Low-dose CT of the thorax in cancer follow-up

Low-dose CT of the thorax in cancer follow-up

European Journal of Radiology 51 (2004) 169–174 Low-dose CT of the thorax in cancer follow-up Takayuki Yamada∗ , Shuichi Ono, Masahiro Tsuboi, Haruo ...

308KB Sizes 31 Downloads 66 Views

European Journal of Radiology 51 (2004) 169–174

Low-dose CT of the thorax in cancer follow-up Takayuki Yamada∗ , Shuichi Ono, Masahiro Tsuboi, Haruo Saito, Akihiro Sato, Toshio Matsuhashi, Tadashi Ishibashi, Shoki Takahashi Department of Diagnostic Radiology, Tohoku University, 1-1 Seiryo-machi Aoba-ku, Sendai 9808574, Japan Received 31 May 2003; received in revised form 26 September 2003; accepted 30 September 2003

Abstract Purpose: To evaluate the quality of low-dose computed tomography (CT) images in the follow-up of cancer patients. Materials and methods: We selected patients with urogenital (n = 7) or esophageal cancer (n = 13) who were attending routine follow-up between April and July 2001. After water and chest phantom studies to decide the scan parameters, postcontrast low-dose CT scans were obtained at 60 mA (45 mA s) with a smoothing kernel. Three radiologists reviewed the CT scans of the thorax independently for overall image quality and anatomic detail in both mediastinal and lung windows. They subjectively rated the images on a four-point scale (0: poor, 1: fair, 2: good, 3: excellent) according to graininess and sharpness. Results: The average score of the low-dose CT for the lung window was 2.85, which was equivalent to control images. The average score for the mediastinal window was 1.77, which was lower than that of the control CT scan (2.62, P < 0.001) and almost identical to that of the chest phantom experiment. Nine of the 20 cases had abnormal findings; low-dose CT scans depicted them well and offered sufficient information for diagnosis. The radiation exposure was reduced by about half. Conclusion: The image quality of low-dose thoracic CT was satisfactory for both mediastinal and lung windows in the follow-up of cancer patients. © 2003 Elsevier Ireland Ltd. All rights reserved. Keywords: Low-dose CT; Thorax; Cancer follow up

1. Introduction Studies in the United States and Europe suggest that computed tomography (CT) scans comprise only 3–11% of all radiological exams, yet contribute 35–45% of the total radiation dose to the patient population [1,2]. Further research into the complex relationship between radiation exposure, image quality, and diagnostic accuracy should be encouraged, in order to establish the minimum radiation dose necessary to provide adequate diagnostic information [2]. Since chest CT images have high contrast, a lower dose can be used for chest CT scans without necessarily causing the radiologist to miss important clinical findings. Several previous studies have evaluated the quality of low-dose chest CT images [3–6], but this technique was applied to pediatric CT [7] or screening for lung cancer [8–10]. For routine follow-up of cancer patients, we are gener-

∗ Corresponding author. Tel.: +81-22-717-7312; fax: +81-22-717-7316. E-mail address: [email protected] (T. Yamada).

ally concerned about recurrence or metastasis rather than carcinogenesis, the induction of genetic effects, or effects due to radiation exposure. Many patients undergo routine follow-up CT after their cancers have been treated, and some examinations are performed in spite of the low rate of metastasis for some cancers [11]. This method is rarely applied for this purpose and a protocol has not been established. In this study, we first decided the scan parameters to use in the clinical study with water and chest phantoms. Then, we performed low-dose thoracic CT for routine follow-up of cancer patients, without obtaining additional images, and evaluated image quality.

