| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Thoracic Imaging |
1 From the Kurt Rossmann Laboratories for Radiologic Image Research, Department of Radiology, MC-2026, University of Chicago, 5841 S Maryland Ave, Chicago, IL 60637 (F.L., H.A., H.M., S.G.A., K.D.); and Azumi General Hospital, Ikeda, Nagano, Japan (S.S.). From the 2001 RSNA scientific assembly. Received August 13, 2001; revision requested September 20; final revision received July 3, 2002; accepted July 10. Supported in part by United States Public Health Service grant CA62625. Address correspondence to F.L. (e-mail: fli@kurt.bsd.uchicago.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Eighty-three primary lung cancers were found during an annual low-dose CT screening program and confirmed histopathologically at either surgery or biopsy. Thirty-two of these lung cancers were missed on 39 CT scans: on 23 scans owing to detection errors and on 16 owing to interpretation errors. The clinical characteristics, CT features, and histopathologic findings of these missed lung cancers were correlated.
RESULTS: All missed cancers were intrapulmonary, and 28 (88%) were stage IA. All 20 detection errors occurred in cases of adenocarcinoma, 17 (85%) of which were well-differentiated tumors and 11 (55%) of which were in nonsmoking women. The mean size of cancers missed owing to detection error, 9.8 mm, was smaller than that of cancers missed owing to interpretation error, 15.9 mm (P < .001). In the detection error group, the percentages of nodules with ground-glass opacity (91%) or judged to be subtle (91%) were greater than those of nodules in the interpretation error group (38% and 25%, respectively) (P < .001). In the detection error group, 83% (19/23) of cancers were overlapped with, obscured by, or similar in appearance to normal structures such as pulmonary vessels. On 14 of the 16 CT scans with which there were interpretation errors, the CT findings mimicked benign disease, and the patients also had underlying lung disease, such as tuberculosis, emphysema, or lung fibrosis.
CONCLUSION: The lung cancers missed at low-dose CT screening in this series generally were very subtle and appeared as small faint nodules, overlapping normal structures, or opacities in a complex background of other disease.
© RSNA, 2002
Index terms: Cancer screening, 60.32 • Lung neoplasms, CT, 60.12111, 60.12115 • Lung neoplasms, diagnosis, 60.31, 60.32 • Lung neoplasms, screening, 60.31, 60.32
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
The high percentage (54%–90%) and large average size (13–16 mm) of lung cancers missed on chest radiographs have been reported in previous studies, and missed cancers have been attributed mainly to impaired visualization due to the superimposition of extra- or intrathoracic normal structures, such as bones, hilar vessels, and the heart (10–12).
The number of lung cancers missed at CT and reported in the literature has been limited, probably because it is difficult to identify the missed cases among the many routine CT examinations performed in most medical centers (13,14). Gurney (13) reported that nine lung cancers that were missed at CT were identified from a monthly tumor registry that was maintained for about 10 years, and five of these tumors were peripheral lung cancers smaller than 3 mm in diameter, which was considered the threshold size for detectability. White et al (14) reported that 14 lung cancers that were overlooked at CT were identified from about 37,500 chest CT scans at more than three institutions, and an endobronchial location was cited as the most common characteristic among these cases.
A more recent study of seven lung cancers missed at low-dose CT was reported on by Kakinuma et al (15); their data were obtained from 5,418 lung cancer CT screening studies performed for more than 3 years. To our knowledge, however, in no previous study have the characteristics of lung cancers missed at CT screening in a general population, including nonsmokers and women, been specifically assessed.
The purpose of our study was to compare the clinical, histopathologic, and imaging features of missed lung cancers in patients screened at low-dose CT.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
30 pack-years), and 4,251 nonsmokers (941 male, 3,310 female). All subjects gave informed consent. Approval for the review and research of the cases used in this study was obtained from the institutional review board of the University of Chicago.
Low-Dose CT Screening and Image Reading
The mobile unit equipped with a CT scanner was used to scan the chest; the following parameters were used: a tube current of 50 mA in 1996 and of 25 mA in 1997 and 1998, a scanning time of 2 seconds per rotation of the x-ray tube (tube rotation time: 2 seconds), a table speed of 10 mm/sec (pitch: 2), 10-mm collimation, and a 10-mm reconstruction interval.
Low-dose CT scans were displayed and interpreted on two monochrome, high-resolution (1,728 x 2,304 matrix) cathode-ray tubes by using three display conditions for adequate examination of the lungs (width: 1,000 HU; level: -700 HU), hilar bronchi (width: 1,500 HU; level: -550 HU), and mediastinum (width: 300 HU; level: 20 HU). The CT scans were displayed primarily in multiformat mode in sets of 12 or 16 images, but four- or single-section views were used as needed. Most scans were interpreted during the evenings, with and without light and without distracting noise. At the primary reading, the first reader (in some cases, F.L.) reviewed the CT scans without referring to previously obtained CT scans, and the second reader (S.S.) reviewed the images while referring to previously obtained CT scans. These primary reviewers were aware of the age and sex of the patients but not of their smoking history.
