Browsing by Author "Yang, Deshan"

Purpose: This project aimed to establish a method for computing 3D cardiac motion given continuous 2D-Cine MRI frames as the inputs. This approach would be useful for continuously monitoring cardiac and respiratory motion during MR-guided cardiac radiation therapy, and thus supporting radiation delivery guidance and gating.Methods: 5D-MRI datasets of seven patients, with each consisting of 3D spatial volumes of the cardiac cycle and respiratory cycle, were used for quantitative evaluation of the heart motion due to respiratory and cardiac movements. This was achieved through deformable image registration (DIR). Subsequently, principal component analysis (PCA) was performed on the computed deformation vector fields (DVF) to extract scores that effectively represent the characteristics of the DVFs. A deep learning model was then trained to predict the cardiac motion PCA scores given the inputs of 2D-Cine MRI. The predicted PCA scores were then transformed into 3D DVFs, which were then used to track 3D target motion. Results: The model’s performance was quantitatively evaluated on ground truth data that were withheld from model training. Across all 7 subjects, the average 3D DVF prediction errors for the heart region consistently remained around 0.3 ± 0.1mm. The predicted target motion, computed from the predicted DVFs, was visually evaluated, and found to be satisfactory. Conclusion: The developed method demonstrated promising potential in accurately computing and tracking real-time 3D cardiac motion given 2D-Cine MRI inputs. This approach presents a viable solution for continuously monitoring the 3D cardiac and respiratory motion of the heart during MR-guided cardiac radiation therapy.

Purpose

Acquisition and validation methods

Data format and usage notes

Potential applications

Purpose: To develop a generic procedure to make 5DCT from ECG 4DCTs and respiratory 4DCTs of cardiac RT patients. The 5DCT, whose dimension consists of 3D volume, cardiac cycle and respiratory cycle, will be used for quantitatively evaluating respiratory and cardiac motion of the heart, and supporting cardiac RT motion management, 5D dose calculation and dosimetry motion assessment. Methods: Images of ECG 4DCTs and respiratory 4DCTs for cardiac RT patients were obtained from the clinical system with IRB approval. For each patient, ten ECG 4DCT phases were registered using the groupwise deformable image registration algorithm GroupRegNet. The results were the template ECG CT representing the accurate average heart anatomy rather than an intensity-averaged CT, and the cardiac 4D DVF (deformation vector field). The ECG CT template and ten respiratory 4DCT phases were registered together using the 2nd groupwise registration to compute the 2nd respiratory 4D DVF. The computed DVFs from two groupwise registrations connected ECG 4DCTs to respiratory 4DCTs. A 10x10 cardiorespiratory 5DCT volume was generated by warping the ECG phases using composed DVFs. The final 5DCT phases were manually evaluated by visually the checking the respiratory and cardiac motion of the heart chambers. Results: The 5DCT generation procedure was implemented using Python and MATLAB, and was successfully applied to 4DCT images from five cardiac RT patients. The registration results were satisfactory based on visual evaluation. The quantitative evaluation and 5D dose calculation are planned for future work. Conclusion: A practical and effective procedure was developed to assess 5D motion of the heart and generate 5DCT phases from the clinical ECG 4DCTs and respiratory 4DCTs. The generated 5DCT could be used in dose calculation to assess the effect of 5D motion of the heart chambers on dosimetry for cardiac RT treatments.

Purpose

Approach

Results

Conclusions

Background

Purpose

Methods

Results

Conclusions

Managing respiratory motion in radiotherapy for abdominal and lung stereotactic body radiation therapy (SBRT) patients is crucial to achieving dose conformity and sparing healthy tissue. While previous studies have utilized principal component analysis (PCA) combined with deep learning to localize lung tumors in real-time, these models lack testing or validation with patient data from later treatment fractions. This study aims to achieve highly accurate 3D tumor localization for abdominal and lung cancer patients using a PCA-based convolutional neural network (CNN) motion model. Another goal is to enhance this prediction model by addressing interfractional motion in lung patients through deep learning and multiple treatment-acquired CT datasets.

The patient's 4D computed tomography (4DCT) image was registered, and resulting deformation vector fields (DVFs) were approximated using PCA. PCA coefficients, linked to diaphragm displacement, controlled breath variability in synthetic CT images. Digitally reconstructed radiographs (DRRs) from synthetic CTs served as network input. The networks were evaluated using a CT image designated for abdominal and lung cancer patient testing. A DRR of the testing CT was input into the model, and the predicted CT image was validated against the test CT to locate the tumor.

This study validated a PCA-based CNN motion model for abdominal and lung cancer patients. Intrafractional motion modeling accurately predicted abdominal tumors with a maximum error of 1.41 mm, whereas lung tumor localization resulted in a maximum error of 2.83 mm. Interfractional motion significantly influenced the accuracy of lung tumor prediction.

Background: Target Registration Error (TRE) calculated based on selected anatomical landmarks is commonly known as the only trustable way to evaluate DIR accuracy. However, manual landmark pair selection is labor intensive and subjected to observer variability. Currently, no DIR benchmark datasets are available for Liver CTs. Manually selected landmark pairs have limited DIR evaluation power due to inadequate landmarks quantity (i.e., ~5 landmark pairs per dataset1 for liver CTs) and positional accuracy. For the purpose of liver DIR verification, there is a great need to establish a large quantity of landmark pairs with good positional accuracy.Purpose: An image processing procedure was developed in this study to automatically and precisely detect landmark pairs on corresponding vessel bifurcations between pairs of intra-patient CT images. With high positional accuracy, the generated landmark pairs can be used to evaluate deformable image registration (DIR) methods quantitatively for liver CTs. Methods: Landmark pairs were detected within the liver between pairs of contrast-enhanced CT scans for 32 patients. For each case, the liver vessel tree was automatically segmented in one image. Landmarks were automatically detected on vessel bifurcations. The corresponding landmarks in the second image were placed using a parametric DIR method (pTVreg). Manual validation was applied to reject outliers and adjust the landmarks’ positions to account for vessel segmentation uncertainty caused by the inconsistent image quality. Landmark pairs' positional accuracy of the procedure was evaluated using digital phantoms on target registration errors (TREs). Results: On average, ~71 landmark pairs per case were detected after manual outlier rejection. The proposed procedure increased the quantity of liver landmark pairs by ~10 times compared to the reported in the literature. A fully manual spot check showed that the reported procedure performed better than or as good as human at landmark pairs positional accuracy. Measured in the digital phantoms, the mean and standard deviation of TREs were 0.67 ± 0.48 mm with 99% of landmark pairs having TREs smaller than 2mm. Conclusion: A large number of liver landmark pairs with high positional accuracy were detected in contrasted enhanced CT image pairs using the reported method. The detected landmark pairs can be used for the quantitative evaluation of DIR methods.

Background

Purpose

Methods

Results

Conclusion