Evaluating performance of a user-trained MR lung tumor autocontouring algorithm in the context of intra- and interobserver variations
|
|
- Gillian Thomas
- 5 years ago
- Views:
Transcription
1 Evaluating performance of a user-trained MR lung tumor autocontouring algorithm in the context of intra- and interobserver variations Eugene Yip a) and Jihyun Yun Department of Medical Physics, Cross Cancer Institute, University Avenue, Edmonton, AB T6G 1Z2, Canada Medical Physics Division, Department of Oncology, University of Alberta, University Avenue, Edmonton, AB T6G 1Z2, Canada Zsolt Gabos, Sarah Baker, and Don Yee Department of Radiation Oncology, Cross Cancer Institute University Avenue, Edmonton, AB T6G 1Z2, Canada Radiation Oncology Division, Department of Oncology, University of Alberta, University Avenue, Edmonton, AB T6G 1Z2, Canada Keith Wachowicz, Satyapal Rathee, and B. Gino Fallone Department of Medical Physics, Cross Cancer Institute, University Avenue, Edmonton, AB T6G 1Z2, Canada Medical Physics Division, Department of Oncology, University of Alberta, University Avenue, Edmonton, AB T6G 1Z2, Canada (Received 17 August 2017; revised 17 October 2017; accepted for publication 10 November 2017; published 15 December 2017) Purpose: Real-time tracking of lung tumors using magnetic resonance imaging (MRI) has been proposed as a potential strategy to mitigate the ill-effects of breathing motion in radiation therapy. Several autocontouring methods have been evaluated against a gold standard of a single human expert user. However, contours drawn by experts have inherent intra- and interobserver variations. In this study, we aim to evaluate our user-trained autocontouring algorithm with manually drawn contours from multiple expert users, and to contextualize the accuracy of these autocontours within intra- and interobserver variations. Methods: Six nonsmall cell lung cancer patients were recruited, with institutional ethics approval. Patients were imaged with a clinical 3 T Philips MR scanner using a dynamic 2D balanced SSFP sequence under free breathing. Three radiation oncology experts, each in two separate sessions, contoured 130 dynamic images for each patient. For autocontouring, the first 30 images were used for algorithm training, and the remaining 100 images were autocontoured and evaluated. Autocontours were compared against manual contours in terms of Dice s coefficient (DC) and Hausdorff distances (d H ). Intra- and interobserver variations of the manual contours were also evaluated. Results: When compared with the manual contours of the expert user who trained it, the algorithm generates autocontours whose evaluation metrics (same session: DC = 0.90(0.03), d H = 3.8(1.6) mm; different session DC = 0.88(0.04), d H = 4.3(1.5) mm) are similar to or better than intraobserver variations (DC = 0.88(0.04), and d H = 4.3(1.7) mm) between two sessions. The algorithm s autocontours are also compared to the manual contours from different expert users with evaluation metrics (DC = 0.87(0.04), d H = 4.8(1.7) mm) similar to interobserver variations (DC = 0.87(0.04), d H = 4.7(1.6) mm). Conclusions: Our autocontouring algorithm delineates tumor contours (<20 ms per contour), in dynamic MRI of lung, that are comparable to multiple human experts (several seconds per contour), but at a much faster speed. At the same time, the agreement between autocontours and manual contours is comparable to the intra- and interobserver variations. This algorithm may be a key component of the real time tumor tracking workflow for our hybrid Linac-MR device in the future American Association of Physicists in Medicine [ Key words: autocontouring, linac-mri hybrid, MRI guidance, observer variability, respiratory motion management, tumor tracking 1. INTRODUCTION Intrafractional tumor motion due to respiration poses a significant challenge to lung cancer radiotherapy. 1 Recently, realtime, magnetic resonance image (MRI) guided radiation therapy has been made possible by the development of novel hybrid devices 2 5 that combine the imaging capability of a MR imager and a radiation therapy device. These systems are capable of real-time tumor imaging for tracking, in which the radiation beam follows a moving tumor in real-time. Numerous groups 6 9 have proposed schemes to acquire dynamic 2D MR images to track tumor motion during radiation delivery. Ideally, the tumor (i.e., gross tumor volume [GTV]) would be contoured in these dynamic images by an expert radiation oncologist such that the radiation field is conformed using dynamic multi-leaf collimator (MLC) to the known shape 307 Med. Phys. 45 (1), January /2018/45(1)/307/ American Association of Physicists in Medicine 307
2 308 Yip et al.: Autocontouring and observer variability 308 and location of tumor. The AAPM Task Group 76 1 report recommends a maximum delay time of 0.5 s for real time tumor tracking. This delay includes times for image acquisition, reconstruction, contouring and MLC repositioning. It is unfeasible for a human expert to continually contour images at a rate of lower than 0.5 s per image for the duration of treatment. To address this, our group 6,10 and others 8 have previously proposed real time autocontouring algorithms. A key feature to imaging and contouring in real time is that it does not rely on regular, predictable motion of lung tumor. Some of these algorithms have been validated using in-vivo images 6,8 against the gold standard contours drawn by a single human expert. However, with a single expert, the impact of the expert s individual preferences and the inherent uncertainties (i.e., intraobserver and inter-observer variations) in the gold standard contour set were not accessed. In this work, 3 human experts in radiation oncology performed manual contouring of the tumor in dynamic lung MR images using identical viewing/display conditions, in two different sessions. A subset of these contours is used to train the algorithm. The two objectives of this study are (a) To evaluate the difference of automatically determined contours from the manual ones drawn by different human experts, and (b) To contextualize the autocontouring algorithm s agreement metrics with intra- and interobserver metrics, which represent inherent uncertainties in manual contours. We hypothesize that our algorithm can generate autocontours, with agreement metrics against standard manual contours that are comparable to intra- and interobserver variation metrics. 2. MATERIALS AND METHODS 2.A. Patient selection and MR image acquisition In this institutional ethics approved pilot study, patients with nonsmall cell lung cancer (NSCLC) undergoing stereotactic radiotherapy are recruited for this study based on selection by a radiation oncologist. Six patients with stage 1 lung cancer, tumor diameter <5 cm were selected. Early stage lung cancer patients are most likely to benefit from the highly conformal, curative radiation treatment provided by MR based tumor tracking. The patients are scanned using a 3 T MRI (Philips Achieva, The Netherlands) using a dynamic, 2D balanced steady state free precession (bssfp) sequence (Matrix size, , Voxel size = 3mm9 3mm9 20 mm, TE = 1.1 ms, TR = 2.2 ms, Imaging time per frame = 270 ms). Each imaging study was performed with the patients undergoing free breathing. 