Search

Target volume delineation (TVD) has been identified as a weakness in the accuracy of radiotherapy, both within and outside of clinical trials due to the intra/interobserver variations affecting the TVD quality. Sources of variations such as poor compliance or protocol violation may have adverse effect on treatment outcomes. In this paper, we present and describe the FIELDFor each site-specific clinical case included in the FIELDResults show the trainees' improvement when reoutlining a case after reviewing the feedback generated from the FIELDWe presented a new open-source software to support education in TVD and ongoing continuous professional development for clinical oncology trainees and consultants. ARENA in combination with FIELDFIELD

Objectives: Target volume delineation (TVD) has been identified as a weakness in the accuracy of radiotherapy, both within and outside of clinical trials due to the intra/interobserver variations affecting the TVD quality. Sources of variations such as poor compliance or protocol violation may have adverse effect on treatment outcomes. In this paper, we present and describe the FIELDRT software developed for the ARENA project to improve the quality of TVD through qualitative and quantitative feedbacks and individual and personalized summary of trainee”s performance. Methods: For each site-specific clinical case included in the FIELDRT software, reference volumes, minimum and maximum “acceptable” volumes and organ at risk were derived by outlines of consultants and senior trainees. The software components currently developed include: (a) user-friendly importing interface (b) analysis toolbox to compute quantitative and qualitative (c) visualiser and (d) structured report generator for personalised feedback. The FIELDRT software was validated by comparing the performance of 63 trainees and by measuring performance over time. In addition, a trainee evaluation day was held in 2019 to collect feedback on FIELDRT. Results: Results show the trainees’ improvement when reoutlining a case after reviewing the feedback generated from the FIELDRT software. Comments and feedback received after evaluation day were positive and confirmed that FIELDRT can be a useful application for training purposes. Conclusion: We presented a new open-source software to support education in TVD and ongoing continuous professional development for clinical oncology trainees and consultants. ARENA in combination with FIELDRT implements site-specific modules with reference target and organs at risk volumes and automatically evaluates individual performance using several quantitative and qualitative feedbacks. Pilot results suggests this software could be used as an education tool to reduce variation in TVD so to guarantee high quality in radiotherapy. Advances in knowledge: FIELDRT is a new easy and free to use software aiming at supporting education in TVD and ongoing continuous professional development. The software provides quantitative/qualitative feedback and an exportable report with an individual and personalised summary of trainee’s performance.

Objective\ud\udStatistical shape modelling (SSM) has established as a powerful method for segmenting the left ventricle in cardiac magnetic resonance (CMR) images However, applying them to segment the right ventricle (RV) is not straightforward because of the complex structure of this chamber. Our aim was to develop a new inter-modality SSM-based approach to detect the RV endocardium in CMR data.\udMethods\ud\udReal-time transthoracic 3D echocardiographic (3DE) images of 219 retrospective patients were used to populate a large database containing 4347 3D RV surfaces and train a model. The initial position, orientation and scale of the model in the CMR stack were semi-automatically derived. The detection process consisted in iteratively deforming the model to match endocardial borders in each CMR plane until convergence was reached. Clinical values obtained with the presented SSM method were compared with gold-standard (GS) corresponding parameters.\udResults\ud\udCMR images of 50 patients with different pathologies were used to test the proposed segmentation method. Average processing time was 2 min (including manual initialization) per patient. High correlations (r2 > 0.76) and not significant bias (Bland-Altman analysis) were observed when evaluating clinical parameters. Quantitative analysis showed high values of Dice coefficient (0.87 ± 0.03), acceptable Hausdorff distance (9.35 ± 1.51 mm) and small point-to-surface distance (1.91 ± 0.26 mm). Conclusion\ud\udA novel SSM-based approach to segment the RV endocardium in CMR scans by using a model trained on 3DE-derived RV endocardial surfaces, was proposed. This inter-modality technique proved to be rapid when segmenting the RV endocardium with an accurate anatomical delineation, in particular in apical and basal regions.

International audience; We evaluate in this paper different strategies for the construction of a statistical shape model (SSM) of the left ventricle (LV) to be used for segmentation in cardiac magnetic resonance (CMR) images. From a large database of LV surfaces obtained throughout the cardiac cycle from 3D echocardiographic (3DE) LV images, different LV shape models were built by varying the considered phase in the cardiac cycle and the registration procedure employed for surface alignment. Principal component analysis was computed to describe the statistical variability of the SSMs, which were then deformed by applying an active shape model (ASM) approach to segment the LV endocardium in CMR images of 45 patients. Segmentation performance was evaluated by comparing LV volumes derived by ASM segmentation with different SSMs and those obtained by manual tracing, considered as a reference. A high correlation (r2 > 0.92) was found in all cases, with better results when using the SSM models comprising more than one frame of the cardiac cycle.