All CT scans were classified in one of seven categories: category A for images depicting unsatisfactory examinations, category B for images depicting normal results, category C for images depicting lung abnormalities of little clinical importance, category D for images depicting noncancerous lung lesions, category Ed for images depicting noncancerous but suspicious (ie, indeterminate) lung lesions, category E for images depicting lesions suspected to be lung cancer, and category F for images depicting small nonspecific lung nodules (<3 mm in diameter). One of five general radiologists (in some cases, F.L.) initially read each CT scan to identify and classify the suspicious abnormalities (categories Ed, E, F) and the noncancerous lesions (category D). A thoracic radiologist (S.S.), the supervisor of the project, then reviewed the CT scans of these suspicious lesions to make a final diagnosis. No second review of the images in categories A–C was performed. An initial radiology report described the features of the suspicious lesions depicted on images in the Ed, E, and F categories and indicated the CT section number, general position of the lesions, and final diagnoses of the lesions depicted on images in the C and D categories. No report was filed for images in the A and B categories. Follow-up diagnostic CT examinations were recommended for patients with suspicious (image categories Ed, E, and F) and noncancerous (image category D) lesions. No follow-up examinations were recommended for patients in whom A–C images were obtained.
Follow-up Diagnostic CT Examination
The follow-up CT examinations of patients with suspicious lesions (Ed, E, F images) were performed primarily at Shinshu University Hospital, Matsumoto, Nagano, Japan, and those of patients with noncancerous lesions (D images) were performed at local hospitals. Of the total of 17,892 low-dose CT screening examinations performed, 819 (4.6%) revealed suspicious lesions (Ed, E, F images). For 780 (95%) of these 819 cases, diagnostic work-up examinations were performed within 3 months after low-dose CT screening by using conventional chest radiography and diagnostic CT, including thin-section scanning, and at follow-up CT 3, 6, 12, 18, and 24 months after screening, as needed. The remaining 39 patients (5%) failed to schedule a follow-up diagnostic examination in the year the lesion was detected owing to reasons such as a busy farm work schedule or the long distance to the hospital, but most of these lesions were proven to be benign on sequential low-dose CT scans.
For follow-up, a helical CT scanner (CT HiSpeed Advantage; GE Medical Systems, Milwaukee, Wis) was used to scan the entire lung with the following standard parameters: 200-mA tube current, 1-second tube rotation, table speed of 10 mm/sec, and 10-mm collimation. In addition, thin-section scans that encompassed the entire lesion were obtained with 1-mm collimation and a bone reconstruction algorithm with a 0.5-mm interval.
The thoracic radiologist (S.S.) reviewed all of the diagnostic CT or thin-section CT scans obtained at Shinshu University Hospital and the indeterminate CT scans obtained at local hospitals. The characteristic findings of lung cancer at thin-section CT included spiculation, lobulation, ground-glass opacity (GGO), and heterogeneous attenuation, including bubble-like air, air bronchogram, cavitation, tumor necrosis, and the halo sign (16–21).
Data Analysis
Lung cancer was confirmed histopathologically in 83 cases: 79 from surgical resection and four from transbronchial biopsy. Thirty-two cases were confirmed to be benign at biopsy. Of the 83 confirmed lung cancers, 39 were suspected to be lung cancer at repeat low-dose CT screening examinations performed in the 2nd or 3rd year of the study.
During image interpretation, two radiologists (S.S., F.L.) reviewed the currently and previously obtained CT scans of the lung cancers and compared the findings with the designated categories and findings on the radiology reports, confirmed the detected and missed lung cancers, and classified the missed lung cancers into two subgroups: those missed owing to detection error and those missed owing to interpretation error. The two radiologists worked in consensus.
Parameters, such as the age, sex, and smoking status of the patients, and the histopathologic findings and American Joint Committee on Cancer stages of the detected and missed lung cancers, were analyzed. The sizes and locations of the detected and missed lung cancers, as well as the sizes of the cancers missed owing to detection and interpretation errors also were compared.
For the purposes of this study, three radiologists (H.M., F.L., H.A.) independently reviewed the low-dose CT scans of the missed and detected lung cancers. The lung cancers were subjectively classified as having one of three patterns: pure GGO, mixed GGO, and solid opacity. The subtlety of each lung cancer with regard to size, attenuation, and conspicuity was classified subjectively into one of three categories: not subtle, subtle, or very subtle. The other findings of the missed lesions in the detection error group also were judged. For example, the similarity in appearance of each missed cancer to normal structures was evaluated, and whether the lesion appeared to overlap with pulmonary vessels or be obscured by hilar structures was determined. The findings of 16 CT scans in the interpretation error group were similar to those of benign lesions, and other confusing findings, such as tuberculosis or other inflammatory lesions, were assessed. The final judgment regarding CT findings was based on agreement among two or more radiologists.