2.B. Manual contouring Three experts (2 radiation oncologists with 10+ years of experience and a radiation oncology resident with 5 years of post graduate experience) manually delineated the tumor contours in the image series of each patient, both to train the algorithm and to serve as a reference for comparison to automatically drawn contours. To minimize bias arising from contouring hardware (i.e., monitor contrast), a common computer terminal, with the freely available CERR 11 (Computation Environment for Radiation Therapy), a MATLAB based software, is used to perform all manual contours. As the CERR platform is not the standard contouring software at the clinic, the contouring experts are given training to familiarize with the software for the given tasks. For each of the 6 patients, each of the 3 experts is asked to manually contour the GTV in 130 dynamic images of the dataset (Session 1). Identical window and level settings (the default CERR setting for the MR images) are used in all contouring sessions to minimize variation; this process mimics the practice of applying standardized CT window/levels for the manual lung tumor contouring at our institute. To ensure independence, each of the contouring experts is blinded to the contours drawn by the other 2 experts. For intraobserver variation assessment, the experts, blinded to their previous contours, are asked to recontour the images at least 1 week after the first round of contouring (Session 2). In summary, a total of 6 patients93 oncologists 9 2 sessions images are manually contoured. 2.C. Autocontouring algorithm The autocontouring algorithm has been discussed in detail previously, 6,12 and it is briefly explained here. Eckhorn 13 introduced a type of neural networks to model the electrochemical mechanism of a cat s visual cortex. A modified model, known as the Pulse Coupled Neural Networks (PCNN), 14 has shown success in image processing applications such as image enhancement, segmentation and classification In this model, a network comprises of a 2D matrix of individual neurons. Each individual neuron receives continuous input signals, but its output is binary (firing/not firing). The neurons, in close proximity to each other, are interlinked such that the output of a neuron depends not only on its own input but also those of its interlinked neighbors, thus, the neurons tend to fire together in clusters. In our MR images, a tumor consists of heterogeneous cluster of hyperintense pixels in comparison to surrounding healthy tissues. Each neuron in the model can emulate an image pixel; its input being the pixel s value in the image. As the output of this neuron is influenced both by its pixel value and the pixel values of its neighbors, the neural network model serves to group heterogeneous tumor pixels together. Applying this algorithm iteratively enhances the contrast of the tumor compared to the surrounding healthy tissues. 5 The current clinical standard for segmenting tumor in an image is the manually drawn contour by the radiation oncologist. The algorithm design accounts for the individual preferences of clinicians, thus, it requires a small number of preparatory images for training. In a prospective implementation of this software, it is anticipated this training process will take place prior to treatment (i.e., during RT simulation), requiring an additional MRI scan. However, for this retrospective validation study, the first 30 images, out of the dataset of 130 images, are used for training (i.e., the training
3 309 Yip et al.: Autocontouring and observer variability 309 images). The remaining 100 images (i.e., the tracking images) are used for evaluating the algorithm. The manually drawn contours (ROI STD ) on the training images are used to optimize the parameters of the algorithm for an individual patient. An example training image with a contour for each of the 6 patients is shown in Fig C.1. Main contouring algorithm The main algorithm can be summarized in the following steps 6 : (a) Normalized cross correlation determines the approximate location of the tumor and yields [Fig. 2(a)], (b) Contrast enhancement, performed using the PCNN 6 algorithm, yields [Fig. 2(b)], (c) Thresholding of the image via the Otsu s Method 18 yields [Fig. 2(c)], (d) Removal of small islands yields [Fig. 2(d)], (e) Smoothing of the tumor shape yields [Fig. 2(e)], (f) Morphology/Shifts, which includes erosion, dilation and translation, yields the final autocontour (ROI AUTO ) shown in [Fig. 2(f)]. The entire process takes <20 ms (Windows 7, Intel i7-2600k, 4GB RAM) per contour. As the algorithm design adapts to the individual expert user`s contouring preferences, step 2 (contrast enhancement by neural network) and step 6 (morphological/linear shift) are governed by the parameters optimized using the individual user s own training contour set. Due to this feature, the algorithm generates autocontours that reflect the individual user s contouring preferences. 2.C.2. Parameter optimization via algorithm training The algorithm requires manually drawn ROI STD on the training images, which are used to find optimized parameters for each individual patient. As described by Yun et al., 12 the adaptive particle swarm optimization 19 determines the optimized parameters in 2 3 hr offline, prior to tracking. During the optimization process, test parameters are entered into the main algorithm which creates autocontours (ROI AUTO ). The FIG. 1. A sample dynamic lung sagittal MR image of each of the 6 patients, with ROI STD contoured on the image. FIG. 2. A summary of the autocontouring process, a normalized cross correlation gives the approximate location of the tumor (a), PCNN based contrast enhancement gives (b), thresholding leads to (c), removal of small islands leads to (d), smoothing leads to (e), and morphology/translation leads to final result in(f). [Color figure can be viewed at wileyonlinelibrary.com]
4 310 Yip et al.: Autocontouring and observer variability 310 optimal set of parameters yield the contours that are best matched to the training ROI STD in terms of the Dice Coefficient (DC), 20 define as follows. T AreaðROI STD ROIAUTO Þ DC ¼ 2 AreaðROI STD ÞþAreaðROI AUTO Þ 2.D. Evaluation of contours For evaluation of the contours in the tracking images, we performed Automatic vs. Manual, and Manual vs. Manual contour comparisons. For automatic vs. manual comparisons, we evaluate the contour agreement between (a) trained automatic contours with its trainer s contours in the same session, (b) trained automatic contours and its trainer s contours in a different session, and (c) trained automatic contours against the contours of a different user. To present the results, the following notation is used: A 12 represents autocontours trained by expert 1 in session 2; whereas M 31 denotes manual contours drawn by expert 3 in session 1. The automatic contouring algorithm, trained using the contours from 30 training images (images 1 30) from each of the 6 manual contouring sessions (3 contourer 9 2 sessions) is used to generate 6 sets of 100 automatic contours (ROI AUTO ) on the 100 tracking images (images ), labeled as A 11,A 12,A 21,A 22,A 31,A 32. These are compared against the 6 sets of manual contours (ROI STD ) drawn on the same tracking images, namely, M 11,M 12,M 21,M 22,M 31,M 32. We group the data into 3 distinct types of comparisons: same user same session (SUSS) match, same user different session (SUDS) match, and different user (DU) match. As an example, A 11,100 ROI AUTO generated with training contours from user 1, session 1, can be compared against M 11, which are 100 ROI STD generated by user 1 in session 1 (i.e., SUSS), or M 12, which are ROI STD generated by the same user at a different session, (i.e., SUDS), or M 21, M 22, M 31, M 32, which are ROI STD generated by different users (i.e., DU). For manual vs. manual comparisons M 11,M 12,M 21,M 22, M 31,M 32 are compared against M 11,M 12,M 21,M 22,M 31, M 32. The SUSS comparison (i.e., M 11 vs. M 11, etc.) is trivial, as they are identical contours. The SUDS manual vs. manual match (i.e., M 11 vs. M 12, etc.) represents the intraobserver variations, while the DU match (i.e., M 11 vs. M 31, etc.) represents interobserver variations. Two metrics for contour agreements are used: Firstly, DC, introduced in Section 2.C.2, evaluated the agreement between two sets of contours, (i.e., ROI AUTO vs. ROI STD for automatic vs. manual comparisons) in the 100 tracking images. It should be noted that our automatic algorithm has no prior knowledge of the 100 manually drawn ROI STD on the tracking images. These 100 ROI STD are compared against corresponding ROI AUTO to validate the algorithm performance only after the autocontouring session is completed. In addition to DC, the Hausdorff distance 21 (d H ) is also used as an alternative metric for comparing contours. To calculate d H between contours A and B, the following steps are taken. For every point on contour A, the shortest distance to any point in contour B is determined. The largest of these distances is denoted as da; ð BÞ. Conversely, from every point on contour B, the shortest distance to any point in contour A is determined. The largest of these distance is denoted as db; ð AÞ. Note that da; ð BÞ is not necessarily equal to db; ð AÞ. The Hausdorff distance d H, is defined as follows. d H ¼ maxðdða; BÞ; db; ð AÞÞ Unlike the area based DC, d H is quite sensitive to small discrepancies in the contours, even though those discrepancies have a small impact in the overall area of the contour and the DC. Thus, d H provides a useful alternative metric for our contour comparisons. 