2. Materials and methods 2.1. Equipment A single-detector helical CT (Siemens, Somatom Plus4/S,) was used in every study. We used the same values for the kilovoltage (140 kVp), helical pitch (1:1), scan time

0720-048X/$ – see front matter © 2003 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.ejrad.2003.09.017

170

T. Yamada et al. / European Journal of Radiology 51 (2004) 169–174

(0.75 s), and slice thickness (10 mm) as in a regular chest CT. We altered both the tube current and kernel; kernel can compensate for image noise, especially for high-frequency space, with a trade-off for a loss in image sharpness. 2.2. Water phantom experiment We evaluated the noise and sharpness of images at different tube currents and kernels using a water phantom. First, we evaluated the noise. CT scans of the water phantom were obtained at four different tube currents (43, 60, 77, and 111 mA) in addition to the regular current (129 mA). All of the images were reconstructed using three kernels: AB30, AB40, and AB50. AB50 is used routinely; AB40 and AB30 have smoothing effects. AB30 has the greatest smoothing effect. Noise was measured in 15 images and represented as the Wiener spectrum. Next, a CT scan of the phantom at 129 mA was reconstructed using the same three kernels. The sharpness was represented using the modulation transfer factor (MTF). Smoothing kernels altered the MTF rather than decreasing the noise (Figs. 1 and 2).

Fig. 2. The modulation transfer factor (MTF) for each kernel. The smoother the effect of the kernel, the lower the MTF became.

2.3. Chest phantom simulation CT scans of a chest phantom (Kyoto-kagaku, Kyoto, Japan) were obtained (Fig. 3). AB30 was not used in this experiment because of the poor MTF observed in the water phantom experiment. CT scans for four lower tube currents were obtained with the AB40 smoothing kernel to decrease image noise. A routine CT scan (129 mA and AB50) was also obtained as a standard. Three radiologists evaluated the five images of the chest phantom independently. They subjectively rated the images on a four-point scale (0: poor, 1: fair, 2: good, and 3: excellent) according to graininess and sharpness. The graininess was observed mainly at the simulated mediastinum and the thoracic wall in the mediastinal window (Fig. 3a). The sharpness was evaluated for simulated vascular structures and simulated disease in the lung window (Fig. 3b–d). 2.4. Clinical study

Fig. 1. The Wiener spectrum (WS) at each lower tube current with smoothing kernels. The regular parameters are 129 mA and AB50. (a) With the kernel AB40. The image noise in the high-frequency area is almost the same as that of regular one. (b) With the kernel AB30. In the high-frequency space image noises of most tube current are superior to that of regular one.

Low-dose CT was performed as part of a routine follow-up study of cancer patients between April and July 2001. We selected routine follow-up of urogenital (n = 7) and esophageal (n = 3) cancers for this study, because these patients frequently undergo repeat CT scanning in our hospital. In all, 20 cancer patients were studied in this clinical trial, which was approved by the Institutional Review Board. These cases constituted part of the follow-up CT for cancer patients done at our institution. Low-dose CT scans were obtained at 60 mA (45 mA s) with kernel AB40, based on the results of the chest phantom study. CT scans were obtained from the thorax to the abdomen or pelvis. No additional scans using different parameters were obtained. All the examinations were postcontrast studies using 100 ml of contrast medium containing 300 mg/l. The contrast medium was injected automatically at 1.0 m/s. All patients underwent regular CT scanning (129 mA and AB50) within 1 year; the images obtained were used as

T. Yamada et al. / European Journal of Radiology 51 (2004) 169–174

171

Fig. 3. CT scans of a chest phantom. (a) Mediastinal window. Great vessels and chest wall are simulated as normal structure. Pleural effusion is simulated as abnormal finding. (b–d) Pulmonary vascular structures are simulated as normal structure. Nodular shadow, peripheral triangular shadow, and emphysematous change are simulated as abnormal finding.

control images for comparison. Abdominal and pelvic CT scans were not evaluated in this study. Three radiologists reviewed the images independently for overall image quality and anatomic detail in both the mediastinal and lung windows. They subjectively rated the images on a four-point scale (0: poor, 1: fair, 2: good, and 3: excellent) according to graininess and sharpness (Figs. 4–6). Graininess was observed mainly at the mediastinum and thoracic wall in the mediastinal window. The sharpness was evaluated using pulmonary vascular shadows in the lung window. Data were analyzed using the Wilcoxon signed-rank test. In addition, the CT dose index (CT DI) was calculated for the regular and low-dose scans.

