Fluoroscopic Image Sequences Research Articles

Fluoroscopic image-based behavior analysis can objectively explain subjective expert assessment of wire navigation skill.

Psychomotor skill and decision-making efficiency in surgical wire navigation can be objectively evaluated by analysis of intraoperative fluoroscopic image sequences. Prior work suggests that such image-based behavior analysis of operating room (OR) performance can predict performer experience level (R2 = 0.62) and agree with expert opinion (the current standard) on the quality of a final implant construct (R2 = 0.59). However, it is unclear how objective image-based evaluation compares with expert assessments for entire technical OR performances. This study examines the relationships between three key variables: (1) objective image-based criteria, (2) expert opinions, and (3) performing surgeon experience level. A paired-comparison survey of seven experts, based upon eight OR fluoroscopic wire navigation image sequences, shows that the experts' preferences are best explained by objective metrics that reflect psychomotor and decision-making behaviors which are counter-productive to successful implant placement, like image count (R2 = 0.83) and behavior tally (R2 = 0.74). One such behavior, adjustments away from goal, uniquely correlated well with all three key variables: a fluoroscopic image-based analysis composite score (R2 = 0.40), expert consensus (R2 = 0.76), and performer experience (R2 = 0.41). These results confirm that experts view less efficient technical behavior as indicative of lesser technical proficiency. While expert assessments of technical skill were reliable and consistent, neither individual nor consensus expert opinion appears to correlate with performer experience (R2 = 0.11).

A Dedicated Simulator Training Curriculum Improves Resident Performance in Surgical Management of Pediatric Supracondylar Humerus Fractures.

The primary goal of including simulation in residency training is to improve technical skills while working outside of the operating room. Such simulation-related skill improvements have seldom been measured in the operating room. This is largely because uncontrolled variables, such as injury severity, patient comorbidity, and anatomical variation, can bias evaluation of an operating surgeon's skill. In this study, performance during the wire navigation phase of pediatric supracondylar humerus fracture fixation was quantitatively compared between 2 groups of orthopaedic residents: a standard training group consisting of residents who participated in a single simulator session of wire navigation training and an expanded training group consisting of residents who participated in a dedicated multifaceted wire navigation simulation training curriculum. To evaluate performance in the operating room, the full sequence of fluoroscopic images collected during wire navigation was quantitatively analyzed. Objective performance metrics included number of fluoroscopic images acquired, duration from placement of the first wire to that of the final wire, and wire spread at the level of the fracture. These metrics were measured from 97 pediatric supracondylar humerus fracture pinning surgeries performed by 28 different orthopaedic residents. No differences were observed between the groups for wire spread in the final fluoroscopic images (t(94) = 0.75, p = 0.45), an important clinical objective of the surgery. Residents who received the expanded simulator training used significantly fewer fluoroscopic images (mean of 46 vs. 61 images, t(85) = 2.25, p < 0.03) and required less time from first to final wire placement (mean of 11.2 vs. 14.9 minutes, t(83) = 2.53, p = 0.013) than the standard training group. A post hoc review of Accreditation Council for Graduate Medical Education case logs for 24 cases from the standard training group and for 21 cases from the expanded training group indicated that, at the time of surgeries, residents who received expanded training had completed fewer comparable cases than residents in the standard training group (mean of 13 vs. 21, t(42) = 2.40 p = 0.02). Further regression analysis indicated that the expanded simulator training produced an effect comparable with that associated with completing 10.5 similar surgical case experiences. This study demonstrates that training on a wire navigation simulator can lead to improved performance in the operating room on a critical skill for all orthopaedic residents. By taking fewer images and less time while maintaining sufficient pin spread, simulator-trained residents were objectively measured to have improved performance in comparison with residents who had not participated in the pediatric elbow simulator curriculum. As programs aim to provide safe and effective training for critical orthopaedic skills such as pinning a pediatric elbow, this study demonstrates a simulator curriculum that has demonstrated the transfer of skill from a learning environment to the operating room.

