1 ratio and the size of the tumour does not fall not below 2cm in diameter in response to treatment. They can also be used for different radiotracers with variable uptake. However, the FCM and C-V methods have the advantage of not requiring calibrations for different scanners and settings.

Radiomics in nuclear medicine: robustness, reproducibility, standardization, and how to avoid data analysis traps and replication crisis

Radiomics in nuclear medicine is rapidly expanding. Reproducibility of radiomics studies in multicentre settings is an important criterion for clinical translation. We therefore performed a meta-analysis to investigate reproducibility of radiomics biomarkers in PET imaging and to obtain quantitative information regarding their sensitivity to variations in various imaging and radiomics-related factors as well as their inherent sensitivity. Additionally, we identify and describe data analysis pitfalls that affect the reproducibility and generalizability of radiomics studies. After a systematic literature search, 42 studies were included in the qualitative synthesis, and data from 21 were used for the quantitative meta-analysis. Data concerning measurement agreement and reliability were collected for 21 of 38 different factors associated with image acquisition, reconstruction, segmentation and radiomics-specific processing steps. Variations in voxel size, segmentation and several reconstruction parameters strongly affected reproducibility, but the level of evidence remained weak. Based on the meta-analysis, we also assessed inherent sensitivity to variations of 110 PET image biomarkers. SUV_mean and SUV_max were found to be reliable, whereas image biomarkers based on the neighbourhood grey tone difference matrix and most biomarkers based on the size zone matrix were found to be highly sensitive to variations, and should be used with care in multicentre settings. Lastly, we identify 11 data analysis pitfalls. These pitfalls concern model validation and information leakage during model development, but also relate to reporting and the software used for data analysis. Avoiding such pitfalls is essential for minimizing bias in the results and to enable reproduction and validation of radiomics studies.

Impact of intensity discretization on textural indices of [18F]FDG-PET tumour heterogeneity in lung cancer patients

Quantifying tumour heterogeneity from [¹⁸F]FDG-PET images promises benefits for treatment selection of cancer patients. Here, the calculation of texture parameters mandates an initial discretization step (binning) to reduce the number of intensity levels. Typically, three types of discrimination methods are used: lesion relative resampling (LRR) with fixed bin number, lesion absolute resampling (LAR) and absolute resampling (AR) with fixed bin widths. We investigated the effects of varying bin widths or bin number using 27 commonly cited local and regional texture indices (TIs) applied on lung tumour volumes. The data set were extracted from 58 lung cancer patients, with three different and robust tumour segmentation methods. In our cohort, the variations of the mean value as the function of the bin widths were similar for TIs calculated with LAR and AR quantification. The TI histograms calculated by LRR method showed distinct behaviour and its numerical values substantially effected by the selected bin number. The correlations of the AR and LAR based TIs demonstrated no principal differences between these methods. However, no correlation was found for the interrelationship between the TIs calculated by LRR and LAR (or AR) discretization method. Visual classification of the texture was also performed for each lesion. This classification analysis revealed that the parameters show statistically significant correlation with the visual score, if LAR or AR discretization method is considered, in contrast to LRR. Moreover, all the resulted tendencies were similar regardless the segmentation methods and the type of textural features involved in this work.

Machine-learned target volume delineation of 18F-FDG PET images after one cycle of induction chemotherapy

Biological tumour volume (GTV_PET) delineation on ¹⁸F-FDG PET acquired during induction chemotherapy (ICT) is challenging due to the reduced metabolic uptake and volume of the GTV_PET. Automatic segmentation algorithms applied to ¹⁸F-FDG PET (PET-AS) imaging have been used for GTV_PET delineation on ¹⁸F-FDG PET imaging acquired before ICT. However, their role has not been investigated in ¹⁸F-FDG PET imaging acquired after ICT. In this study we investigate PET-AS techniques, including ATLAAS a machine learned method, for accurate delineation of the GTV_PET after ICT. Twenty patients were enrolled onto a prospective phase I study (FiGaRO). PET/CT imaging was acquired at baseline and 3 weeks following 1 cycle of induction chemotherapy. The GTV_PET was manually delineated by a nuclear medicine physician and clinical oncologist. The resulting GTV_PET was used as the reference contour. The ATLAAS original statistical model was expanded to include images of reduced metabolic activity and the ATLAAS algorithm was re-trained on the new reference dataset. Estimated GTV_PET contours were derived using sixteen PET-AS methods and compared to the GTV_PET using the Dice Similarity Coefficient (DSC). The mean DSC for ATLAAS, 60% Peak Thresholding (PT60), Adaptive Thresholding (AT) and Watershed Thresholding (WT) was 0.72, 0.61, 0.63 and 0.60 respectively. The GTV_PET generated by ATLAAS compared favourably with manually delineated volumes and in comparison, to other PET-AS methods, was more accurate for GTV_PET delineation after ICT. ATLAAS would be a feasible method to reduce inter-observer variability in multi-centre trials.

Deep Learning for Variational Multimodality Tumor Segmentation in PET/CT

Positron emission tomography/computed tomography (PET/CT) imaging can simultaneously acquire functional metabolic information and anatomical information of the human body. How to rationally fuse the complementary information in PET/CT for accurate tumor segmentation is challenging. In this study, a novel deep learning based variational method was proposed to automatically fuse multimodality information for tumor segmentation in PET/CT. A 3D fully convolutional network (FCN) was first designed and trained to produce a probability map from the CT image. The learnt probability map describes the probability of each CT voxel belonging to the tumor or the background, and roughly distinguishes the tumor from its surrounding soft tissues. A fuzzy variational model was then proposed to incorporate the probability map and the PET intensity image for an accurate multimodality tumor segmentation, where the probability map acted as a membership degree prior. A split Bregman algorithm was used to minimize the variational model. The proposed method was validated on a non-small cell lung cancer dataset with 84 PET/CT images. Experimental results demonstrated that: 1). Only a few training samples were needed for training the designed network to produce the probability map; 2). The proposed method can be applied to small datasets, normally seen in clinic research; 3). The proposed method successfully fused the complementary information in PET/CT, and outperformed two existing deep learning-based multimodality segmentation methods and other multimodality segmentation methods using traditional fusion strategies (without deep learning); 4). The proposed method had a good performance for tumor segmentation, even for those with Fluorodeoxyglucose (FDG) uptake inhomogeneity and blurred tumor edges (two major challenges in PET single modality segmentation) and complex surrounding soft tissues (one major challenge in CT single modality segmentation), and achieved an average dice similarity indexes (DSI) of 0.86 ± 0.05, sensitivity (SE) of 0.86 ± 0.07, positive predictive value (PPV) of 0.87 ± 0.10, volume error (VE) of 0.16 ± 0.12, and classification error (CE) of 0.30 ± 0.12.