3. Results 3.1. Water phantom experiment Smoothing kernels decrease the Wiener spectrum as compared with regular images, especially in high-frequency space (Fig. 1a and b), which can prove beneficial for observing fine structures, such as the peripheral lung parenchyma in lung windows. In exchange for decreasing image noise, the MTF is also decreased (Fig. 2), especially in the case of AB30, which caused a markedly worse MTF. Therefore, we selected kernel AB40 for the chest phantom experiment in order to decrease the noise while maintaining image sharpness. In kernel AB40, the image noise of the high-frequency

172

T. Yamada et al. / European Journal of Radiology 51 (2004) 169–174

Fig. 4. Comparison between control and low-dose CT scans for the lung window. A 69-year-old patient with esophageal cancer postoperatively. (a) Control CT scan. The image quality was judged “excellent”. (b) Low-dose CT scan. The image quality was also judged “excellent”. The nodular opacity in the right upper lobe and emphysematous changes showed up clearly. There was no significant interval change between the two scans.

Fig. 5. Comparison between control and low-dose CT scans for the mediastinal window. A 75-year-old patient with esophageal cancer postoperatively. (a) Control CT scan. The image quality was judged “excellent”. (b) Low-dose CT scan. The image quality was also judged “excellent” and was equivalent to the control image.

3.3. Clinical study area at 60 and 77 mA was almost the same as that of regular images (Fig. 1a). 3.2. Chest phantom simulation The average image quality scores are shown in Table 1. The scores for all the images in the lung window were the same. By contrast, in the mediastinal window, the image score decreased with the tube current, by one rank at 60 mA and by nearly two ranks at 43 mA. Since the mediastinum including the cardiovascular system and thoracic wall are not fine structures, we considered that a decrease of one rank as compared with the control scan was acceptable for the mediastinal window. Therefore, we selected a tube current of 60 mA for the clinical protocol.

The average image quality scores are shown in Table 2. Although the score in the lung window was significantly lower than that of the control images (P = 0.007), this Table 1 The score of the image quality at each tube current I (mA)

Lung window

Mediastinal window

129a 111b 77b 60b 43b

3 3 3 3 3

2.67 (0.58) 2.67 (0.58) 2.33 (0.58) 1.67 (0.58) 1 (0)

(0) (0) (0) (0) (0)

Numbers in parentheses are S.D. a The image was reconstructed with kernel AB50. b The images were reconstructed with kernel AB40.

T. Yamada et al. / European Journal of Radiology 51 (2004) 169–174

173

value by about one rank (P < 0.001). This result was almost identical to that of the chest phantom experiment. Positive findings were seen in 9 of the 20 cases. In the lung window, they included linear shadows (n = 5), small nodular or granular shadows (n = 4) (Fig. 4), emphysematous change (n = 1) (Fig. 4), and ground-glass opacity (n = 1). All three radiologists detected them. In the mediastinal window, the findings included pleural thickening (n = 2), pericardial thickening (n = 1), pleural effusion (n = 1) (Fig. 6), lung collapse (n = 1), and recurrent mediastinal mass (n = 1) (Fig. 6). Although the score of the low-dose CT was just below the rank “good”, all three radiologists detected these findings. The CT DI was 0.656 mGy at 129 mA and 0.303 mGy at 60 mA. Therefore, this clinical protocol reduced the radiation exposure by about one half.