Fluoroscopic 3D Image Generation from Patient-Specific PCA Motion Models Derived from 4D-CBCT Patient Datasets: A Feasibility Study.

A method for generating fluoroscopic (time-varying) volumetric images using patient-specific motion models derived from four-dimensional cone-beam CT (4D-CBCT) images was developed. 4D-CBCT images acquired immediately prior to treatment have the potential to accurately represent patient anatomy and respiration during treatment. Fluoroscopic 3D image estimation is performed in two steps: (1) deriving motion models and (2) optimization. To derive motion models, every phase in a 4D-CBCT set is registered to a reference phase chosen from the same set using deformable image registration (DIR). Principal components analysis (PCA) is used to reduce the dimensionality of the displacement vector fields (DVFs) resulting from DIR into a few vectors representing organ motion found in the DVFs. The PCA motion models are optimized iteratively by comparing a cone-beam CT (CBCT) projection to a simulated projection computed from both the motion model and a reference 4D-CBCT phase, resulting in a sequence of fluoroscopic 3D images. Patient datasets were used to evaluate the method by estimating the tumor location in the generated images compared to manually defined ground truth positions. Experimental results showed that the average tumor mean absolute error (MAE) along the superior–inferior (SI) direction and the 95th percentile in two patient datasets were 2.29 and 5.79 mm for patient 1, and 1.89 and 4.82 mm for patient 2. This study demonstrated the feasibility of deriving 4D-CBCT-based PCA motion models that have the potential to account for the 3D non-rigid patient motion and localize tumors and other patient anatomical structures on the day of treatment.

Mathematical Models for Blood Flow Quantification in Dialysis Access Using Angiography: A Comparative Study.

Blood flow rate in dialysis (vascular) access is the key parameter to examine patency and to evaluate the outcomes of various endovascular interve7ntions. While angiography is extensively used for dialysis access–salvage procedures, to date, there is no image-based blood flow measurement application commercially available in the angiography suite. We aim to calculate the blood flow rate in the dialysis access based on cine-angiographic and fluoroscopic image sequences. In this study, we discuss image-based methods to quantify access blood flow in a flow phantom model. Digital subtraction angiography (DSA) and fluoroscopy were used to acquire images at various sampling rates (DSA—3 and 6 frames/s, fluoroscopy—4 and 10 pulses/s). Flow rates were computed based on two bolus tracking algorithms, peak-to-peak and cross-correlation, and modeled with three curve-fitting functions, gamma variate, lagged normal, and polynomial, to correct errors with transit time measurement. Dye propagation distance and the cross-sectional area were calculated by analyzing the contrast enhancement in the vessel. The calculated flow rates were correlated versus an in-line flow sensor measurement. The cross-correlation algorithm with gamma-variate curve fitting had the best accuracy and least variability in both imaging modes. The absolute percent error (mean ± SEM) of flow quantification in the DSA mode at 6 frames/s was 21.4 ± 1.9%, and in the fluoroscopic mode at 10 pulses/s was 37.4 ± 3.6%. The radiation dose varied linearly with the sampling rate in both imaging modes and was substantially low to invoke any tissue reactions or stochastic effects. The cross-correlation algorithm and gamma-variate curve fitting for DSA acquisition at 6 frames/s had the best correlation with the flow sensor measurements. These findings will be helpful to develop a software-based vascular access flow measurement tool for the angiography suite and to optimize the imaging protocol amenable for computational flow applications.

Deep learning-based recognition of cell structures in fluorescence microscopy sequences with respect to their morphology on cells infected with Marburg virus

Abstract The Institute of Virology at the Philipps-Universität Marburg is currently researching possible drugs to combat the Marburg virus. This involves classifying cell structures based on fluoroscopic microscopic image sequences. Conventionally, membranes of cells must be marked for better analysis, which is time consuming. In this work, an approach is presented to identify cell structures in images that are marked for subviral particles. It could be shown that there is a correlation between the distribution of subviral particles in an infected cell and the position of the cell’s structures. The segmentation is performed with a "Mask-R-CNN" algorithm, presented in this work. The model (a region-based convolutional neural network) is applied to enable a robust and fast recognition of cell structures. Furthermore, the network architecture is described. The proposed method is tested on data evaluated by experts. The results show a high potential and demonstrate that the method is suitable.