3. RESULTS 3.A. Automatic vs. manual comparisons The six autocontour datasets, A 11,A 12,A 21,A 22,A 31,A 32 are compared against the 6 sets of manual contours drawn on the same tracking images, namely, M 11,M 12,M 21,M 22,M 31, M 32. DC is shown in Table I, with the diagonal element representing the automatic vs. manual SUSS DC, bolded elements representing automatic vs. manual SUDS DC, and underlined elements representing automatic vs. manual DU DC. The overall mean and standard deviation (i.e., mean (SD)) values for automatic vs. manual SUSS DC is 0.90 (0.03), SUDS DC is 0.88(0.04), DU DC is 0.87(0.04). The equivalent analysis for automatic vs. manual Hausdorff distance (d H ) is shown in Table II. Overall, the mean (SD) values for SUSS d H is 3.8(1.6) mm, the SUDS d H is 4.3 (1.5) mm, and the DU match d H is 4.8(1.7) mm. TABLE I. Dice Coefficient (Mean(SD)) for comparing automatic vs. manual contours. Diagonal elements represent SUSS comparisons. Bolded elements represent SUDS comparisons. Underlined elements represent DU comparisons. A (0.03) 0.90(0.03) 0.86(0.04) 0.87(0.05) 0.87(0.04) 0.87(0.04) A (0.03) 0.91(0.03) 0.84(0.04) 0.83(0.05) 0.85(0.04) 0.85(0.04) A (0.04) 0.90(0.04) 0.89(0.04) 0.86(0.05) 0.87(0.04) 0.88(0.04) A (0.05) 0.87(0.04) 0.85(0.05) 0.89(0.04) 0.88(0.04) 0.86(0.05) A (0.04) 0.88(0.04) 0.87(0.05) 0.88(0.04) 0.90(0.03) 0.89(0.04) A (0.04) 0.88(0.03) 0.88(0.04) 0.86(0.05) 0.89(0.03) 0.90(0.03)
5 311 Yip et al.: Autocontouring and observer variability B. Manual vs. manual comparisons (Intra- and Interobserver variations) The manual contours for the 100 tracking images (i.e., images in each data set), labelled M 11,M 12,M 21,M 22, M 31,M 32, are compared against each other. DC is shown in Table III. The diagonal column represents the SUSS manual vs. manual comparisons, which always returns 1 as they are identical contours. The bolded elements represent manual vs. manual SUDS match (i.e., intraobserver variations). Underlined elements represent the manual vs. manual DU match (i.e., interobserver variations). The overall mean(sd) values for SUDS/intraobserver DC is 0.88(0.04), and the DU/interobserver DC is 0.87(0.04). The manual to manual Hausdorff distances (d H, in mm) are shown in Table IV. The diagonal elements represent comparisons of identical contours (SUSS) thus always yields 0. The bolded elements represent SUDS (i.e., intraobserver) d H, while underlined elements represent DU (i.e., interobserver) d H. The overall mean(sd) values for SUDS (i.e., intraobserver) d H is 4.3(1.7) mm and the DU/interobserver d H is 4.7 (1.6) mm. A summary of the SUSS, SUDS and DC metrics for manual vs. manual and manual vs. automatic comparisons is shown in Table V. 4. DISCUSSION Intra- and interobserver variations for manual contours of lung tumors are well documented in the literature, in CT, 22 in TABLE II. Hausdorff distances (Mean(SD), mm) for comparing automatic vs. manual contours. Diagonal elements represent SUSS comparisons. Bolded elements represent SUDS comparisons. Underlined elements represent DU comparisons. A (1.7) 3.4(1.4) 5.7(1.9) 4.6(1.9) 4.7(1.6) 5.2(1.9) A (1.4) 3.3(1.1) 6.8(1.7) 5.7(2.1) 5.7(1.8) 6.2(1.9) A (1.9) 3.9(1.6) 4.6(2.0) 4.3(1.6) 4.2(1.4) 4.6(1.9) A (1.6) 4.2(1.6) 4.9(1.8) 3.5(1.5) 4.0(1.4) 4.8(1.9) A (1.9) 4.4(1.7) 4.8(1.9) 3.9(1.4) 3.5(1.3) 4.3(1.8) A (1.7) 4.4(1.7) 4.3(1.8) 4.2(1.4) 3.6(1.2) 3.8(1.7) PET/CT fusion 23 and in 4DCT 24 images. In terms of MRI of lung cancer, intra/inter observer metrics have been evaluated for breath held lung volumes in MR/CT images, 25 as well diffusion metrics. 26,27 However, in terms of lung tumor target delineation, to the best knowledge of the authors (i.e., PubMed/Google Scholar), this is the first assessment of intraobserver and interobserver variations in the dynamic lung MR images for monitoring the changes in tumor shape and location in real-time. This is also the first study in which a user-trained autocontouring algorithm has been benchmarked against variation metrics of multiple expert users, as suggested by Valentini et al. 28 on dynamic lung MR images. The key objective of this study is to validate the autocontouring algorithm in the context of the variations in contours drawn by experts. If one simply assumed a single expert s manual contours to be a gold standard, an approach TABLE IV. Hausdorff Distance (mm, Mean(SD)) comparing manual vs. manual contours. Bolded elements represent intraobserver (SUDS) comparisons while underlined elements represent interobserver (DU) comparisons. TABLE V. Summary of SUSS, SUDS and DU metrics for manual vs. manual agreements and automatic vs. manual agreements. In manual vs. manual metrics, SUDS and DU represents intra- and interobserver variations. Automatic vs. Manual DC Mean (SD) d H (mm) Mean (SD) DC Mean(SD) Manual vs. Manual SUSS 0.90(0.03) 3.8(1.6) 1 a 0 a SUDS/ intraobserver d H (mm) Mean (SD) 0.88(0.04) 4.3(1.5) 0.88(0.04) 4.3(1.7) DU/interobserver 0.87(0.04) 4.8(1.7) 0.87(0.04) 4.7(1.6) a Identical contours. M (1.6) 4.4(1.4) 4.8(1.8) 5.0(1.6) 4.3(1.4) M (1.6) 0 5.8(1.7) 4.9(2.0) 4.7(1.7) 5.3(1.9) M (1.4) 5.8(1.7) 0 4.2(1.8) 4.5(1.7) 4.0(1.4) M (1.8) 4.9(2.0) 4.2(1.80) 0 3.8(1.4) 4.4(1.6) M (1.6) 4.7(1.7) 4.5(1.7) 3.8(1.4) 0 3.8(1.6) M (1.4) 5.3(1.9) 4.0(1.4) 4.4(1.6) 3.8(1.6) 0 TABLE III. Dice Coefficient (Mean(SD)) comparing manual vs. manual contours. Bolded elements represent intraobserver (SUDS) comparisons while underlined elements represent interobserver (DU) comparisons. M (0.03) 0.87(0.04) 0.85(0.05) 0.86(0.04) 0.88(0.04) M (0.03) (0.04) 0.85(0.05) 0.86(0.04) 0.87(0.04) M (0.04) 0.86(0.04) (0.06) 0.87(0.04) 0.88(0.04) M (0.05) 0.85(0.05) 0.86(0.06) (0.04) 0.86(0.05) M (0.04) 0.86(0.04) 0.87(0.04) 0.87(0.04) (0.04) M (0.04) 0.87(0.04) 0.88(0.04) 0.86(0.05) 0.89(0.04) 1