4. Discussion

Fig. 6. Comparison between control and low-dose CT scans for the mediastinal window. A 53-year-old patient with esophageal cancer postoperatively. (a) Control CT scan. The image quality was judged “good”. (b) Low-dose CT scan obtained 2 months after the control examination. The image quality was judged “fair”. However, the increased right pleural effusion and enlarged mediastinal mass were both recognized and there was sufficient information to diagnose a recurrence.

was because all of the control scans were rated “excellent”. Nevertheless, the image quality of the low-dose CT for the lung window was equivalent to that of the control CT scans. Conversely, the average score of the low-dose CT in the mediastinal window was 1.77, which was just below the rank “good”. This was significantly lower than the control Table 2 The comparison of the image quality between low-dose and control CT scans

Lung window Mediastinal window

Low dose

Control



Pa

2.85 (0.40) 1.77 (0.67)

3.00 (0) 2.62 (0.49)

0.15 0.85

0.007 <0.001

Numbers in parentheses are S.D. a Wilcoxon signed-rank test.

In this study, we showed that low-dose CT can be used for routine follow-up of cancer patients without significant loss of information. Before starting the clinical study, we first investigated the Wiener spectrum (image noise) and sharpness (MTF) in a water phantom study. Our method is similar to that reported by Frush et al. for pediatric abdominal CT [12]. They evaluated the noise using a water phantom at different tube currents and determined the noise as the standard deviation of the CT values. However, no studies have investigated the MTF (sharpness) simultaneously. Image sharpness is considered important for observing peripheral structures, such as the pulmonary vasculature. Moreover, our method has three good points. Since complete chest CT scans cannot be obtained at a variety of radiation exposures because of ethical considerations, we performed a chest phantom study to decide the scan parameters to use in the clinical study. Our method enabled us to determine the parameters and to evaluate the image quality of the complete CT scans in a clinical setting. Since prior studies tried evaluating the image quality of low-dose CT in a clinical setting [3,4,6], one or a few image slices had to be added to the regular scans. In a study of follow-up chest CT for extrathoracic malignancy, a low dose of 20 mA was used, and every image was evaluated; however, the authors did not state why they adopted this tube current [13]. Second, we were also able to evaluate the image quality at the same level as the contrast enhancement in a typical study. In a prior study, two levels with both regular and reduced tube currents were added after the ordinary scanning to minimize differences in the level of contrast enhancement [3]. This limited the number of images evaluated and resulted in a different level of contrast enhancement from the ordinary scans. Third, most previous studies changed only the tube current [3,4,6]. However, parameters other than tube current also affect image noise and sharpness, including kernel, helical pitch [5], and slice thickness. As it is difficult to predict

174

T. Yamada et al. / European Journal of Radiology 51 (2004) 169–174

changes in image quality if several parameters are changed simultaneously, the more parameters to be changed, the more useful a phantom simulation is likely to be. In our study, the image quality in both the lung and mediastinal windows was considered practical for routine follow-up of cancer patients. The scores of the low-dose CT scans for the lung window were equivalent to control images. The image quality of the abnormal findings was also satisfactory, as all three radiologists observed the changes. In fact, lower doses (e.g., 20 and 50 mA) have been used in lung cancer screening or follow-up of extrapulmonary malignancies of lymphatic origin [8,9,13]. Therefore, our result for the lung window should not be surprising. The mediastinal window images suffered from increased noise with the low-dose CT. The tube current that we used gave an image quality just below “good”, which was one rank poorer than the quality of the control images. However, the image quality of abnormal findings including recurrent mediastinal mass was also sufficient for all three radiologists to detect the abnormalities. All the studies were postcontrast CT scans, because the examination included the abdomen and sometimes the pelvis. The contrast media likely improved the contrast between the vascular structures and the lesion, whereas our simulation involved a non-contrast phantom experiment and did not evaluate contrast. Therefore, it might be possible to reduce the tube current further. This study used only a single-detector helical CT. Recently, multi-detector row CT (MDCT) scanners have come into wide use. MDCT could contribute to increasing the radiation dose delivered to patients. Current MDCT scanners have software that modulates the tube current during tube rotation as a function of the projection angle. Although the radiation dose is reduced by as much as 20%, initial decreases in the mA s preset by the physician should be considered the primary tool [14]. There have been two reports on dose reduction using MDCT [15,16]. In one, dose-reduced thoracic MDCT was used to restage lung cancer and gave equivalent visualization of anatomic and pathologic structures and was of diagnostic quality [16]. However, the appropriate effective mA s still remains unclear. Our method can also be applied to MDCT and could help to determine the scan parameters of low-dose MDCT of the chest for routine follow-up of cancer patients. 5. Conclusion After simulating the scan parameters using water and chest phantoms, we performed low-dose chest CT for rou-