OA101] Assessment of dose efficiency of a modern X-ray fluoroscopy system using a model observer

Purpose The aim of this work was to measure image quality in terms of the square of the accumulation rate of the signal to noise ratio of low-contrast details, SNR rate 2 , kerma area product rate, P KA , rate and to quantify dose efficiency, SNR rate 2 / P KA , rate on a modern fluoroscopy system while varying the imaging conditions such as contrast detail size, X-ray beam size, dose rate, and object thickness. Methods The methodology derived by Tapiovaara et al. (Phys. Med. Biol., 38, 1761–1788, 1993, and STUK-A196, 2003) was used and implemented in MatlabTM. Measurement of model observer SNR rate 2 and P KA , rate were derived from stored fluoroscopy image sequences and the built-in kerma-area product reading on a Siemens Axiom Artis Zee MP system, respectively. 1024 image frames with and without the contrast detail were stored and subsequently analyzed by a model observer. SNR was calculated from the mean and standard deviation of the distributions of the decision variable. Finally, the long-term stability of the imaging system in terms of dose efficiency and the usefulness of the method as a quality control measure was evaluated. Results The precision in SNR rate 2 estimate was proportional to the inverse square root of the number of frames and SNR rate 2 was proportional to the area and squared thickness of the low-contrast detail. The dose efficiency increased with a factor of three with decreasing X-ray beam size from 242 cm2 to 182 cm2 and object thickness from 25 cm to 20 cm, hence quantifying the large dose efficiency gains by collimation and compression. The dose efficiency increased by 50% using the reduced dose rate compared to high dose rate. While the long-term stability of dose efficiency was measured with an uncertainty of ± 8%, the method is presently too time-consuming for routine quality control, mainly due to the slow data transfer of the large data files to the analysis server. Conclusions The results of the dose efficiency measurements under varying imaging conditions, agree with common radiological protection knowledge and allows changes in dose efficiency to be quantified which facilitate dose optimization in interventional procedures.

Improving subviral particle tracks in fluoroscopic image sequences by global motion compensation

Abstract Automatic detection and tracking of subviral particles in image sequences is an indispensable supportive method for modern medicine research programs. This paper describes the development of a highly adaptable camera-to-world system motion invariant tracking algorithm. A translation compensation is obtained by cross correlations. Particles are detected by an implemented existing algorithm. The detected particles are linked by solving a Linear Assignment Problem. For highly stable results the tracks are improved by Kalman filtering. The algorithm is tested on simulated sequences. The results show a great ability for stable tracking.

Extract-and-match geometric corner and step pattern approach for registration of fluoroscopic X-ray sequences.

This paper presents a method to extract-and-match robust corner features based on connecting edges from the edge maps, mainly formed by coronary vascular junctions in fluoroscopic x-ray sequence images. Such images are challenging due to the aperture problem. To overcome this, existing approaches attempt to extract vessels for registration. However, they are ineffective in poor quality images. Our approach describes the extracted robust corner features in a rotation invariant manner using step patterns, followed by matching them effectively. Experimental results show that our approach performs very well (above 80%) in a dataset of poor quality fluoroscopic x-ray image sequences without extensive processing such as segmentation or learning.