6 312 Yip et al.: Autocontouring and observer variability 312 previously used by our group 6,12 and others, 8 then the accuracy of algorithm is reflected in the same user same session match (SUSS) DC of 0.90(0.03) and d H of 3.8(1.6) mm. Clearly, the algorithm does not generate a perfect match (DC = 1, d H = 0) with the gold standard. However, this error may be attributed to the inherent variations in manual contouring process. If the same expert on a different session, or another expert, creates another set of manual contours on the same set of images that is compared against the original gold standard, significant variation can be observed, (i.e., intraobserver DC = 0.88(0.04), d H = 4.3(1.7) mm, interobserver DC = 0.87(0.04), d H = 4.7(1.6) mm). Since there is no way to determine which set of manual contours is the actual truth, the gold standard is only as accurate as these intra- and interobserver variations indicate. Hence, we argue that, if a user s manually drawn contours are used as the standard for comparison, an autocontour is considered to be accurate if it has same-user automatic vs. manual agreement metrics (SUSS/SUDS) comparable to intraobserver variations of that individual user, and a different-user automatic vs. manual agreement metrics (DU) comparable to interobserver variations between that user and other users. Table V gives a side by side comparison of equivalent metrics for manual vs. manual and automatic vs. manual comparisons. In our limited study of a modest number of patients (6), observers (3), and sessions (2), these results indicate that our user trained automatic algorithm is indeed as accurate as the human experts, in both the same-user and different-user scenarios. Since the patients in this study were imaged on a 3 T MRI system, a few comments on the general applicability of this study are warranted. In particular, the contrast with noise ratio of tumor depends on the main magnetic field strength 29 of the MRI and the Linac-MR hybrids typically operate at lower field strengths 2 5 than 3 T. Therefore, the quality of the acquired images and contouring performance may be different for the actual Linac-MR systems. Additionally, other factors, such as choice of MRI sequence, scan parameters (i.e., TE/TR, voxel size, etc.), and presence of image artifacts from various sources (including bias field artifacts, which are not corrected for in this study) may all affect image quality and contouring performance. However, in our previous work we have tested this autocontouring algorithm with images degraded retrospectively by additional noise and k-space undersampling, and has shown that the algorithm works reasonably well with lower image quality scenarios, albeit with slightly poorer performance metrics. Our intra/interobserver agreement (mean DC: 0.88/0.87) are higher compared to the some previous lung tumor contouring studies on CT (mean DC:0.51/0.51) vs. 4DCT (mean DC: 0.80/0.80) images, 24 as well as on nonregistered PET- CT (median DC: 0.58/0.61) vs. registered PET-CT (median DC: 0.71/0.70) images. 23 However, there are major differences in the study protocols (i.e., 2D contouring vs. 3D contouring) and in imaging modality (dynamic MRI vs. CT/ 4DCT/PET-CT). Therefore, a direct comparison of the results in this study to the previous studies is not advised. In summary, our trained algorithm s ability to match its expert user s contours (SUSS, SUDS) is comparable to the uncertainties from intraobserver variations from those experts, while our algorithm s ability to match to a different expert s contours (DU) is comparable to the uncertainties from interobserver variations. These results show that the autocontouring algorithm can, in less than 20 ms, produce contours on dynamic lung MR images that are comparable, in terms of accuracy, to that of human experts who generally takes several seconds to produce a manual contour. However, our interpretations of these results are limited to that of lung tumors, where there is a naturally a large amount of contrast between tumor and background. The ability of our algorithm to perform in other tumor sites (i.e., liver, pancreas) would require further investigations. In terms of implementation of the autocontouring algorithm on Linac-MR hybrids, we have integrated a prior version of this software with a preclinical 0.2 T Linac-MR prototype 2 and demonstrated its capabilities to deliver conformal radiation to a moving phantom by imaging/autocontouring/mlc reshaping in real time. 10 For real time tumor tracking radiation delivery in-vivo, further studies are needed to validate the algorithm in a larger cohort of patients, with the ultimate aim of integrating this software to perform real time tumor tracking with a 0.5 T clinical Linac- MR system 3 (Aurora RT, MagnetT x Oncology Solutions). 5. CONCLUSIONS In this study, we evaluated our individual user-trained autocontouring algorithm on dynamic lung MR images for 6 NSCLC patients with multiple users. Autocontours from the trained algorithm for a particular expert agrees with the manual contours of the same user within the intraobserver variations. Autocontours from the trained algorithm by a particular expert agrees with the manual contours of different experts within the interobserver variations. These results suggest that the user trained autocontouring algorithm is capable of tracking a moving tumor on MR images as accurately as human experts, but with much faster speed (<20 ms). This algorithm may be a key component of the real time tumor tracking workflow for our hybrid Linac-MR device in the future. ACKNOWLEDGMENTS This work is supported by Alberta Innovates: Health Solutions, CRIO Team grant. CONFLICTS OF INTEREST Fallone is a cofounder and CEO of MagnetT x Oncology Solutions. a) Author to whom correspondence should be addressed. Electronic mail: eyip@ualberta.ca.
7 313 Yip et al.: Autocontouring and observer variability 313 REFERENCES 1. Keall PJ, Mageras GS, Balter JM, et al. The management of respiratory motion in radiation oncology report of AAPM Task Group 76. Med Phys. 2006;33: Fallone BG, Murray B, Rathee S, et al. First MR images obtained during megavoltage photon irradiation from a prototype integrated linac- MR system. Med Phys. 2009;36: Fallone BG. The rotating biplanar linac magnetic resonance imaging system. Semin Radiat Oncol. 2014;24: Raaymakers B, Lagendijk J, Overweg J, et al. Integrating a 1.5 T MRI scanner with a 6 MV accelerator: proof of concept. Phys Med Biol. 2009;54:N229 N Mutic S, Dempsey JF. The viewray system: magnetic resonance guided and controlled radiotherapy. Semin Radiat Oncol. 2014;24: Yun J, Yip E, Gabos Z, Wachowicz K, Rathee S, Fallone BG. Neuralnetwork based autocontouring algorithm for intrafractional lung-tumor tracking using Linac-MR. Med Phys. 2015;42: Cervino L, Du J, Jiang S. MRI-guided tumor tracking in lung cancer radiotherapy. Phys Med Biol. 2011;56: Bourque AE, Bedwani S, Filion E, Carrier J. A particle filter based autocontouring algorithm for lung tumor tracking using dynamic magnetic resonance imaging. Med Phys. 2016;43: Sawant A, Keall P, Pauly KB, et al. Investigating the feasibility of rapid MRI for image-guided motion management in lung cancer radiotherapy. Biomed Res Int. 2014; 2014: Yun J, Wachowicz K, Mackenzie M, Rathee S, Robinson D, Fallone B. First demonstration of intrafractional tumor-tracked irradiation using 2D phantom MR images on a prototype linac-mr. Med Phys. 2013;40: Deasy JO, Blanco AI, Clark VH. CERR: a computational environment for radiotherapy research. Med Phys. 2003;30: Yun J, Yip E, Gabos Z, Wachowicz K, Rathee S, Fallone B. Improved lung tumor autocontouring algorithm for intrafractional tumor tracking using 0.5 T linac-mr. Biomed Phys Eng Express. 2016; 2: Eckhorn R, Reitboeck H, Arndt M, Dicke P. Feature Linking Via Stimulus-Evoked Oscillations: Experimental Results From Cat Visual Cortex and Functional Implications From a Network Model. Presented at Proc in Washington, DC: ICNN; Johnson JL, Padgett ML. PCNN models and applications. IEEE Trans Neural Networks. 1999;10: Lindblad T, Kinser JM, Taylor J. Image Processing Using Pulse- Coupled Neural Networks, (3rd edn). Berlin: Springer; Ma Y, Zhan K, Wang Z. Applications of Pulse-Coupled Neural Networks. Berlin, Heidelberg: Springer, Kuntimad G, Ranganath HS. Perfect image segmentation using pulse coupled neural networks. IEEE Trans Neural Networks. 1999;10: Otsu N. A threshold selection method from gray-level histograms. Automatica. 1975;11: Zhan Z, Zhang J, Li Y, Chung HS. Adaptive particle swarm optimization. IEEE Trans Syst Man Cybern B Cybern. 2009;39: Dice LR. Measures of the amount of ecologic association between species. Ecology. 1945;26: Huttenlocher DP, Klanderman GA, Rucklidge WJ. Comparing images using the Hausdorff distance. IEEE Trans Pattern Anal Mach Intell. 1993;15: Erasmus JJ, Gladish GW, Broemeling L, et al. Interobserver and intraobserver variability in measurement of non small-cell carcinoma lung lesions: implications for assessment of tumor response. J Clin Oncol. 