tine follow-up of some cancer patients, without subjecting the patients to additional scans. The image quality of the low-dose thoracic CT was satisfactory for both the mediastinal and lung windows and was considered practical for observing abnormal findings. The radiation exposure was reduced by about half.

References [1] Imhof H, Schibany N, Ba-Ssalamah A, Czerny C, Hojreh A, Kainberger F, et al. Spiral CT and radiation dose. Eur J Radiol 2003;47:29–37. [2] Mayo JR, Aldrich J, Muller NL. Radiation exposure at chest CT: a statement of the Fleischner society. Radiology 2003;228:15–21. [3] Naidich DP, Marshall CH, Gribbin C, Arams RS, McCauley DI. Low-dose CT of the lungs: preliminary observations. Radiology 1990;175:729–31. [4] Revenel JG, Scalzetti EM, Huda W, Garrisi W. Radiation exposure and image quality in chest CT examinations. Am J Roentgenol 2001;177:279–84. [5] Mahesh M, Scatarige JC, Cooper J, Fishman EK. Dose and pitch relationship for a particular multislice CT scanner. Am J Roentgenol 2001;177:1273–5. [6] Mayo JR, Whittall KP, Leung AN, Hartman TE, Park CS, Primack SL, et al. Simulated dose reduction in conventional chest CT: validation study. Radiology 1997;202:453–7. [7] Brenner DJ, Elliston CD, Hall EJ, Berdon WE. Estimated risks of radiation-induced fatal cancer from pediatric CT. Am J Roentgenol 2001;176:289–96. [8] Sone S, Takhasima S, Li F, et al. Mass screening for lung cancer with mobile spiral computed tomography scanner. Lancet 1998;351:1242– 5. [9] Bach PB, Kelly MJ, Tate RC, McCrory DC. Screening for lung cancer. A review for current literature. Chest 2003;123:72S–82S. [10] Garg K, Keith RL, Byers T, et al. Randomized controlled trial with low-dose spiral CT for lung cancer screening: feasibility study and preliminary results. Radiology 2002;225:506–10. [11] Sella T, Rosenbaum E, Edelmann DZ, Agid R, Bloom AI, Libson E. Value of chest CT scans in routine ovarian carcinoma follow-up. Am J Roentgenol 2001;177:857–9. [12] Frush DP, Slack CC, Hollingsworth CL, et al. Computer-simulated radiation dose reduction for abdominal multidetector CT for pediatric patients. Am J Roentgenol 2002;179:1107–13. [13] Sonnenshein M, Dinkel HP, Vock P. Low-dose multislice-CT of the thorax in the follow-up of malignancies of extrathoracic or lymphatic origin B-0061 ECR 2001. [14] Tack D, Maertelaer VD, Gevenois PA. Dose reduction in multidetector CT using attenuation-based online tube current modulation. Am J Roentgenol 2003;181:331–4. [15] Tack D, Sourtzis S, Delpierre I, Maertelaer VD, Gevenois PA. Low-dose unenhanced multidetector CT of patients with suspected renal colic. Am J Roentgenol 2003;180:305–11. [16] Rehbock B, Hieckel HG. Chest examination protocol with a reduced dose using a multi-slice spiral CT. Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr 2003;175:963–6.