Motion-adaptive 3D nonlocal means filter based on stochastic distance for low-dose X-ray fluoroscopy

Abstract Low-dose X-ray fluoroscopy avoids radiation risks, but X-ray fluoroscopic image sequences are contaminated by quantum noise. In this paper, a 3D nonlocal means (NLM) filter based on stochastic distance that incorporates motion information is proposed, which can be applied to the denoising of X-ray fluoroscopic image sequences. First, the stochastic distance is obtained for use as the NLM filter similarity measure. This facilitates the removal of Poisson noise between the patches to be denoised within motion-compensated 3D volumes for spatio-temporal filtering. Second, motion state detection and motion-adaptive weights are proposed, which preserve the details of the motion that can occur during medical procedures. Experimental results obtained by applying the proposed method to real X-ray fluoroscopic image sequence images are shown in visual and numerical comparison with state-of-the-art denoising methods for spatio-temporal filtering techniques. The quantitative and qualitative results confirm that the proposed novel frameworks outperform other techniques in terms of objective criteria, as well as the subjective visual perception of the real X-ray fluoroscopic image sequences.

Gaussian mixture models based 2D-3D registration of bone shapes for orthopedic surgery planning.

In orthopedic surgery, precise kinematics assessment helps the diagnosis and the planning of the intervention. The correct placement of the prosthetic component in the case of knee replacement is necessary to ensure a correct load distribution and to avoid revision of the implant. 3D reconstruction of the knee kinematics under weight-bearing conditions becomes fundamental to understand existing in vivo loads and improve the joint motion tracking. Existing methods rely on the semiautomatic positioning of a shape previously segmented from a CT or MRI on a sequence of fluoroscopic images acquired during knee flexion. We propose a method based on statistical shape models (SSM) automatically superimposed on a sequence of fluoroscopic datasets. Our method is based on Gaussian mixture models, and the core of the algorithm is the maximization of the likelihood of the association between the projected silhouette and the extracted contour from the fluoroscopy image. We evaluated the algorithm using digitally reconstructed radiographies of both healthy and diseased subjects, with a CT-extracted shape and a SSM as the 3D model. In vivo tests were done with fluoroscopically acquired images and subject-specific CT shapes. The results obtained are in line with the literature, but the computational time is substantially reduced.

3D localization of electrophysiology catheters from a single x-ray cone-beam projection.

X-ray images allow the visualization of percutaneous devices such as catheters in real time but inherently lack depth information. The provision of 3D localization of these devices from cone beam x-ray projections would be advantageous for interventions such as electrophysiology (EP), whereby the operator needs to return a device to the same anatomical locations during the procedure. A method to achieve real-time 3D single view localization (SVL) of an object of known geometry from a single x-ray image is presented. svl exploits the change in the magnification of an object as its distance from the x-ray source is varied. The x-ray projection of an object of interest is compared to a synthetic x-ray projection of a model of said object as its pose is varied. svl was tested with a 3 mm spherical marker and an electrophysiology catheter. The effect of x-ray acquisition parameters on svl was investigated. An independent reference localization method was developed to compare results when imaging a catheter translated via a computer controlled three-axes stage. svl was also performed on clinical fluoroscopy image sequences. A commercial navigation system was used in some clinical image sequences for comparison. svl estimates exhibited little change as x-ray acquisition parameters were varied. The reproducibility of catheter position estimates in phantoms denoted by the standard deviations, (σ(x), σ(y), σ(z)) = (0.099 mm, 0.093 mm, 2.2 mm), where x and y are parallel to the detector plane and z is the distance from the x-ray source. Position estimates (x, y, z) exhibited a 4% systematic error (underestimation) when compared to the reference method. The authors demonstrated that EP catheters can be tracked in clinical fluoroscopic images. It has been shown that EP catheters can be localized in real time in phantoms and clinical images at fluoroscopic exposure rates. Further work is required to characterize performance in clinical images as well as the sensitivity to clinical image quality.