2003;21: Fox JL, Rengan R, O Meara W, et al. Does registration of PET and planning CT images decrease interobserver and intraobserver variation in delineating tumor volumes for non small-cell lung cancer?. Int J Radiat Oncol Biol Phys. 2005;62: Louie AV, Rodrigues G, Olsthoorn J, et al. Inter-observer and intraobserver reliability for lung cancer target volume delineation in the 4D- CT era. Radiother Oncol. 2010;95: Kaza E, Dunlop A, Panek R, et al. Lung volume reproducibility under ABC control and self-sustained breath-holding. J Appl Clin Med Phys. 2017;18: Cui L, Yin J, Hu C, Gong S, Xu J, Yang J. Inter-and intraobserver agreement of ADC measurements of lung cancer in free breathing, breathhold and respiratory triggered diffusion-weighted MRI. Clin Imaging. 2016;40: Bernardin L, Douglas N, Collins D, Giles S, O Flynn E, Orton M. Diffusion-weighted magnetic resonance imaging for assessment of lung lesions: repeatability of the apparent diffusion coefficient measurement. Eur Radiol. 2014;24: Valentini V, Boldrini L, Damiani A, Muren LP. Recommendations on how to establish evidence from auto-segmentation software in radiotherapy. Radiother Oncol. 2014;112: Wachowicz K, De Zanche N, Yip E, Volotovskyy V, Fallone BG. CNR considerations for rapid real-time MRI tumor tracking in radiotherapy hybrid devices: effects of B0 field strength. Med Phys. 2016;43: Yip E, Yun J, Wachowicz K, et al. Prior data assisted compressed sensing: a novel MR imaging strategy for real time tracking of lung tumors. Med Phys. 2014;41: Yip E, Yun J, Wachowicz K, Gabos Z, Rathee S, Fallone BG. Sliding window prior data assisted compressed sensing for MRI tracking of lung tumors. Med Phys. 2017;44: Dietz B, Yip E, Yun J, Fallone BG, Wachowicz K. Real-time dynamic MR image reconstruction using compressed sensing and principal component analysis (CS-PCA): demonstration in lung tumor tracking. Med Phys. 2017;44:
Sliding window prior data assisted compressed sensing for MRI tracking of lung tumors
Sliding window prior data assisted compressed sensing for MRI tracking of lung tumors Eugene Yip Department of Oncology, Medical Physics Division, University of Alberta, 11560 University Avenue, Edmonton,
More informationState-of-the-Art IGRT
in partnership with State-of-the-Art IGRT Exploring the Potential of High-Precision Dose Delivery and Real-Time Knowledge of the Target Volume Location Antje-Christin Knopf IOP Medical Physics Group Scientific
More informationImage Guidance and Beam Level Imaging in Digital Linacs
Image Guidance and Beam Level Imaging in Digital Linacs Ruijiang Li, Ph.D. Department of Radiation Oncology Stanford University School of Medicine 2014 AAPM Therapy Educational Course Disclosure Research
More informationA fluence convolution method to account for respiratory motion in three-dimensional dose calculations of the liver: A Monte Carlo study
A fluence convolution method to account for respiratory motion in three-dimensional dose calculations of the liver: A Monte Carlo study Indrin J. Chetty, a) Mihaela Rosu, Neelam Tyagi, Lon H. Marsh, Daniel
More informationMotion artifact detection in four-dimensional computed tomography images
Motion artifact detection in four-dimensional computed tomography images G Bouilhol 1,, M Ayadi, R Pinho, S Rit 1, and D Sarrut 1, 1 University of Lyon, CREATIS; CNRS UMR 5; Inserm U144; INSA-Lyon; University
More informationFiber Selection from Diffusion Tensor Data based on Boolean Operators
Fiber Selection from Diffusion Tensor Data based on Boolean Operators D. Merhof 1, G. Greiner 2, M. Buchfelder 3, C. Nimsky 4 1 Visual Computing, University of Konstanz, Konstanz, Germany 2 Computer Graphics
More information8/3/2017. Contour Assessment for Quality Assurance and Data Mining. Objective. Outline. Tom Purdie, PhD, MCCPM
Contour Assessment for Quality Assurance and Data Mining Tom Purdie, PhD, MCCPM Objective Understand the state-of-the-art in contour assessment for quality assurance including data mining-based techniques
More informationLucy Phantom MR Grid Evaluation
Lucy Phantom MR Grid Evaluation Anil Sethi, PhD Loyola University Medical Center, Maywood, IL 60153 November 2015 I. Introduction: The MR distortion grid, used as an insert with Lucy 3D QA phantom, is
More informationMR-guided radiotherapy: Vision, status and research at the UMC Utrecht. Dipl. Ing. Dr. Markus Glitzner
MR-guided radiotherapy: Vision, status and research at the UMC Utrecht Dipl. Ing. Dr. Markus Glitzner About myself Training Medizintechnik TU Graz PhD UMC Utrecht Clinical work Software implementation
More informationImprovement and Evaluation of a Time-of-Flight-based Patient Positioning System
Improvement and Evaluation of a Time-of-Flight-based Patient Positioning System Simon Placht, Christian Schaller, Michael Balda, André Adelt, Christian Ulrich, Joachim Hornegger Pattern Recognition Lab,
More informationANALYSIS OF PULMONARY FIBROSIS IN MRI, USING AN ELASTIC REGISTRATION TECHNIQUE IN A MODEL OF FIBROSIS: Scleroderma
ANALYSIS OF PULMONARY FIBROSIS IN MRI, USING AN ELASTIC REGISTRATION TECHNIQUE IN A MODEL OF FIBROSIS: Scleroderma ORAL DEFENSE 8 th of September 2017 Charlotte MARTIN Supervisor: Pr. MP REVEL M2 Bio Medical
More informationImage-based Monte Carlo calculations for dosimetry
Image-based Monte Carlo calculations for dosimetry Irène Buvat Imagerie et Modélisation en Neurobiologie et Cancérologie UMR 8165 CNRS Universités Paris 7 et Paris 11 Orsay, France buvat@imnc.in2p3.fr
More informationRADIOMICS: potential role in the clinics and challenges
27 giugno 2018 Dipartimento di Fisica Università degli Studi di Milano RADIOMICS: potential role in the clinics and challenges Dr. Francesca Botta Medical Physicist Istituto Europeo di Oncologia (Milano)
More information7/31/2011. Learning Objective. Video Positioning. 3D Surface Imaging by VisionRT
CLINICAL COMMISSIONING AND ACCEPTANCE TESTING OF A 3D SURFACE MATCHING SYSTEM Hania Al-Hallaq, Ph.D. Assistant Professor Radiation Oncology The University of Chicago Learning Objective Describe acceptance
More informationAn investigation of temporal resolution parameters in cine-mode four-dimensional computed tomography acquisition
JOURNAL OF APPLIED CLINICAL MEDICAL PHYSICS, VOLUME 9, NUMBER 4, FALL 2008 An investigation of temporal resolution parameters in cine-mode four-dimensional computed tomography acquisition Yildirim D. Mutaf
More informationAuto-Segmentation Using Deformable Image Registration. Disclosure. Objectives 8/4/2011
Auto-Segmentation Using Deformable Image Registration Lei Dong, Ph.D. Dept. of Radiation Physics University of Texas MD Anderson Cancer Center, Houston, Texas AAPM Therapy Educational Course Aug. 4th 2011
More informationHelical 4D CT pitch management for the Brilliance CT Big Bore in clinical practice
JOURNAL OF APPLIED CLINICAL MEDICAL PHYSICS, VOLUME 16, NUMBER 3, 2015 Helical 4D CT pitch management for the Brilliance CT Big Bore in clinical practice Guido Hilgers, a Tonnis Nuver, André Minken Department
More informationUse of MRI in Radiotherapy: Technical Consideration
Use of MRI in Radiotherapy: Technical Consideration Yanle Hu, PhD Department of Radiation Oncology, Mayo Clinic Arizona 04/07/2018 2015 MFMER slide-1 Conflict of Interest: None 2015 MFMER slide-2 Objectives
More informationIntraoperative Prostate Tracking with Slice-to-Volume Registration in MR
Intraoperative Prostate Tracking with Slice-to-Volume Registration in MR Sean Gill a, Purang Abolmaesumi a,b, Siddharth Vikal a, Parvin Mousavi a and Gabor Fichtinger a,b,* (a) School of Computing, Queen
More informationAn Automated Image-based Method for Multi-Leaf Collimator Positioning Verification in Intensity Modulated Radiation Therapy
An Automated Image-based Method for Multi-Leaf Collimator Positioning Verification in Intensity Modulated Radiation Therapy Chenyang Xu 1, Siemens Corporate Research, Inc., Princeton, NJ, USA Xiaolei Huang,
More informationThe team. Disclosures. Ultrasound Guidance During Radiation Delivery: Confronting the Treatment Interference Challenge.
