Inter-rater agreement in glioma segmentations on longitudinal MRI.

BACKGROUND: Tumor segmentation of glioma on MRI is a technique to monitor, quantify and report disease progression. Manual MRI segmentation is the gold standard but very labor intensive. At present the quality of this gold standard is not known for different stages of the disease, and prior work has mainly focused on treatment-naive glioblastoma. In this paper we studied the inter-rater agreement of manual MRI segmentation of glioblastoma and WHO grade II-III glioma for novices and experts at three stages of disease. We also studied the impact of inter-observer variation on extent of resection and growth rate. METHODS: In 20 patients with WHO grade IV glioblastoma and 20 patients with WHO grade II-III glioma (defined as non-glioblastoma) both the enhancing and non-enhancing tumor elements were segmented on MRI, using specialized software, by four novices and four experts before surgery, after surgery and at time of tumor progression. We used the generalized conformity index (GCI) and the intra-class correlation coefficient (ICC) of tumor volume as main outcome measures for inter-rater agreement. RESULTS: For glioblastoma, segmentations by experts and novices were comparable. The inter-rater agreement of enhancing tumor elements was excellent before surgery (GCI 0.79, ICC 0.99) poor after surgery (GCI 0.32, ICC 0.92), and good at progression (GCI 0.65, ICC 0.91). For non-glioblastoma, the inter-rater agreement was generally higher between experts than between novices. The inter-rater agreement was excellent between experts before surgery (GCI 0.77, ICC 0.92), was reasonable after surgery (GCI 0.48, ICC 0.84), and good at progression (GCI 0.60, ICC 0.80). The inter-rater agreement was good between novices before surgery (GCI 0.66, ICC 0.73), was poor after surgery (GCI 0.33, ICC 0.55), and poor at progression (GCI 0.36, ICC 0.73). Further analysis showed that the lower inter-rater agreement of segmentation on postoperative MRI could only partly be explained by the smaller volumes and fragmentation of residual tumor. The median interquartile range of extent of resection between raters was 8.3% and of growth rate was 0.22 mm/year. CONCLUSION: Manual tumor segmentations on MRI have reasonable agreement for use in spatial and volumetric analysis. Agreement in spatial overlap is of concern with segmentation after surgery for glioblastoma and with segmentation of non-glioblastoma by non-experts.

Three-dimensional heart dose reconstruction to estimate normal tissue complication probability after breast irradiation using portal dosimetry.

Irradiation of the heart is one of the major concerns during radiotherapy of breast cancer. Three-dimensional (3D) treatment planning would therefore be useful but cannot always be performed for left-sided breast treatments, because CT data may not be available. However, even if 3D dose calculations are available and an estimate of the normal tissue damage can be made, uncertainties in patient positioning may significantly influence the heart dose during treatment. Therefore, 3D reconstruction of the actual heart dose during breast cancer treatment using electronic imaging portal device (EPID) dosimetry has been investigated. A previously described method to reconstruct the dose in the patient from treatment portal images at the radiological midsurface was used in combination with a simple geometrical model of the irradiated heart volume to enable calculation of dose-volume histograms (DVHs), to independently verify this aspect of the treatment without using 3D data from a planning CT scan. To investigate the accuracy of our method, the DVHs obtained with full 3D treatment planning system (TPS) calculations and those obtained after resampling the TPS dose in the radiological midsurface were compared for fifteen breast cancer patients for whom CT data were available. In addition, EPID dosimetry as well as 3D dose calculations using our TPS, film dosimetry, and ionization chamber measurements were performed in an anthropomorphic phantom. It was found that the dose reconstructed using EPID dosimetry and the dose calculated with the TPS agreed within 1.5% in the lung/heart region. The dose-volume histograms obtained with EPID dosimetry were used to estimate the normal tissue complication probability (NTCP) for late excess cardiac mortality. Although the accuracy of these NTCP calculations might be limited due to the uncertainty in the NTCP model, in combination with our portal dosimetry approach it allows incorporation of the actual heart dose. For the anthropomorphic phantom, and for fifteen patients for whom CT data were available to test our method, the average difference between the NTCP values obtained with our method and those resulting from the dose distributions calculated with the TPS was 0.1% +/- 0.3% (1 SD). Most NTCP values were 1%-2% lower than those obtained using the method described by Hurkmans et al. [Radiother. Oncol. 62, 163-171 (2002)], using the maximum heart distance determined from a simulator image as a single pre-treatment parameter. A similar difference between the two methods was found for twelve patients using in vivo EPID dosimetry; the average NTCP value obtained with EPID dosimetry was 0.9%, whereas an average NTCP value of 2.2% was derived using the method of Hurkmans et al. The results obtained in this study show that EPID dosimetry is well suited for in vivo verification of the heart dose during breast cancer treatment, and can be used to estimate the NTCP for late excess cardiac mortality. To the best of our knowledge, this is the first study using portal dosimetry to calculate a DVH and NTCP of an organ at risk.

The long-term stability of amorphous silicon flat panel imaging devices for dosimetry purposes.

