MRI and CBCT for lymph node identification and registration in patients with NSCLC undergoing radical radiotherapy

Deep learning-based lung image registration: A review

Lung image registration can effectively describe the relative motion of lung tissues, thereby helping to solve series problems in clinical applications. Since the lungs are soft and fairly passive organs, they are influenced by respiration and heartbeat, resulting in discontinuity of lung motion and large deformation of anatomic features. This poses great challenges for accurate registration of lung image and its applications. The recent application of deep learning (DL) methods in the field of medical image registration has brought promising results. However, a versatile registration framework has not yet emerged due to diverse challenges of registration for different regions of interest (ROI). DL-based image registration methods used for other ROI cannot achieve satisfactory results in lungs. In addition, there are few review articles available on DL-based lung image registration. In this review, the development of conventional methods for lung image registration is briefly described and a more comprehensive survey of DL-based methods for lung image registration is illustrated. The DL-based methods are classified according to different supervision types, including fully-supervised, weakly-supervised and unsupervised. The contributions of researchers in addressing various challenges are described, as well as the limitations of these approaches. This review also presents a comprehensive statistical analysis of the cited papers in terms of evaluation metrics and loss functions. In addition, publicly available datasets for lung image registration are also summarized. Finally, the remaining challenges and potential trends in DL-based lung image registration are discussed.

What do we want? Training! When do we want it? Now? A training needs analysis for adaptive radiotherapy for therapeutic radiographers

INTRODUCTION

Therapeutic radiographers (TRs) have adapted to the changing requirements and demands of the oncology service and in response to advanced techniques such as on-line adaptive MRI-guided radiotherapy (MRIgRT). The skills required for MRIgRT would benefit many TRs not just those involved in this technique. This study presents the results of a training needs analysis (TNA) for the required MRIgRT skills in readiness for training TRs for current and future practice.

METHODS

A UK-based TNA was used to ask TRs about their knowledge and experience with essential skills required for MRIgRT based on previous investigations into the topic. A five-point Likert scale was used for each of the skills and the difference in values were used to calculate the training need for current and future practice.

RESULTS

261 responses were received (n = 261). The skill rated the most important to current practice was CBCT/CT matching and/or fusion. The current highest priority needs were radiotherapy planning and radiotherapy dosimetry. The skill rated the most important to future practice was CBCT/CT matching and/or fusion. The future highest priority needs were MRI acquisition and MRI Contouring. Over 50% of participants wanted training or additional training in all skills. There was an increase in all values for skills investigated from current to future roles.

CONCLUSION

Although the examined skills were viewed as important to current roles, the future training needs, both overall and high priority, were different compared to current roles. As the 'future' of radiotherapy can arrive rapidly, it is essential that training is delivered appropriately and timely. Before this can occur, there must be investigations into the method and delivery of this training.

IMPLICATIONS FOR PRACTICE

Role development. Education changes for therapeutic radiographers.

Feasibility of a deep-learning based anatomical region labeling tool for Cone-Beam Computed Tomography scans in radiotherapy

Background and purpose

Currently, there is no robust indicator within the Cone-Beam Computed Tomography (CBCT) DICOM headers as to which anatomical region is present on the scan. This can be a predicament to CBCT-based algorithms trained on specific body regions, such as auto-segmentation and radiomics tools used in the radiotherapy workflow. We propose an anatomical region labeling (ARL) algorithm to classify CBCT scans into four distinct regions: head & neck, thoracic-abdominal, pelvis, and extremity.

Materials and methods

Algorithm training and testing was performed on 3,802 CBCT scans from 596 patients treated at our radiotherapy center. The ARL model, which consists of a convolutional neural network, makes use of a single CBCT coronal slice to output a probability of occurrence for each of the four classes. ARL was evaluated on the test dataset composed of 1,090 scans and compared to a support vector machine (SVM) model. ARL was also used to label CBCT treatment scans for 22 consecutive days as part of a proof-of-concept implementation. A validation study was performed on the first 100 unique patient scans to evaluate the functionality of the tool in the clinical setting.

Results

ARL achieved an overall accuracy of 99.2% on the test dataset, outperforming the SVM (91.5% accuracy). Our validation study has shown strong agreement between the human annotations and ARL predictions, with accuracies of 99.0% for all four regions.

Conclusion

The high classification accuracy demonstrated by ARL suggests that it may be employed as a pre-processing step for site-specific, CBCT-based radiotherapy tools.

The impact of an educational tool in cervix image registration across three imaging modalities

OBJECTIVES

Accurate image registration is vital in cervical cancer where changes in both planning target volume (PTV) and organs at risk (OARs) can make decisions regarding image registration complicated. This work aims to determine the impact of a dedicated educational tool compared with experience gained in MR-guided radiotherapy (MRgRT).

METHODS

10 therapeutic radiographers acted as observers and were split into two groups based on previous experience with MRgRT and Monaco treatment planning system. Three CBCT-CT, three MR-CT and two MR-MR registrations were completed per patient by each observer. Observers recorded translations, time to complete image registration and confidence. Data were collected in two phases; prior to and following the introduction of a cervix registration guide.