Ultrasound Guidance During Radiation Delivery: Confronting the Treatment Interference Challenge Dimitre Hristov Radiation Oncology Stanford University The team Renhui Gong 1 Magdalena Bazalova-Carter 1
More informationA Generation Methodology for Numerical Phantoms with Statistically Relevant Variability of Geometric and Physical Properties
A Generation Methodology for Numerical Phantoms with Statistically Relevant Variability of Geometric and Physical Properties Steven Dolly 1, Eric Ehler 1, Yang Lou 2, Mark Anastasio 2, Hua Li 2 (1) University
More informationQuantitative imaging for clinical dosimetry
Quantitative imaging for clinical dosimetry Irène Buvat Laboratoire d Imagerie Fonctionnelle U678 INSERM - UPMC CHU Pitié-Salpêtrière, Paris buvat@imed.jussieu.fr http://www.guillemet.org/irene Methodology
More informationICARO Vienna April Implementing 3D conformal radiotherapy and IMRT in clinical practice: Recommendations of IAEA- TECDOC-1588
ICARO Vienna April 27-29 2009 Implementing 3D conformal radiotherapy and IMRT in clinical practice: Recommendations of IAEA- TECDOC-1588 M. Saiful Huq, Ph.D., Professor and Director, Dept. of Radiation
More informationClinical Prospects and Technological Challenges for Multimodality Imaging Applications in Radiotherapy Treatment Planning
Clinical Prospects and Technological Challenges for Multimodality Imaging Applications in Radiotherapy Treatment Planning Issam El Naqa, PhD Assistant Professor Department of Radiation Oncology Washington
More informationVirtual Phantoms for IGRT QA
TM Virtual Phantoms for IGRT QA Why ImSimQA? ImSimQA was developed to overcome the limitations of physical phantoms for testing modern medical imaging and radiation therapy software systems, when there
More information3DVH : SUN NUCLEAR On The Accuracy Of The corporation Planned Dose Perturbation Algorithm Your Most Valuable QA and Dosimetry Tools *Patent Pending
3DVH : On The Accuracy Of The Planned Dose Perturbation Algorithm SUN NUCLEAR corporation Your Most Valuable QA and Dosimetry Tools *Patent Pending introduction State-of-the-art IMRT QA of static gantry
More informationRespiratory Motion Estimation using a 3D Diaphragm Model
Respiratory Motion Estimation using a 3D Diaphragm Model Marco Bögel 1,2, Christian Riess 1,2, Andreas Maier 1, Joachim Hornegger 1, Rebecca Fahrig 2 1 Pattern Recognition Lab, FAU Erlangen-Nürnberg 2
More informationAutomated Quality Assurance for Image-Guided Radiation Therapy
JOURNAL OF APPLIED CLINICAL MEDICAL PHYSICS, VOLUME 10, NUMBER 1, WINTER 2009 Automated Quality Assurance for Image-Guided Radiation Therapy Eduard Schreibmann, a Eric Elder, Tim Fox Department of Radiation
More informationREAL-TIME ADAPTIVITY IN HEAD-AND-NECK AND LUNG CANCER RADIOTHERAPY IN A GPU ENVIRONMENT
REAL-TIME ADAPTIVITY IN HEAD-AND-NECK AND LUNG CANCER RADIOTHERAPY IN A GPU ENVIRONMENT Anand P Santhanam Assistant Professor, Department of Radiation Oncology OUTLINE Adaptive radiotherapy for head and
More informationUsing a research real-time control interface to go beyond dynamic MLC tracking
in partnership with Using a research real-time control interface to go beyond dynamic MLC tracking Dr. Simeon Nill Joint Department of Physics at The Institute of Cancer Research and the Royal Marsden
More informationChapter 3 Set Redundancy in Magnetic Resonance Brain Images
16 Chapter 3 Set Redundancy in Magnetic Resonance Brain Images 3.1 MRI (magnetic resonance imaging) MRI is a technique of measuring physical structure within the human anatomy. Our proposed research focuses
More informationImage Co-Registration II: TG132 Quality Assurance for Image Registration. Image Co-Registration II: TG132 Quality Assurance for Image Registration
Image Co-Registration II: TG132 Quality Assurance for Image Registration Preliminary Recommendations from TG 132* Kristy Brock, Sasa Mutic, Todd McNutt, Hua Li, and Marc Kessler *Recommendations are NOT
More informationADVANCING CANCER TREATMENT
The RayPlan treatment planning system makes proven, innovative RayStation technology accessible to clinics that need a cost-effective and streamlined solution. Fast, efficient and straightforward to use,
More informationUvA-DARE (Digital Academic Repository) Motion compensation for 4D PET/CT Kruis, M.F. Link to publication
UvA-DARE (Digital Academic Repository) Motion compensation for 4D PET/CT Kruis, M.F. Link to publication Citation for published version (APA): Kruis, M. F. (2014). Motion compensation for 4D PET/CT General
More informationAdaptive algebraic reconstruction technique
Adaptive algebraic reconstruction technique Wenkai Lua) Department of Automation, Key State Lab of Intelligent Technology and System, Tsinghua University, Beijing 10084, People s Republic of China Fang-Fang
More informationarxiv: v1 [cs.cv] 6 Jun 2017
Volume Calculation of CT lung Lesions based on Halton Low-discrepancy Sequences Liansheng Wang a, Shusheng Li a, and Shuo Li b a Department of Computer Science, Xiamen University, Xiamen, China b Dept.
More informationRespiratory Motion Compensation for Simultaneous PET/MR Based on Strongly Undersampled Radial MR Data
Respiratory Motion Compensation for Simultaneous PET/MR Based on Strongly Undersampled Radial MR Data Christopher M Rank 1, Thorsten Heußer 1, Andreas Wetscherek 1, and Marc Kachelrieß 1 1 German Cancer
More informationPrototype of Silver Corpus Merging Framework
www.visceral.eu Prototype of Silver Corpus Merging Framework Deliverable number D3.3 Dissemination level Public Delivery data 30.4.2014 Status Authors Final Markus Krenn, Allan Hanbury, Georg Langs This
More informationTomoTherapy Related Projects. An image guidance alternative on Tomo Low dose MVCT reconstruction Patient Quality Assurance using Sinogram
TomoTherapy Related Projects An image guidance alternative on Tomo Low dose MVCT reconstruction Patient Quality Assurance using Sinogram Development of A Novel Image Guidance Alternative for Patient Localization
More informationTechnical aspects of SPECT and SPECT-CT. John Buscombe
Technical aspects of SPECT and SPECT-CT John Buscombe What does the clinician need to know? For SPECT What factors affect SPECT How those factors should be sought Looking for artefacts For SPECT-CT Issues
More informationSpatial-temporal Total Variation Regularization (STTVR) for 4D-CT Reconstruction
Spatial-temporal Total Variation Regularization (STTVR) for 4D-CT Reconstruction Haibo Wu a, b, Andreas Maier a, Rebecca Fahrig c, and Joachim Hornegger a, b a Pattern Recognition Lab (LME), Department
More informationVolumetric Cine Imaging for On-board Target Localization in Radiation Therapy. Wendy Beth Harris. Graduate Program in Medical Physics Duke University
Volumetric Cine Imaging for On-board Target Localization in Radiation Therapy by Wendy Beth Harris Graduate Program in Medical Physics Duke University Date: Approved: Lei Ren, Supervisor Fang-Fang Yin,
More informationEstimating 3D Respiratory Motion from Orbiting Views
Estimating 3D Respiratory Motion from Orbiting Views Rongping Zeng, Jeffrey A. Fessler, James M. Balter The University of Michigan Oct. 2005 Funding provided by NIH Grant P01 CA59827 Motivation Free-breathing
More informationClassification of Subject Motion for Improved Reconstruction of Dynamic Magnetic Resonance Imaging
1 CS 9 Final Project Classification of Subject Motion for Improved Reconstruction of Dynamic Magnetic Resonance Imaging Feiyu Chen Department of Electrical Engineering ABSTRACT Subject motion is a significant
More informationADVANCING CANCER TREATMENT
3 ADVANCING CANCER TREATMENT SUPPORTING CLINICS WORLDWIDE RaySearch is advancing cancer treatment through pioneering software. We believe software has un limited potential, and that it is now the driving
More informationPET/CT multimodality imaging for radiotherapy planning in lung cancer The medical physicist point of view Isabelle Gardin Rouen CHB and Quant.I.
PET/CT multimodality imaging for radiotherapy planning in lung cancer The medical physicist point of view Isabelle Gardin Rouen CHB and Quant.I.F (EA4108 FR CNRS 3638) Outline Specific acquisition conditions
More informationAbbie M. Diak, PhD Loyola University Medical Center Dept. of Radiation Oncology
Abbie M. Diak, PhD Loyola University Medical Center Dept. of Radiation Oncology Outline High Spectral and Spatial Resolution MR Imaging (HiSS) What it is How to do it Ways to use it HiSS for Radiation
More informationAutomatic Segmentation of Parotids from CT Scans Using Multiple Atlases
Automatic Segmentation of Parotids from CT Scans Using Multiple Atlases Jinzhong Yang, Yongbin Zhang, Lifei Zhang, and Lei Dong Department of Radiation Physics, University of Texas MD Anderson Cancer Center
More informationImad Ali 1 Nesreen Alsbou 2 Justin Jaskowiak 1 Salahuddin Ahmad 1. Abstract RADIATION ONCOLOGY PHYSICS
Received: 9 May 2017 Revised: 2 October 2017 Accepted: 21 November 2017 DOI: 10.1002/acm2.12246 RADIATION ONCOLOGY PHYSICS Quantitative evaluation of the performance of different deformable image registration
More informationPROSTATE CANCER DETECTION USING LABEL IMAGE CONSTRAINED MULTIATLAS SELECTION
PROSTATE CANCER DETECTION USING LABEL IMAGE CONSTRAINED MULTIATLAS SELECTION Ms. Vaibhavi Nandkumar Jagtap 1, Mr. Santosh D. Kale 2 1 PG Scholar, 2 Assistant Professor, Department of Electronics and Telecommunication,
More informationDeviceless respiratory motion correction in PET imaging exploring the potential of novel data driven strategies
g Deviceless respiratory motion correction in PET imaging exploring the potential of novel data driven strategies Presented by Adam Kesner, Ph.D., DABR Assistant Professor, Division of Radiological Sciences,
More informationFinite Element Simulation of Moving Targets in Radio Therapy
Finite Element Simulation of Moving Targets in Radio Therapy Pan Li, Gregor Remmert, Jürgen Biederer, Rolf Bendl Medical Physics, German Cancer Research Center, 69120 Heidelberg Email: pan.li@dkfz.de Abstract.
