Search

Background and purpose Transperineal ultrasound (TPUS) is used clinically for directly assessing prostate motion. Factors affecting accuracy and precision in TPUS motion estimation must be assessed to realise its full potential. Methods and materials Patients were imaged using volumetric TPUS during the Clarity-Pro trial (NCT02388308). Prostate motion was measured online at patient set-up and offline by experienced observers. Cone beam CT with markers was used as a comparator and observer performance was also quantified. The influence of different clinical factors was examined to establish specific recommendations towards efficacious ultrasound guided radiotherapy. Results From 330 fractions in 22 patients, offline observer random errors were 1.5 mm, 1.3 mm, 1.9 mm (left–right, superior-inferior, anteroposterior respectively). Errors increased in fractions exhibiting poor image quality to 3.3 mm, 3.3 mm and 6.8 mm. Poor image quality was associated with inconsistent probe placement, large anatomical changes and unfavourable imaging conditions within the patient. Online matching exhibited increased observer errors of: 3.2 mm, 2.9 mm and 4.7 mm. Four patients exhibited large systematic residual errors, of which three had poor quality images. Patient habitus showed no correlation with observer error, residual error, or image quality. Conclusions TPUS offers the unique potential to directly assess inter- and intra-fraction motion on conventional linacs. Inconsistent image quality, inexperienced operators and the pressures of the clinical environment may degrade precision and accuracy. Experienced operators are essential and cross-centre standards for training and QA should be established that build upon current guidance. Greater use of automation technologies may further minimise uncertainties.

Highlights • Automated matching improves accuracy and precision of ultrasound guided radiotherapy. • Image registration software lowers setup errors and sonography training requirements. • Spatial regularisation improves registration algorithm performance., Background and purpose Daily image guidance is standard care for prostate radiotherapy. Innovations which improve the accuracy and efficiency of ultrasound guidance are needed, particularly with respect to reducing interobserver variation. This study explores automation tools for this purpose, demonstrated on the Elekta Clarity Autoscan®. The study was conducted as part of the Clarity-Pro trial (NCT02388308). Materials and methods Ultrasound scan volumes were collected from 32 patients. Prostate matches were performed using two proposed workflows and the results compared with Clarity’s proprietary software. Gold standard matches derived from manually localised landmarks provided a reference. The two workflows incorporated a custom 3D image registration algorithm, which was benchmarked against a third-party application (Elastix). Results Significant reductions in match errors were reported from both workflows compared to standard protocol. Median (IQR) absolute errors in the left–right, anteroposterior and craniocaudal axes were lowest for the Manually Initiated workflow: 0.7(1.0) mm, 0.7(0.9) mm, 0.6(0.9) mm compared to 1.0(1.7) mm, 0.9(1.4) mm, 0.9(1.2) mm for Clarity. Median interobserver variation was ≪0.01 mm in all axes for both workflows compared to 2.2 mm, 1.7 mm, 1.5 mm for Clarity in left–right, anteroposterior and craniocaudal axes. Mean matching times was also reduced to 43 s from 152 s for Clarity. Inexperienced users of the proposed workflows attained better match precision than experienced users on Clarity. Conclusion Automated image registration with effective input and verification steps should increase the efficacy of interfraction ultrasound guidance compared to the current commercially available tools.

Ultrasound imaging is a commonly used technology for visualising patient anatomy in real-time during diagnostic and therapeutic procedures. High operator dependency and low reproducibility make ultrasound imaging and interpretation challenging with a steep learning curve. Automatic image classification using deep learning has the potential to overcome some of these challenges by supporting ultrasound training in novices, as well as aiding ultrasound image interpretation in patient with complex pathology for more experienced practitioners. However, the use of deep learning methods requires a large amount of data in order to provide accurate results. Labelling large ultrasound datasets is a challenging task because labels are retrospectively assigned to 2D images without the 3D spatial context available in vivo or that would be inferred while visually tracking structures between frames during the procedure. In this work, we propose a multi-modal convolutional neural network (CNN) architecture that labels endoscopic ultrasound (EUS) images from raw verbal comments provided by a clinician during the procedure. We use a CNN composed of two branches, one for voice data and another for image data, which are joined to predict image labels from the spoken names of anatomical landmarks. The network was trained using recorded verbal comments from expert operators. Our results show a prediction accuracy of 76% at image level on a dataset with 5 different labels. We conclude that the addition of spoken commentaries can increase the performance of ultrasound image classification, and eliminate the burden of manually labelling large EUS datasets necessary for deep learning applications., Submitted to IEEE TMI