Guidewire contrast enhancement by a two-stage order statistic filter in X-ray fluoroscopy

As a result of an increased number of applications, X-ray fluoroscopy has become one of major sources of diagnostic X-ray dose to patients and staff. It is important to lower X-ray dose while maintaining or improving image quality. A two-stage order statistic filter has been proposed. The first part of the filter is designed to detect the orientation of the guidewire and to reduce the image noise. The second part of the filter is to implement a non-linear discriminator to obtain an optimal output from the outputs generated by the oriented filters used in the first stage. We apply the filter to the X-ray synthetic fluoroscopy image sequences with and without clinical backgrounds. A human observer model is applied to quantitatively assess detection performance of guidewire in unfiltered and filtered image sequences. Our experiment has shown that the filter has effectively improved guidewire visibility in the image sequences.

A kernel-based method for markerless tumor tracking in kV fluoroscopic images

Markerless tracking of respiration-induced tumor motion in kilo-voltage (kV) fluoroscopic image sequence is still a challenging task in real time image-guided radiation therapy (IGRT). Most of existing markerless tracking methods are based on a template matching technique or its extensions that are frequently sensitive to non-rigid tumor deformation and involve expensive computation. This paper presents a kernel-based method that is capable of tracking tumor motion in kV fluoroscopic image sequence with robust performance and low computational cost. The proposed tracking system consists of the following three steps. To enhance the contrast of kV fluoroscopic image, we firstly utilize a histogram equalization to transform the intensities of original images to a wider dynamical intensity range. A tumor target in the first frame is then represented by using a histogram-based feature vector. Subsequently, the target tracking is then formulated by maximizing a Bhattacharyya coefficient that measures the similarity between the tumor target and its candidates in the subsequent frames. The numerical solution for maximizing the Bhattacharyya coefficient is performed by a mean-shift algorithm. The proposed method was evaluated by using four clinical kV fluoroscopic image sequences. For comparison, we also implement four conventional template matching-based methods and compare their performance with our proposed method in terms of the tracking accuracy and computational cost. Experimental results demonstrated that the proposed method is superior to conventional template matching-based methods.

SU‐D‐18A‐03: Markerless Motion Tracking of Lung Tumors Using Dual‐Energy Fluoroscopy

Purpose:To evaluate the efficacy of dual energy (DE) vs. single energy (SE) fluoroscopic imaging in lung tumors using a markerless template‐based tracking algorithm.Methods:Ten representative patient breathing patterns were programmed into a Quasar™ motion phantom. The phantom was modified by affixing pork ribs to the surface, and a cedar insert with a small spherical volume was used to simulate lung and tumor, respectively. Gated 60kVp(6mA) and 120kVp(1.5mA) fluoroscopic sequences were acquired using the Varian TrueBeam Integrated Imaging Solution. These high and low energy frames were subsequently aligned by phase and amplitude. Frame‐by‐frame weighted logarithmic subtraction was performed resulting in a DE fluoroscopic image sequence that highlighted soft tissue structures. A template‐based algorithm was then used to track tumor motion throughout the sequence for DE and SE images (120kVp). Tracking coordinates were evaluated against ground truth tumor locations as identified by a physician. A displacement of greater than 3 mm was considered a non‐successful track. The same process was also repeated for images obtained from 3 SBRT lung patients, none of which had implanted fiducials.Results:For phantom imaging, a total of 1925 frames were analyzed. The algorithm successfully tracked the target on 91.6%(1763/1925) of DE frames vs. 74.9%(1441/1923) SE images (p&lt;0.001). The displacement between tracking coordinates and ground truth for the phantom was 1.1mm+/−0.7mm for DE and 1.5mm+/− 0.6mm for SE (p&lt;0.001). Images from 3 patients were analyzed with the tumor tracked on 84.9%(478/563) of frames for DE vs. 53.5%(301/563) for SE (p&lt;0.001). The average difference to ground truth for successfully tracked DE frames was 0.6mm+/−0.6mm vs. 1.7mm+/−0.7mm SE (p&lt;0.001).Conclusion:This work indicates the potential for markerless tumor tracking utilizing DE fluoroscopy. With DE imaging, the algorithm showed improved detectability vs. SE fluoroscopy and was able to accurately track the tumor in a majority of cases.This work was supported by Varian Medical Systems

Markerless Lung Tumor Motion Tracking by Dynamic Decomposition of X-Ray Image Intensity.