The aim of this work was to investigate radiomic analysis of contrast and non-contrast enhanced planning CT images of oesophageal cancer (OC) patients in terms of stability, dimensionality and contrast agent dependency. The prognostic significance of CT-based radiomic features was also evaluated. Different 2D and 3D radiomic features were extracted from contrast and non-contrast enhanced CT images of 213 patients from the multi-centre SCOPE1 randomised controlled trial (RCT) in OC. Feature stability was evaluated by randomly dividing patients into three groups and identifying textures with similar distributions among groups with a Kruskal-Wallis analysis. A paired two-sided Wilcoxon signed rank test was used to assess for significant differences in the remaining corresponding 2D and 3D stable features. A prognostic model was constructed using clinical characteristics and remaining filtered features. The discriminative ability of significant variables was tested using Kaplan-Meier analysis. A total of 238 2D and 3D radiomic features were computed from oesophageal CT images. More than 75 features were stable if extracted from homogeneous cohort (contrast or non-contrast enhanced CT images) and inhomogeneous cohort (contrast and non-contrast enhanced CT images). Among the remaining corresponding stable features computed from both cohorts, only 4 features did not show a statistically significant difference if obtained in 2D or in 3D (p-value < 0.05). A Cox regression model constructed using 5 clinical variables (age, sex, tumour, node and metastasis (TNM) stage, WHO performance status and contrast administration) and 4 radiomic variables (inverse varianceGLCM, large distance emphasisGLDZM, zone distance non uniformity normGLDZM, zone distance varianceGLDZM), identified one radiomic feature (zone distance varianceGLDZM) that was significantly associated with overall survival (p-value = 0.032, HR = 1.25, 95% CI = 1.02–1.52). A significant difference in overall survival between groups was found when considering a threshold of zone distance varianceGLDZM equals to 1.70 (X2 = 7.692, df = 1, p-value = 0.006). Zone distance varianceGLDZM was identified as the only stable CT radiomic feature statistically correlated with overall survival, independent of dimensionality and contrast administration. This feature was able to identify high-risk patients and if validated, could be the subject of a future clinical trial aiming to improve clinical decision making and personalise OC treatment.

Object We present in this paper the application of a statistical shape model of the left ventricle (LV) built from transthoracic real time 3D echocardiography (3DE) to segment the LV endocardium and epicardium in cardiac magnetic resonance (CMR) images. Material and methods The LV model was built from a training database constituted by over 9000 surfaces obtained from retrospectively selected 3DE examination of 435 patients with various pathologies. Three-dimensional segmentation of the endocardium and the epicardium was obtained by processing CMR images acquired in 30 patients with a dedicated active shape modelling (ASM) algorithm using the proposed LV model. Results The segmentation results obtained with the proposed method were compared with those obtained by the manual reference technique; similarity was proven by computing: i) point to surface distance ( 89%), iii) Hausdorff distance ( ∼ 5 mm). This was furthermore confirmed by equivalence testing, linear regression and Bland Altman analysis applied on derived clinical parameters, such as LV volumes and mass. Conclusions This study showed the potential usefulness of the proposed inter-modal ASM approach featuring a 3DE-based LV model for the 3D segmentation of the LV myocardium in CMR images.

Recent advances in imaging technology have enabled the non-invasive study of the structure and the function of the heart, the valves and the vascular system. Different imaging modalities are routinely used to provide specific and complementary diagnostic and prognostic information. Computerized analysis plays a crucial role to in quantifying cardiac function from non-invasive imaging. To this respect, model-based techniques, such as statistical shape models (SSMs), have become a popular solution for the detection of different cardiac structures. In this two-steps approach, a statistical model, trained on a set of samples to encode the morphology and the statistical variability of the structure of interest, is applied to segment the same structure in new images constraining the possible deformations only to plausible shapes observed in the training set. The aim of this chapter is to give a summary of the current state-of-the-art of SSMs in cardiac imaging. In particular, the most relevant and recent SSMs applications proposed for a specific structure (left ventricle, right ventricle, atria and valves) or more cardiac structures together (left and right ventricles, four chambers and entire heart) will be discussed. Furthermore, the potential usefulness of this technique as well as its robustness when applied to different imaging modalities are reviewed.

Statistical shape models (SSMs) represent a powerful tool used in patient-specific modeling to segment medical images because they incorporate a-priori knowledge that guide the model during deformation. Our aim was to evaluate segmentation accuracy in terms of left ventricular (LV) volumes obtained using four different SSMs versus manual gold standard tracing on cardiac magnetic resonance (CMR) images. A database of 3D echocardiographic (3DE) LV surfaces obtained in 435 patients was used to generate four different SSMs, based on cardiac phase selection. Each model was scaled and deformed to detect LV endocardial contours in the end-diastolic (ED) and end-systolic (ES) frames of a CMR short-axis (SAX) stack for 15 patients with normal LV function. Linear correlation and Bland-Altman analyses versus gold-standard showed in all cases high correlation (r2>0.95), non-significant biases and narrow limits of agreement.