Endoscopic ultrasound (EUS) is a challenging procedure that requires skill, both in endoscopy and ultrasound image interpretation. Classification of key anatomical landmarks visible on EUS images can assist the gastroenterologist during navigation. Current applications of deep learning have shown the ability to automatically classify ultrasound images with high accuracy. However, these techniques require a large amount of labelled data which is time consuming to obtain, and in the case of EUS, is also a difficult task to perform retrospectively due to the lack of 3D context. In this paper, we propose the use of an image-to-image translation method to create synthetic EUS (sEUS) images from CT data, that can be used as a data augmentation strategy when EUS data is scarce. We train a cycle-consistent adversarial network with unpaired EUS images and CT slices extracted in a manner such that they mimic plausible EUS views, to generate sEUS images from the pancreas, aorta and liver. We quantitatively evaluate the use of sEUS images in a classification sub-task and assess the Frechet Inception Distance. We show that synthetic data, obtained from CT data, imposes only a minor classification accuracy penalty and may help generalization to new unseen patients. The code and a dataset containing generated sEUS images are available at: https://ebonmati.github.io.

Effective transperineal ultrasound image guidance in prostate external beam radiotherapy requires consistent alignment between probe and prostate at each session during patient set-up. Probe placement and ultrasound image interpretation are manual tasks contingent upon operator skill, leading to interoperator uncertainties that degrade radiotherapy precision. We demonstrate a method for ensuring accurate probe placement through joint classification of images and probe position data. Using a multi-input multi-task algorithm, spatial coordinate data from an optically tracked ultrasound probe is combined with an image classifier using a recurrent neural network to generate two sets of predictions in real-time. The first set identifies relevant prostate anatomy visible in the field of view using the classes: outside prostate, prostate periphery, prostate centre. The second set recommends a probe angular adjustment to achieve alignment between the probe and prostate centre with the classes: move left, move right, stop. The algorithm was trained and tested on 9,743 clinical images from 61 treatment sessions across 32 patients. We evaluated classification accuracy against class labels derived from three experienced observers at 2/3 and 3/3 agreement thresholds. For images with unanimous consensus between observers, anatomical classification accuracy was 97.2% and probe adjustment accuracy was 94.9%. The algorithm identified optimal probe alignment within a mean (standard deviation) range of 3.7° (1.2°) from angle labels with full observer consensus, comparable to the 2.8° (2.6°) mean interobserver range. We propose such an algorithm could assist radiotherapy practitioners with limited experience of ultrasound image interpretation by providing effective real-time feedback during patient set-up.

DeepReg (https://github.com/DeepRegNet/DeepReg) is a community-supported open-source toolkit for research and education in medical image registration using deep learning., Comment: Accepted in The Journal of Open Source Software (JOSS)

Radiotherapy (RT) aims to deliver a spatially conformal dose of radiation to tumours while maximizing the dose sparing to healthy tissues. However, the internal patient anatomy is constantly moving due to respiratory, cardiac, gastrointestinal and urinary activity. The long term goal of the RT community to ‘see what we treat, as we treat’ and to act on this information instantaneously has resulted in rapid technological innovation. Specialized treatment machines, such as robotic or gimbal-steered linear accelerators (linac) with in-room imaging suites, have been developed specifically for real-time treatment adaptation. Additional equipment, such as stereoscopic kilovoltage (kV) imaging, ultrasound transducers and electromagnetic transponders, has been developed for intrafraction motion monitoring on conventional linacs. Magnetic resonance imaging (MRI) has been integrated with cobalt treatment units and more recently with linacs. In addition to hardware innovation, software development has played a substantial role in the development of motion monitoring methods based on respiratory motion surrogates and planar kV or Megavoltage (MV) imaging that is available on standard equipped linacs. In this paper, we review and compare the different intrafraction motion monitoring methods proposed in the literature and demonstrated in real-time on clinical data as well as their possible future developments. We then discuss general considerations on validation and quality assurance for clinical implementation. Besides photon RT, particle therapy is increasingly used to treat moving targets. However, transferring motion monitoring technologies from linacs to particle beam lines presents substantial challenges. Lessons learned from the implementation of real-time intrafraction monitoring for photon RT will be used as a basis to discuss the implementation of these methods for particle RT.