More informationAutomated segmentation methods for liver analysis in oncology applications
University of Szeged Department of Image Processing and Computer Graphics Automated segmentation methods for liver analysis in oncology applications Ph. D. Thesis László Ruskó Thesis Advisor Dr. Antal
More informationIs deformable image registration a solved problem?
Is deformable image registration a solved problem? Marcel van Herk On behalf of the imaging group of the RT department of NKI/AVL Amsterdam, the Netherlands DIR 1 Image registration Find translation.deformation
More informationA Non-Linear Image Registration Scheme for Real-Time Liver Ultrasound Tracking using Normalized Gradient Fields
A Non-Linear Image Registration Scheme for Real-Time Liver Ultrasound Tracking using Normalized Gradient Fields Lars König, Till Kipshagen and Jan Rühaak Fraunhofer MEVIS Project Group Image Registration,
More information3/27/2012 WHY SPECT / CT? SPECT / CT Basic Principles. Advantages of SPECT. Advantages of CT. Dr John C. Dickson, Principal Physicist UCLH
3/27/212 Advantages of SPECT SPECT / CT Basic Principles Dr John C. Dickson, Principal Physicist UCLH Institute of Nuclear Medicine, University College London Hospitals and University College London john.dickson@uclh.nhs.uk
More informationMachine Learning for Medical Image Analysis. A. Criminisi
Machine Learning for Medical Image Analysis A. Criminisi Overview Introduction to machine learning Decision forests Applications in medical image analysis Anatomy localization in CT Scans Spine Detection
More informationAn Investigation of a Model of Percentage Depth Dose for Irregularly Shaped Fields
Int. J. Cancer (Radiat. Oncol. Invest): 96, 140 145 (2001) 2001 Wiley-Liss, Inc. Publication of the International Union Against Cancer An Investigation of a Model of Percentage Depth Dose for Irregularly
More informationThe MSKCC Approach to IMRT. Outline
The MSKCC Approach to IMRT Spiridon V. Spirou, PhD Department of Medical Physics Memorial Sloan-Kettering Cancer Center New York, NY Outline Optimization Field splitting Delivery Independent verification
More informationAutomatic Detection and Segmentation of Kidneys in Magnetic Resonance Images Using Image Processing Techniques
Biomedical Statistics and Informatics 2017; 2(1): 22-26 http://www.sciencepublishinggroup.com/j/bsi doi: 10.11648/j.bsi.20170201.15 Automatic Detection and Segmentation of Kidneys in Magnetic Resonance
More informationValidation of GEANT4 for Accurate Modeling of 111 In SPECT Acquisition
Validation of GEANT4 for Accurate Modeling of 111 In SPECT Acquisition Bernd Schweizer, Andreas Goedicke Philips Technology Research Laboratories, Aachen, Germany bernd.schweizer@philips.com Abstract.
More informationEffects of the difference in tube voltage of the CT scanner on. dose calculation
Effects of the difference in tube voltage of the CT scanner on dose calculation Dong Joo Rhee, Sung-woo Kim, Dong Hyeok Jeong Medical and Radiological Physics Laboratory, Dongnam Institute of Radiological
More informationInitial Clinical Experience with 3D Surface Image Guidance
Initial Clinical Experience with 3D Surface Image Guidance Amanda Havnen-Smith, Ph.D. Minneapolis Radiation Oncology Ridges Radiation Therapy Center Burnsville, MN April 20 th, 2012 Non-funded research
More informationArtefakt-resistente Bewegungsschätzung für die bewegungskompensierte CT
Artefakt-resistente Bewegungsschätzung für die bewegungskompensierte CT Marcus Brehm 1,2, Thorsten Heußer 1, Pascal Paysan 3, Markus Oehlhafen 3, and Marc Kachelrieß 1,2 1 German Cancer Research Center
More information8/2/2016. Measures the degradation/distortion of the acquired image (relative to an ideal image) using a quantitative figure-of-merit
Ke Li Assistant Professor Department of Medical Physics and Department of Radiology School of Medicine and Public Health, University of Wisconsin-Madison This work is partially supported by an NIH Grant
More informationTHE SIMULATION OF THE 4 MV VARIAN LINAC WITH EXPERIMENTAL VALIDATION
2007 International Nuclear Atlantic Conference - INAC 2007 Santos, SP, Brazil, September 30 to October 5, 2007 ASSOCIAÇÃO BRASILEIRA DE ENERGIA NUCLEAR - ABEN ISBN: 978-85-99141-02-1 THE SIMULATION OF
More informationData. ModuLeaf Mini Multileaf Collimator Precision Beam Shaping for Advanced Radiotherapy
Data ModuLeaf Mini Multileaf Collimator Precision Beam Shaping for Advanced Radiotherapy ModuLeaf Mini Multileaf Collimator Precision Beam Shaping for Advanced Radiotherapy The ModuLeaf Mini Multileaf
More informationAnalysis of Acquisition Parameters That Caused Artifacts in Four-dimensional (4D) CT Images of Targets Undergoing Regular Motion
Original Article PROGRESS in MEDICAL PHYSICS Vol. 24, No. 4, December, 2013 http://dx.doi.org/10.14316/pmp.2013.24.4.243 Analysis of Acquisition Parameters That Caused Artifacts in Four-dimensional (4D)
More informationRespiratory Motion Compensation for C-arm CT Liver Imaging
Respiratory Motion Compensation for C-arm CT Liver Imaging Aline Sindel 1, Marco Bögel 1,2, Andreas Maier 1,2, Rebecca Fahrig 3, Joachim Hornegger 1,2, Arnd Dörfler 4 1 Pattern Recognition Lab, FAU Erlangen-Nürnberg
More informationOptimization of CT Simulation Imaging. Ingrid Reiser Dept. of Radiology The University of Chicago
Optimization of CT Simulation Imaging Ingrid Reiser Dept. of Radiology The University of Chicago Optimization of CT imaging Goal: Achieve image quality that allows to perform the task at hand (diagnostic
More informationCHAPTER 6 MODIFIED FUZZY TECHNIQUES BASED IMAGE SEGMENTATION
CHAPTER 6 MODIFIED FUZZY TECHNIQUES BASED IMAGE SEGMENTATION 6.1 INTRODUCTION Fuzzy logic based computational techniques are becoming increasingly important in the medical image analysis arena. The significant
More informationMoving Metal Artifact Reduction for Cone-Beam CT (CBCT) Scans of the Thorax Region
Moving Metal Artifact Reduction for Cone-Beam CT (CBCT) Scans of the Thorax Region Andreas Hahn 1,2, Sebastian Sauppe 1,2, Michael Knaup 1, and Marc Kachelrieß 1,2 1 German Cancer Research Center (DKFZ),
More information8/3/2016. Image Guidance Technologies. Introduction. Outline
8/3/26 Session: Image Guidance Technologies and Management Strategies Image Guidance Technologies Jenghwa Chang, Ph.D.,2 Department of Radiation Medicine, Northwell Health 2 Hofstra Northwell School of
More informationFIRST DEMONSTRATION OF COMBINED KV/MV IMAGE-GUIDED REAL-TIME DYNAMIC MULTILEAF-COLLIMATOR TARGET TRACKING
doi:10.1016/j.ijrobp.2009.02.012 Int. J. Radiation Oncology Biol. Phys., Vol. 74, No. 3, pp. 859 867, 2009 Copyright Ó 2009 Elsevier Inc. Printed in the USA. All rights reserved 0360-3016/09/$ see front
More informationComputer-Aided Diagnosis in Abdominal and Cardiac Radiology Using Neural Networks
Computer-Aided Diagnosis in Abdominal and Cardiac Radiology Using Neural Networks Du-Yih Tsai, Masaru Sekiya and Yongbum Lee Department of Radiological Technology, School of Health Sciences, Faculty of
More informationAdvanced Targeting Using Image Deformation. Justin Keister, MS DABR Aurora Health Care Kenosha, WI