We propose a new markerless tracking technique of lung tumor motion by using an X-ray fluoroscopic image sequence for real-time image-guided radiation therapy (IGRT). A core innovation of the new technique is to extract a moving tumor intensity component from the fluoroscopic image intensity. The fluoroscopic intensity is the superimposition of intensity components of all the structures passed through by the X-ray. The tumor can then be extracted by decomposing the fluoroscopic intensity into the tumor intensity component and the others. The decomposition problem for more than two structures is ill posed, but it can be transformed into a well-posed one by temporally accumulating constraints that must be satisfied by the decomposed moving tumor component and the rest of the intensity components. The extracted tumor image can then be used to achieve accurate tumor motion tracking without implanted markers that are widely used in the current tracking techniques. The performance evaluation showed that the extraction error was sufficiently small and the extracted tumor tracking achieved a high and sufficient accuracy less than 1 mm for clinical datasets. These results clearly demonstrate the usefulness of the proposed method for markerless tumor motion tracking.

Esophagus Silhouette Extraction and Reconstruction From Fluoroscopic Views for Cardiac Ablation Procedure Guidance

Cardiac ablation involves the risk of serious complications when thermal injury to the esophagus occurs. This paper proposes to reduce the risk of such injuries by a proactive visualization technique, improving physician awareness of the esophagus location in the absence of or in addition to a reactive monitoring device such as a thermal probe. This is achieved by combining a graphical representation of the esophagus with live fluoroscopy. Toward this goal, we present an automated method to reconstruct and visualize a 3-D esophagus model from fluoroscopy image sequences acquired using different C-arm viewing directions. In order to visualize the esophagus under fluoroscopy, it is first biomarked by swallowing a contrast agent such as barium. Images obtained in this procedure are then used to automatically extract the 2-D esophagus silhouette and reconstruct a 3-D surface of the esophagus internal wall. Once the 3-D representation has been computed, it can be visualized using fluoroscopy overlay techniques. Compared to 3-D esophagus imaging using CT or C-arm CT, our proposed fluoroscopy method requires low radiation dose and enables a simpler workflow on geometry-calibrated standard C-arm systems.

Image-Guided Transapical Aortic Valve Implantation Sensorless Tracking of Stenotic Valve Landmarks in Live Fluoroscopic Images

Aortic valve stenosis is one of the most frequently acquired valvular heart diseases, accounting for almost 70% of valvular cardiac surgery. Transapical transcatheter aortic valve implantation has recently become a suitable minimally invasive technique for high-risk and elderly patients with severe aortic stenosis. In this article, we aim to automatically define a target area of valve implantation, namely, the area between the coronary ostia and the lowest points of two aortic valve cusps. Therefore, we present a new image-based tracking method of these aortic landmarks to assist in the placement of aortic valve prosthesis under live 2D fluoroscopy guidance. We propose a rigid intensity-based image registration technique for tracking valve landmarks in 2D fluoroscopic image sequences, based on a real-time alignment of a contrast image including the initialized manual valve landmarks to each image of sequence. The contrast image is automatically detected to visualize aortic valve features when the aortic root is filled with a contrast agent. Our registration-based tracking method has been retrospectively applied to 10 fluoroscopic image sequences from routine transapical aortic valve implantation procedures. Most of all tested fluoroscopic images showed a successful tracking of valve landmarks, especially for the images without contrast agent injections. A new intraoperative image-based method has been developed for tracking aortic valve landmarks in live 2D fluoroscopic images to assist transapical aortic valve implantations and to increase the overall safety of surgery as well.