RESULTS

No statistically significant differences were noted between imaging modalities. Each group was assessed independently pre- and post-education, no statistically significant differences were noted in either CBCT-CT or MR-CT imaging. Group 1 MR-MR imaging showed a statistically significant reduction in interobserver variability (p=0.04), in Group 2, the result was not statistically significant (p=0.06). Statistically significant increases in confidence were seen in all three modalities (p≤0.05).

CONCLUSIONS

At The Christie NHS Foundation Trust, radiographers consistently registered images across three different imaging modalities regardless of their previous experience. The implementation of an image registration guide had limited impact on inter- and intraobserver variability. Radiographers' confidence showed statistically significant improvements following the use of the registration manual.

ADVANCES IN KNOWLEDGE

This work helps evaluate training methods for novel roles that are developing in MRgRT.

To see or not to see: Evaluation of magnetic resonance imaging sequences for use in MR Linac-based radiotherapy treatment

BACKGROUND/PURPOSE

This work evaluated the suitability of MR derived sequences for use in online adaptive RT workflows on a 1.5 Tesla (T) MR-Linear Accelerator (MR Linac).

MATERIALS/METHODS

Non-patient volunteers were recruited to an ethics approved MR Linac imaging study. Participants attended 1-3 imaging sessions in which a combination of DIXON, 2D and 3D volumetric T1 and T2 weighted images were acquired axially, with volunteers positioned using immobilisation devices typical for radiotherapy to the anatomical region being scanned. Images from each session were appraised by three independent reviewers to determine optimal sequences over six anatomical regions: head and neck, female and male pelvis, thorax (lung), thorax (breast/chest wall) and abdomen. Site specific anatomical structures were graded by the perceived ability to accurately contour a typical organ at risk. Each structure was independently graded on a 4-point Likert scale as 'Very Clear', 'Clear', 'Unclear' or 'Not visible' by observers, consisting of radiographers (therapeutic and diagnostic) and clinicians.

RESULTS

From July 2019 to September 2019, 18 non-patient volunteers underwent 24 imaging sessions in the following anatomical regions: head and neck (n=3), male pelvis (n=4), female pelvis (n=5), lung/oesophagus (n=5) abdomen (n=4) and chest wall/breast (n=3). T2 sequences were the most preferred for perceived ability to contour anatomy in both male and female pelvis. For all other sites T1 weighted DIXON sequences were most favourable.

CONCLUSION

This study has determined the preferential sequence selection for organ visualisation, as a pre-requisite to our institution adopting MR-guided radiotherapy for a more diverse range of disease sites.

Impact of Cold Weather on Setup Errors in Radiotherapy

Objective

To investigate the influence of cold weather on setup errors of patients with chest and pelvic disease in radiotherapy.

Methods

The image-guided data of the patients were collected from the Radiotherapy Center of Cancer Hospital Affiliated to Guangxi Medical University from October 2020 to February 2021. During this period, the cold weather days were December 15, 16, and 17, 2020, and January 7 and 8, 2021. For body fixation in radiotherapy, an integrated plate and a thermoplastic mold were employed in 18 patients with chest disease, while an integrated plate and a vacuum pad were applied in 19 patients with pelvic disease. All patients underwent cone beam computed tomography (CBCT) scans in the first five treatments and once a week thereafter. The obtained data were registered to the planning CT image to get the setup errors of the patient in the translational direction including X, Y, and Z axes and rotational direction including R _X , R _Y , and R _Z . Then, the Mann-Whitney U test was performed. The expansion boundary values of the chest and pelvis were calculated according to the formula M _PTV=2.5∑+0.7δ.

Results

A total of 286 eligible results of CBCT scans were collected. There were 138 chest CBCT scans, including 26 taken in cold weather and 112 in usual weather, and 148 pelvic CBCT scans, including 33 taken in cold weather and 115 in usual weather. The X-, Y-, and Z-axis translational setup errors of patients with chest disease in the cold weather group were 0.16 (0.06, 0.32) cm, 0.25 (0.17, 0.52) cm, and 0.35 (0.21, 0.47) cm, respectively, and those in the usual weather group were 0.14 (0.08, 0.29) cm, 0.23 (0.13, 0.37) cm, and 0.18 (0.1, 0.35) cm, respectively. The results indicated that there was a statistical difference in the Z-axis translational error between the cold weather group and the usual weather group (U = 935.5; p=0.005 < 0.05), while there was no statistical difference in the rotational error between the two groups. The external boundary values of X, Y, and Z axes in the cold weather group were 0.57 cm, 0.92 cm, and 0.99 cm, respectively, and those in the usual weather group were 0.57 cm, 0.78 cm, and 0.68 cm, respectively. There was no significant difference in the translational and rotational errors of patients with pelvic disease between the cold weather group and the usual weather group (p < 0.05). The external boundary values of X, Y, and Z axes were 0.63 cm, 0.79 cm, and 0.68 cm in the cold weather group and 0.61 cm, 0.79 cm, and 0.61 cm in the usual weather group, respectively.

Conclusion

The setup error of patients undergoing radiotherapy with their bodies fixed by an integrated plate and a thermoplastic mold was greater in cold weather than in usual weather, especially in the ventrodorsal direction.