Conformal treatment planning using 4DCT can decrease ipsilateral lung dose and improve tumor coverage: A prospective 4DCT treatment planning study

Conformal treatment planning using 4DCT can decrease ipsilateral lung dose and improve tumor coverage: A prospective 4DCT treatment planning study

S286 I. J. Radiation Oncology 1050 ● Biology ● Physics Volume 60, Number 1, Supplement, 2004 Quantitation of the Four-Dimensional Computed Tomogr...

90KB Sizes 0 Downloads 63 Views

S286

I. J. Radiation Oncology

1050

● Biology ● Physics

Volume 60, Number 1, Supplement, 2004

Quantitation of the Four-Dimensional Computed Tomography Process

1

W. Lu, P. Parikh,1 I. ElNaqa,1 M. Nystrom,1 J. Hubenschmidt,1 S. Wahab,1 S. Mutic,1 A. Singh,1 G. Christensen,2 J. D. Bradley,1 D. A. Low1 1 Department of Radiation Oncology, Washington University, St. Louis, MO, 2Electrical and Computer Engineering, University of Iowa, Iowa City, IA Purpose/Objective: The process of four-dimensional computed tomography (4D CT) uses multislice CT scans that have been sorted or gated using a metric, spirometry in our case. The purpose of this work is to quantify the reconstructed dataset quality and especially the level of correlation between spirometry and internal-object motion. Good correlations will imply that CT reconstructions and motion measurements will be precise and accurate for treatment planning purposes. Materials/Methods: The 4D CT process uses multislice CT scans sorted by spirometry-measured tidal volume to reconstruct 4D dataset. A multislice CT scanner (1.5 mm slices) operated in cine´ mode was used to acquire 12 contiguous slices in each couch position for 15 consecutive scans (0.5 s rotation) while the patient underwent simultaneous spirometry measurements. A critical factor for 4D CT is quantifying the reconstructed dataset quality and especially the level of correlation between the metric used relative to internal-object motion. For this study, the internal air content within the lung was used as a surrogate for internal position measurements. Thresholding and image morphological operations were applied to delineate the aircontaining tissues (lungs, trachea, bronchi) from each CT slice. An internal air content (V) was then calculated. The relationship of this independent internal measure with spirometer-measured tidal volume (v) was found to be quite linear throughout the lungs and was used to determine the overall accuracy and precision of spirometry-sorted 4DCT. Results: The air content is sensitive to even very small changes in anatomical shapes that correspond to breathing. Inspection of the CT-scan air content as a function of spirometry-measured tidal volume showed excellent correlations (typically r ⬎ 0.99) throughout the lung volume. Because of the linear relationship, the ratio of internal air content to spirometry volume was indicative of the fraction of air change in each couch position. If the process worked correctly, the sum of these ratios would equal the ratio of room to lung air densities (1.05). For 11 patients, the mean value was 1.07⫾0.09 (Table 1), indicating the high quality of the spirometry-based image sorting. The statistical deviation from a linear relationship between V and v was used to determine the process precision. For most patients, the precision was better than 8%, with a mean value of 5.8% ⫾ 2.4%. The table shows the results for the individual patients. Conclusions: This quantitative analysis highlights the value of using spirometry as the metric in sorting CT scans. With respect to the commonly used 4DCT metrics, the accuracy measurement could only have been provided by spirometry. The precision was excellent and is approximately a factor of three better than that obtained using the abdominal height metric. The 4D CT reconstruction provides the CT data required to measure the 3D trajectory of the tumor and tissue in the lung during breathing. This work supported in part by R01CA976679.

1051

Conformal Treatment Planning Using 4DCT Can Decrease Ipsilateral Lung Dose and Improve Tumor Coverage: A Prospective 4DCT Treatment Planning Study