A robust and efficient approach to reconstruction of the descending thoracic aorta from contiguous 3-D transesophageal echocardiographic (TEE) images is proposed. An ad hoc image acquisition protocol was designed to acquire ordered and partially overlapped 3-D TEE data sets, followed by dedicated image processing to align and fuse all acquired data sets. Alignment strategy implemented pairwise rigid registration guided by a priori knowledge, and it was validated using artificially misaligned images. Image fusion was finally performed to enable visualization and analysis of extended field-of-view of the acquired aorta. The application of different fusion techniques was also investigated. The method was applied to a population of 17 consecutive patients. Qualitative and quantitative results supported the feasibility and accuracy of the proposed approach. In a clinical scenario, its application could allow the quantitative assessment of aortic plaque burden in the descending thoracic aorta from 3-D TEE images.

Aims To propose a nearly automated left ventricular (LV) three-dimensional (3D) surface segmentation procedure, based on active shape modelling (ASM) and built on a database of 3D echocardiographic (3DE) LV surfaces, for cardiac magnetic resonance (CMR) images, and to test its accuracy for LV volumes computation compared with ‘gold standard’ manual tracings and discs-summation method. Methods and results The ASM was created based on segmented LV surfaces (4D LV analysis, Tomtec) from 3DE datasets of 205 patients. Then, it was applied to the cardiac magnetic resonance imaging short-axis (SAX) images stack of 12 consecutive patients. After proper realignment using two- and four-chambers CMR long-axis views both as reference and for initializing LV apex and base (six points in total), the ASM was iteratively and automatically updated to match the information of all the SAX planes contemporaneously, resulting in an endocardial LV 3D mesh from which volume was directly derived. The same CMR images were analysed by an experienced cardiologist to derive end-diastolic and end-systolic volumes. Linear correlation and Bland–Altman analyses were applied vs. the manual ‘gold standard’. Active shape modelling results showed high correlations with manual values both for LV volumes ( r 2 > 0.98) and ejection fraction (EF) ( r 2 > 0.90), non-significant biases and narrow limits of agreement. Conclusion The proposed method resulted in accurate detection of 3D LV endocardial surfaces, which lead to fast and reliable measurements of LV volumes and EF when compared with manual tracing of CMR SAX images. The segmented 3D mesh, including a realistic LV apex and base, could constitute a novel starting point for more realistic patient-specific finite element modelling.

Background Aortic atherosclerosis is a risk factor for cerebrovascular events. Two-dimensional transesophageal echocardiographic quantification of descending aortic plaques is time-consuming and underestimates plaque burden. The aim of this study was to assess the feasibility and accuracy of a novel semiautomated program that uses three-dimensional (3D) transesophageal echocardiography to identify and quantify aortic plaque severity as determined by plaque thickness, volume, and number. The relationship between maximum plaque thickness and volume was also examined. Methods Descending aortic 3D transesophageal echocardiographic images from 58 patients were analyzed for plaque thickness, volume, and number using semiautomated custom software. The reference standard was manual assessment by an expert reader using 3D multiplanar reconstructions. Agreement and κ values were calculated to determine the program's accuracy against the reference standard. Correlation and bias were examined using linear regression and Bland-Altman statistics. Pearson's correlation was used to examine the relationship between maximum plaque thickness and volume. Results Analysis was possible in all patients. Overall agreement for the absolute presence or absence of plaque per patient was 95%. Agreement regarding the number of plaques per patient and plaque severity was high at 95% and 85%, respectively. Plaque volume was slightly underestimated by the program compared with manual measurements. The correlation between plaque thickness and volume was 0.56. Conclusions The results of this study demonstrate that semiautomated plaque analysis of 3D transesophageal echocardiographic descending aortic data sets is feasible and accurate in determining plaque severity as measured by plaque thickness, volume, and number. This methodology allows the standardization of plaque quantification, which will improve its utility in clinical trials. A greater understanding of the importance of plaque thickness versus volume is needed.

Statistical shape modelling (SSM) approaches have been proposed as a powerful tool to segment the left ventricle in cardiac magnetic resonance (CMR) images. Our aim was to extend this method to segment the RV cavity in CMR images and validate it compared to the conventional gold-standard (GS) manual tracing. A SSM of the RV was built using a database of 4347 intrinsically 3D surfaces, extracted from transthoracic 3D echo cardiographic (3DE) images of 219 retrospective patients. The SSM was then scaled and deformed on the base of some features extracted, with different strategies, from each short-axis plane until a stable condition was reached. The proposed approach, tested on 14 patients, resulted in a high correlation (r2=0.97) and narrow limits of agreement (± 17% error) when comparing the semiautomatic volumes to the GS, confirming the accuracy of this approach in segmenting the RV endocardium.

Aortic atherosclerosis is an important risk factor for embolic cerebrovascular events. Quantification of descending aorta atheromas with 2D transesophageal echocardiography (TEE) is time-consuming and likely underestimates the 3D plaque burden. We aimed to assess the feasibility and accuracy of a