Aortic valve prosthesis tracking for transapical aortic valve implantation

Transapical aortic valve implantation (TA-AVI) is a new minimally invasive surgical treatment of aortic stenosis for high-risk patients. The placement of aortic valve prosthesis (AVP) is performed under 2D X-ray fluoroscopic guidance. Difficult clinical complications can arise if the implanted valve is misplaced. Therefore, we present a method to track the AVP in 2D X-ray fluoroscopic images in order to improve the accuracy of the TA-AVI. The proposed tracking method includes the template matching approach to estimate the position of AVP and a shape model of the prosthesis to extract the corner points of the AVP in each image of sequence. To start the AVP tracking procedure, an initialization step is performed by manually defining the corner points of the prosthesis in the first image of sequence to provide the required algorithm parameters such as the AVP model parameters. We evaluated the AVP tracking method on six 2D intra-operative fluoroscopic image sequences. The results of automatic AVP localization agree well with manually defined AVP positions. The maximum localization errors of tracked prosthesis are less than 1mm and within the clinical accepted range. For assisting the TA-AVI, a method for tracking the AVP in 2D X-ray fluoroscopic image sequences has been developed. Our AVP tracking method is a first step toward automatic optimal placement of the AVP during the TA-AVI.

SU‐GG‐J‐55: Simultaneous Estimation of Lung Tumor Image and Intrafractional Motion without Implanted Markers Using KV‐X‐Ray Fluoroscopy for Image‐Guided Radiotherapy

Purpose: The aim of this research is to develop a novel image processing method for accurate estimation of intrafractional lung tumor respiratory motion using kV‐X‐ray fluoroscopy without implanted fiducial makers that have been used and needed for image guided radiotherapy, but their implantation run the risk of pulmonary emphysema and pneumothorax. Method and Materials: We propose a technique for extracting the lung tumor image from the kV‐X‐ray fluoroscopic observation. An essential core of the method is in a simultaneous estimation of tumor image and motion by using sequences of fluoroscopic images that can provide pieces of information needed for the estimation. The tumor images extracted can then easily be used for motion estimation of the tumor by using conventional motion estimation techniques such as block matching (BM) and template matching (TM). Results: We have evaluated the new extraction method with TM using three patient kV‐X‐ray fluoroscopy data. An example of the tumor image extraction is shown in the supporting document. The average error and the standard deviation of the motion estimation were, respectively, 0.72 [mm] and 0.37 [mm] by the proposed method. This is clearly smaller than the average of 1.02 [mm] with the standard deviation 0.23[mm] by the conventional TM. Note that the comparison between the proposed method with TM and the conventional TM demonstrates the effectiveness of the new extraction technique. In addition, this result implies that the method can achieve within 1 [mm] of motion estimation accuracy without implanted markers and it's clinically acceptable. Conclusion: We have developed intrafractional motion estimation method of lung tumor without implanted markers. Any motion estimation technique can be incorporated into the proposed extraction method for accuracy improvement. The proposed method can achieve clinically sufficient accuracy estimation without implanted makers and thus can fundamentally reduce any patient's risk caused by the implantation.

The diaphragm as an anatomic surrogate for lung tumor motion

Lung tumor motion due to respiration poses a challenge in the application of modern three-dimensional conformal radiotherapy. Direct tracking of the lung tumor during radiation therapy is very difficult without implanted fiducial markers. Indirect tracking relies on the correlation of the tumor's motion and the surrogate's motion. The present paper presents an analysis of the correlation between tumor motion and diaphragm motion in order to evaluate the potential use of diaphragm as a surrogate for tumor motion. We have analyzed the correlation between diaphragm motion and superior–inferior lung tumor motion in 32 fluoroscopic image sequences from ten lung cancer patients. A simple linear model and a more complex linear model that accounts for phase delays between the two motions have been used. Results show that the diaphragm is a good surrogate for tumor motion prediction for most patients, resulting in an average correlation factor of 0.94 and 0.98 with each model respectively. The model that accounts for delays leads to an average localization prediction error of 0.8 mm and an error at the 95% confidence level of 2.1 mm. However, for one patient studied, the correlation is much weaker compared to other patients. This indicates that, before using diaphragm for lung tumor prediction, the correlation should be examined on a patient-by-patient basis.