Advanced Targeting Using Image Deformation Justin Keister, MS DABR Aurora Health Care Kenosha, WI History of Targeting The advance of IMRT and CT simulation has changed how targets are identified in radiation
More informationDevelopment and Optimization of Fourdimensional Magnetic Resonance Imaging (4D-MRI) for Radiation Therapy
Development and Optimization of Fourdimensional Magnetic Resonance Imaging (4D-MRI) for Radiation Therapy by Yilin Liu Medical Physics Graduate Program Duke University Date: Approved: Jing Cai, co-advisor
More informationRobust Lung Ventilation Assessment
Fifth International Workshop on Pulmonary Image Analysis -75- Robust Lung Ventilation Assessment Sven Kabus 1, Tobias Klinder 1, Tokihiro Yamamoto 2, Paul J. Keall 3, Billy W. Loo, Jr. 4, and Cristian
More informationMutual information based CT registration of the lung at exhale and inhale breathing states using thin-plate splines
Mutual information based CT registration of the lung at exhale and inhale breathing states using thin-plate splines Martha M. Coselmon, a) James M. Balter, Daniel L. McShan, and Marc L. Kessler Department
More informationUNIVERSITY OF SOUTHAMPTON
UNIVERSITY OF SOUTHAMPTON PHYS2007W1 SEMESTER 2 EXAMINATION 2014-2015 MEDICAL PHYSICS Duration: 120 MINS (2 hours) This paper contains 10 questions. Answer all questions in Section A and only two questions
More informationImage Quality Assessment and Quality Assurance of Advanced Imaging Systems for IGRT. AAPM Penn-Ohio Chapter Sep 25, 2015 Soyoung Lee, PhD
Image Quality Assessment and Quality Assurance of Advanced Imaging Systems for IGRT AAPM Penn-Ohio Chapter Sep 25, 2015 Soyoung Lee, PhD 1 Outline q Introduction q Imaging performances in 4D-CBCT Image
More informationA MORPHOLOGY-BASED FILTER STRUCTURE FOR EDGE-ENHANCING SMOOTHING
Proceedings of the 1994 IEEE International Conference on Image Processing (ICIP-94), pp. 530-534. (Austin, Texas, 13-16 November 1994.) A MORPHOLOGY-BASED FILTER STRUCTURE FOR EDGE-ENHANCING SMOOTHING
More informationFinancial disclosure. Onboard imaging modality for IGRT
Tetrahedron Beam Computed Tomography Based On Multi-Pixel X- Ray Source and Its Application in Image Guided Radiotherapy Tiezhi Zhang, Ph.D. Advanced X-ray imaging Lab Financial disclosure Patent royalty
More informationVALIDATION OF DIR. Raj Varadhan, PhD, DABMP Minneapolis Radiation Oncology
VALIDATION OF DIR Raj Varadhan, PhD, DABMP Minneapolis Radiation Oncology Overview Basics: Registration Framework, Theory Discuss Validation techniques Using Synthetic CT data & Phantoms What metrics to
More informationIterative regularization in intensity-modulated radiation therapy optimization. Carlsson, F. and Forsgren, A. Med. Phys. 33 (1), January 2006.
Iterative regularization in intensity-modulated radiation therapy optimization Carlsson, F. and Forsgren, A. Med. Phys. 33 (1), January 2006. 2 / 15 Plan 1 2 3 4 3 / 15 to paper The purpose of the paper
More informationIterative CT Reconstruction Using Curvelet-Based Regularization
Iterative CT Reconstruction Using Curvelet-Based Regularization Haibo Wu 1,2, Andreas Maier 1, Joachim Hornegger 1,2 1 Pattern Recognition Lab (LME), Department of Computer Science, 2 Graduate School in
More informationProstate Detection Using Principal Component Analysis
Prostate Detection Using Principal Component Analysis Aamir Virani (avirani@stanford.edu) CS 229 Machine Learning Stanford University 16 December 2005 Introduction During the past two decades, computed
More informationA simple method to test geometrical reliability of digital reconstructed radiograph (DRR)
JOURNAL OF APPLIED CLINICAL MEDICAL PHYSICS, VOLUME 11, NUMBER 1, WINTER 2010 A simple method to test geometrical reliability of digital reconstructed radiograph (DRR) Stefania Pallotta, a Marta Bucciolini
More informationiplan RT Image Advanced Contouring Workstation - Driving Physician Collaboration
iplan RT Image Advanced Contouring Workstation - Driving Physician Collaboration The iplan Contouring Workstation offers unique and innovative capabilities for faster contouring and consistent segmentation
More informationReal Time Tumor Motion Tracking with CyberKnife
Real Time Tumor Motion Tracking with CyberKnife Martina Descovich, Ph.D University of California San Francisco July 16, 2015 Learning objectives Review the principles of real-time tumor motion tracking
More informationPosition accuracy analysis of the stereotactic reference defined by the CBCT on Leksell Gamma Knife Icon
Position accuracy analysis of the stereotactic reference defined by the CBCT on Leksell Gamma Knife Icon WHITE PAPER Introduction An image guidance system based on Cone Beam CT (CBCT) is included in Leksell
More informationCreating a Knowledge Based Model using RapidPlan TM : The Henry Ford Experience
DVH Estimates Creating a Knowledge Based Model using RapidPlan TM : The Henry Ford Experience Karen Chin Snyder, MS, DABR AAMD Region V Meeting October 4, 2014 Disclosures The Department of Radiation Oncology
More informationBrilliance CT Big Bore.
1 2 2 There are two methods of RCCT acquisition in widespread clinical use: cine axial and helical. In RCCT with cine axial acquisition, repeat CT images are taken each couch position while recording respiration.
More informationTumor motion during liver SBRT
Tumor motion during liver SBRT - projects at Aarhus University Hospital - Per Poulsen, Esben Worm, Walther Fledelius, Morten Høyer Aarhus University Hospital, Denmark SBRT: Stereotactic Body Radiation
More informationSimple quality assurance method of dynamic tumor tracking with the gimbaled linac system using a light field
JOURNAL OF APPLIED CLINICAL MEDICAL PHYSICS, VOLUME 17, NUMBER 5, 2016 Simple quality assurance method of dynamic tumor tracking with the gimbaled linac system using a light field Hideharu Miura, 1a Shuichi
More information8/4/2016. Emerging Linac based SRS/SBRT Technologies with Modulated Arc Delivery. Disclosure. Introduction: Treatment delivery techniques
Emerging Linac based SRS/SBRT Technologies with Modulated Arc Delivery Lei Ren, Ph.D. Duke University Medical Center 2016 AAPM 58 th annual meeting, Educational Course, Therapy Track Disclosure I have
More informationInterior Tomography Approach for MRI-guided Radiation Therapy
June 2017, Xi'an Interior Tomography Approach for MRI-guided Radiation Therapy Xun Jia and Steve B. Jiang Department of Radiation Oncology, University of Texas Southwestern Medical Center, Dallas, TX 75390
More informationIMRT site-specific procedure: Prostate (CHHiP)
IMRT site-specific procedure: Prostate (CHHiP) Scope: To provide site specific instructions for the planning of CHHIP IMRT patients Responsibilities: Radiotherapy Physicists, HPC Registered Therapy Radiographers
More information3D Voxel-Based Volumetric Image Registration with Volume-View Guidance
3D Voxel-Based Volumetric Image Registration with Volume-View Guidance Guang Li*, Huchen Xie, Holly Ning, Deborah Citrin, Jacek Copala, Barbara Arora, Norman Coleman, Kevin Camphausen, and Robert Miller
More information