Evaluation of deep learning based implanted fiducial markers tracking in pancreatic cancer patients

Abstract

Real-time target position verification during pancreas stereotactic body radiation therapy (SBRT) is important for the detection of unplanned tumour motions. Fast and accurate fiducial marker segmentation is a requirement of real-time marker-based verification. Deep learning (DL) segmentation techniques are ideal because they don’t require additional learning imaging or prior marker information (e.g., shape, orientation). In this study, we evaluated three DL frameworks for marker tracking applied to pancreatic cancer patient data. The DL frameworks evaluated were 1) a convolutional neural network (CNN) classifier with sliding window, 2) a pretrained you-only-look-once (YOLO) version-4 architecture, and 3) a hybrid CNN-YOLO. Intrafraction kV images collected during pancreas SBRT treatments were used as training data (44 fractions, 2017 frames). All patients had 1-4 implanted fiducial markers. Each model was evaluated on unseen kV images (42 fractions, 2517 frames). The ground truth was calculated from manual segmentation and triangulation of markers in orthogonal paired kV/MV images. The sensitivity, specificity, and area under the precision-recall curve (AUC) were calculated. In addition, the mean-absolute-error (MAE), root-mean-square-error (RMSE) and standard-error-of-mean (SEM) were calculated for the centroid of the markers predicted by the models, relative to the ground truth. The sensitivity and specificity of the CNN model were 99.41% and 99.69%, respectively. The AUC was 0.9998. The average precision of the YOLO model for different values of recall was 96.49%. The MAE of the three models in the left-right, superior-inferior, and anterior-posterior directions were under 0.88±0.11 mm, and the RMSE were under 1.09±0.12 mm. The detection times per frame on a GPU were 48.3, 22.9, and 17.1 milliseconds for the CNN, YOLO, and CNN-YOLO, respectively. The results demonstrate the accuracy and precision of marker position prediction by the DL models from the ground truth was submillimeter, and detection time was fast enough to meet the requirements for real-time application.

Deep learning-based target tracking with X-ray images for radiotherapy: a narrative review

“…Some studies ( 52 , 54 , 57 , 63 ) have used real 2D kV X-ray images to train their models, thus mitigating the potential effects resulting from discrepancies between the simulated images and real images; however, some limitations persisted. It should be noted that the tracking target of reference ( 52 ) is the fiducial marker, and the question of whether it is efficient for marker-less target tracking requires further investigation.…”

“…And Kim et al ( 54 ) localized the lumbar vertebrae only and reported a mean center position error of 5.07±2.17 mm; however, this slightly large error could limit the application of the model in the clinic. Hirai et al ( 57 ) and Ahmed et al ( 63 ) also used real 2D kV images, but the input of their models was the sub-images cropped from the real images. The use of sub-images as input may limit the ability of the deep-learning model to extract global features and possibly decrease the tracking accuracy, especially when tumor and organ positions inter-fractionally change.…”

“…Ahmed et al ( 63 ) used a CNN for automatic fiducial marker tracking. In their study, a sliding window technique was employed to determine the search area from the input kV images, and the sub-images cropped from the search area were then input into the CNN to classify them as either fiducial markers or background.…”