P. Parikh, S. Wahab, W. Lu, R. Nantz, J. Hubenschmidt, M. Nystrom, I. ElNaqa, B. Pierburg, J.D. Bradley, D.A. Low Radiation Oncology, Washington University School of Medicine, St. Louis, MO Purpose/Objective: To determine the full range of lung tumor motion in non-coached, free-breathing lung cancer patients and to use the data to evaluate 3D conformal radiation treatment planning techniques. Materials/Methods: 20 patients were enrolled in a prospective study of lung tumor motion. Each patient was immobilized in the treatment position and underwent quantitative spirometry during a high resolution CT scan. The CT scanner was operated in cine mode and acquired 15 sets of twelve contiguous 1.5 mm slices. The spirometer measured tidal volume and was compared with the imaged air volume using a semiautomatic autothreshold to correct for measurement drift. Each image was associated a tidal volume measurement, subdivided by inhalation or exhalation phase. Inhalation, mid-inhalation, exhalation, and mid-exhalation image datasets were fused with a standard helical CT scan acquired during the same session. The inhalation and

Proceedings of the 46th Annual ASTRO Meeting

exhalation tumor volumes (GTV) were combined to produce an internal target volume (ITVl) that implicitly assumed linear motion. A second ITV (ITVh) was produced by combining the ITVl and the mid-inspiration and mid-exhalation GTVs. This enabled an evaluation of the effects of internal hysteresis on the ITV definition. 3D conformal radiation therapy plans were generated to 1) prescribe 70 Gy to the isocenter of the GTV defined using the standard helical scan (GTVs) using a 2 cm block margin; and 2) prescribe 66.5 Gy (95% of 70 Gy) to cover 100% of ITVl and 3) prescribe 66.5 Gy to cover 100% of ITVh using customized margins for 2) and 3). Results: To date, 12 patients have been analyzed. 8 patients had upper lobe tumors and 8 patients had peripheral tumors. The tumors moved an average of 0.9 (range 0.3 to 1.5) cm. The ratio of ITVl to GTVs was 1.0 – 4.0 (mean 1.95, median 1.70) and the average ratio of ITVh to ITVl was 1.16. This indicated that the GTVs inaccurately modeled the moving tumor, while incorporation of hysteresis did not substantially impact the definition of ITV in all but one patient. Block margins required to cover the ITVh and ITVl ranged from 1.0 to 1.5 cm. The 2 cm margin around the GTVs overcompensated for tumor motion in at least one direction in all patients. This resulted in a 5% decrease (range 2% - 8%) in V20 of the ipsilateral lung for the ITV-based plans. In 3 patients, the GTVs was an inaccurate model for the ITV and even the 2 cm margin was insufficient to adequately cover the tumor (Figure 1). Conclusions: Motion analysis can decrease the ipsilateral lung dose in most patients by allowing margin reduction compared to standard clinical practice. It also shows inadequate tumor coverage in some patients. In certain patients, full tumor motion may not be adequately represented by inhalation and exhalation reconstructions. Supported by R01CA96679

1052

Fluoroscopic Characterization of Lung Tumor Breathing Motion During Respiratory Gating

S. Korreman,1 H. Mostafavi,2 A. Grow,1 A. Boyer,1 Q. Le1 Department of Radiation Oncology, Stanford University School of Medicine, Stanford, CA, 2Gintzon Research Center, Varian Medical Systems, Mountain View, CA 1

Purpose/Objective: This study aims to investigate the pattern of internal lung tumor breathing motion during respiratory gating based on external chest-wall motion monitoring. Materials/Methods: As part of an IRB approved study, patients underwent placement of gold fiducial markers directly into the tumor under CT-guidance. The RPM system (Varian Medical Systems, Palo Alto CA) was used to record several consecutive orthogonal fluoroscopic movies of internal fiducial movement during breathing, synchronously with the movement of an external marker placed on the chest-wall. All measurements were performed during audio breathing instructions customized to a comfortable breathing pattern for each patient. To date, seven patients have been included in this study. Gating thresholds were retrospectively superposed on the external marker breathing traces, and the internal movement of the gold fiducials within those gating intervals was analyzed along all three axis of motion. Amplitude and phase approaches to assignment of gating thresholds was compared. Gating thresholds were applied to both inspiration and expiration points of the breathing cycle, respectively. Breathing motion range was quantified as limited by the 5% and the 95% fractiles of the total breathing motion. Results: In general, very high correlation was observed between external and internal respiration traces (cross-correlation coefficients larger than 0.8). The patients ability to comply with breathing instructions was not a limiting factor for high internal-external motion correlation. Correspondingly, the reduction of internal movement by gating was mimicked well by the chosen gating reduction in external movement. For all cases, amplitude and/or phase gating gave a relative reduction in internal

S287