This study was carried out to determine the stability of the response of amorphous silicon (a-Si)-flat panel imagers for dosimetry applications. Measurements of the imager's response under reference conditions were performed on a regular basis for four detectors of the same manufacturer. We found that the ambient temperature influenced the dark-field, while the gain of the imager signal was unaffected. Therefore, temperature fluctuations were corrected for by applying a "dynamic" darkfield correction. This correction method also removed the influence of a small, irreversible increase of the dark-field current, which was equal to 0.5% of the dynamic range of the imager per year and was probably caused by mild radiation damage to the a-Si array. By applying a dynamic dark-field correction, excellent stability of the response over the entire panel of all imagers of 0.5% (1 SD) was obtained over an observation period up to 23 months. However, two imagers had to be replaced after several months. For one imager, an image segment stopped functioning, while the image quality of the other imager degraded significantly. We conclude that the tested a-Si EPIDs have a very stable response and are therefore well suited for dosimetry. We recommend, however, applying quality assurance tests dedicated to both imaging and dosimetry.

The stability of liquid-filled matrix ionization chamber electronic portal imaging devices for dosimetry purposes.

This study was performed to determine the stability of liquid-filled matrix ionization chamber (LiFi-type) electronic portal imaging devices (EPID) for dosimetric purposes. The short- and long-term stability of the response was investigated, as well as the importance of factors influencing the response (e.g., temperature fluctuations, radiation damage, and the performance of the electronic hardware). It was shown that testing the performance of the electronic hardware as well as the short-term stability of the imagers may reveal the cause of a poor long-term stability of the imager response. In addition, the short-term stability was measured to verify the validity of the fitted dose-response curve immediately after beam startup. The long-term stability of these imagers could be considerably improved by correcting for room temperature fluctuations and gradual changes in response due to radiation damage. As a result, the reproducibility was better than 1% (1 SD) over a period of two years. The results of this study were used to formulate recommendations for a quality control program for portal dosimetry. The effect of such a program was assessed by comparing the results of portal dosimetry and in vivo dosimetry using diodes during the treatment of 31 prostate patients. The improvement of the results for portal dosimetry was consistent with the deviations observed with the reproducibility tests in that particular period. After a correction for the variation in response of the imager, the average difference between the measured and prescribed dose during the treatment of prostate patients was -0.7%+/-1.5% (1 SD), and -0.6%+/-1.1% (1 SD) for EPID and diode in vivo dosimetry, respectively. It can be concluded that a high stability of the response can be achieved for this type of EPID by applying a rigorous quality control program.

Dose-response and ghosting effects of an amorphous silicon electronic portal imaging device.

The purpose of this study was to investigate the dose-response characteristics, including ghosting effects, of an amorphous silicon-based electronic portal imaging device (a-Si EPID) under clinical conditions. EPID measurements were performed using one prototype and two commercial a-Si detectors on two linear accelerators: one with 4 and 6 MV and the other with 8 and 18 MV x-ray beams. First, the EPID signal and ionization chamber measurements in a mini-phantom were compared to determine the amount of buildup required for EPID dosimetry. Subsequently, EPID signal characteristics were studied as a function of dose per pulse, pulse repetition frequency (PRF) and total dose, as well as the effects of ghosting. There was an over-response of the EPID signal compared to the ionization chamber of up to 18%, with no additional buildup layer over an air gap range of 10 to 60 cm. The addition of a 2.5 mm thick copper plate sufficiently reduced this over-response to within 1% at clinically relevant patient-detector air gaps (> 40 cm). The response of the EPIDs varied by up to 8% over a large range of dose per pulse values, PRF values and number of monitor units. The EPID response showed an under-response at shorter beam times due to ghosting effects, which depended on the number of exposure frames for a fixed frame acquisition rate. With an appropriate build-up layer and corrections for dose per pulse, PRF and ghosting, the variation in the a-Si EPID response can be reduced to well within +/- 1%.

The computation of MR image distortions caused by tissue susceptibility using the boundary element method.

Static field inhomogeneity in magnetic resonance (MR) imaging produces geometrical distortions which restrict the clinical applicability of MR images, e.g., for planning of precision radiotherapy. The authors describe a method to compute distortions which are caused by the difference in magnetic susceptibility between the scanned object and the surrounding air. Such a method is useful for understanding how the distortions depend on the object geometry, and for correcting for geometrical distortions, and thereby improving MR/CT registration algorithms. The geometric distortions in MR can be directly computed from the magnetic field inhomogeneity and the applied gradients. The boundary value problem of computing the magnetic field inhomogeneity caused by susceptibility differences is analyzed. It is shown that the boundary element method (BEM) has several advantages over previously applied methods to compute the magnetic field. Starting from the BEM and the assumption that the susceptibilities are very small (typically O(10(-5)) or less), a formula is derived to compute the magnetic field directly, without the need to solve a large system of equations. The method is computationally very efficient when the magnetic field is needed at a limited number of points, e.g., to compute geometrical distortions of a set of markers or a single surface. In addition to its computational advantage the method proves to be efficient to correct for the lack of data outside the scan which normally causes large artifacts in the computed magnetic field. These artifacts can be reduced by assuming that at the scan boundary the object extends to infinity in the form of a generalized cylinder. With the adaptation of the BEM this assumption is equivalent to simply omitting the scan boundary from the computations. To the authors' knowledge, no such simple correction method exists for other computation methods. The accuracy of the algorithm was tested by comparing the BEM solution with the analytical solution for a sphere. When the applied homogeneous field is 1.5 T the agreement between both methods was within 0.11.10(-6) T. As an example, the method was applied to compute the displacement vector field of the surface of a human head, derived from an MR imaging data set. This example demonstrates that the distortions can be as large as 3 mm for points just outside the head when a gradient strength of 3 mT/m is used. It was also observed that distortion within the head can be described accurately as a linear scaling in the axial direction.