Statistical analysis was performed by using the Student t test to compare differences in patient age and lesion size between the missed and detected cancer cases and between the detection error and interpretation error cases. The 
2 test for independence and the Fisher exact test were performed independently to compare differences in the sex and smoking status of the patients and differences in the location, histopathologic findings, cancer stage, radiologic pattern, and subtlety of the lesions in these groups (ie, missed and detected cancer groups and detection and interpretation error groups). Statistical significance was defined as a P value of less than .05.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Of the 39 lung cancers assigned to image categories E, Ed, or F at first or second repeat CT, 11 either could not be identified or were barely discernible owing to their small (<2-mm) size on the screening images obtained the previous year, and 28 were visible retrospectively in the corresponding location. Thirty-two lung cancers in 32 patients, including the 28 cancers just described and the four that were misclassified as probably benign (D images) at final CT, were considered to be lung cancers that were missed during the 3-year lung cancer CT screening project.
Of the 76 detected lung cancers in 76 patients, four were central (endobronchial tumors in or proximal to a segmental bronchus) and 72 were intrapulmonary (distal to segmental bronchi). All of the missed lung cancers in this study were intrapulmonary tumors. Surgery was performed in 71 of the patients with detected intrapulmonary cancers (one patient refused to undergo surgery) and in one of the patients with central cancers (three cancers were deemed noncurable with surgery). Therefore, in the 76 patients with detected lung cancer, 72 lesions were verified at surgical staging (ie, histopathologic analysis) and four were identified at radiologic staging.
The 32 missed lung cancers were confirmed surgically in 31 cases. In the one patient who refused to undergo surgery, the diagnosis was based on transbronchial biopsy findings, whereas the stage of this malignancy was based on radiologic findings.
The clinical and lesion characteristics of the 32 patients with missed lung cancers, as compared with those of the 76 patients with detected lung cancers, are shown in Table 1. No significant difference in the mean age or smoking status of patients or in tumor location was found between the missed and detected lung cancer cases.
|
|
Detection and Interpretation Errors
Of the 32 missed cancer cases, 20 (20 patients, 23 scans) were classified as detection errors with previously obtained low-dose CT scans because the cases were not mentioned in the radiologists’ report, and 13 (13 patients, 16 scans: four obtained during the present study and 12 obtained previously) were classified as interpretation errors (ie, reported but misinterpreted). For one patient, there was a detection error with the first scan and an interpretation error with the second scan. For three patients, there were detection errors with two scans. For one patient, there were interpretation errors with two scans. For one patient, there were interpretation errors with three scans.
All lesions with which there were detection errors were also missed on the previously obtained CT scans. Among the 23 detection error cases, the numbers of cases involving B, C, and either D or Ed images were 15, six, and two, respectively. The lesions depicted on CT scans in categories C and either D or Ed in the detection error group were judged to be lesions other than lung cancer. Among the interpretation error cases, four errors (on D images) were missed on currently obtained CT scans and 12 errors were missed on previously obtained CT scans (D images in three cases and C images in nine cases). The erroneous diagnosis in the 16 interpretation error cases included tuberculosis in eight cases (two of which consisted of pleural thickening attributed to residual tuberculosis) and inflammatory lesions in another eight cases. Diffuse lung diseases, including fibrosis in two patients, emphysema in one patient, fibrosis and emphysema in one patient, silicosis in one patient, and asbestosis in one patient, were included in the interpretation error group.
The clinical findings in 20 patients with whom there were detection errors, as compared with those in 13 patients with whom there were interpretation errors, are shown in Table 3. The mean age of patients in the detection error group (61.7 years) was younger than that of patients in the interpretation error group (71.2 years) (P = .001). Although we observed no significant difference between smokers and nonsmokers in either subgroup, there were more women (11/20 [55%]), all of whom were nonsmokers, in the detection error group than in the interpretation error group (3/13 [23%]) (P = .087). No significant difference in lesion location between the detection error and interpretation error groups was noted.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
In most previous studies (2,4,6,10,11,15) of lung cancer screening programs, the subjects had similar backgrounds—that is, they were considered to be at increased risk of developing lung cancer: They were heavy smokers, of advanced age, predominantly men, and from metropolitan populations in the United States and Japan. The subjects in our CT lung cancer screening program, however, were from a general population in Nagano, a rural area of Japan (5,7,8), and approximately half of them were nonsmokers and women. Most of these individuals were farmers who had lived most of their life in mountainous areas with little pollution. Austin et al (12) and Gurney (13) reported on 27 lung cancers that were missed at radiography and nine lung cancers that were overlooked at CT, respectively. The investigators in both studies identified more missed lung cancers in women (67% of lung cancers in Austin et al study and 78% of lung cancers in Gurney study), most of whom were heavy smokers. Adenocarcinoma accounted for 30% and 56% of the missed cancers, respectively, in the two studies.
In the present study, the overall prevalence of cancer in the nonsmoking group (45 [1%] of 4,251 subjects) was comparable to that in the smoking group (38 [1%] of 3,596 subjects); similar findings have been reported in previous studies (5,8). In the detection error group, however, lesions were frequently missed in the nonsmoking women (n = 11, 55%), and nearly all of these lesions were well-differentiated adenocarcinomas. These findings are different from those in previous studies (12–15), and the reasons for these differences are unclear but probably related to the specific population group in our CT screening project.
Our study results demonstrate the correlation between the histopathologic findings and the radiologic features of missed lung cancers. The well-differentiated adenocarcinomas commonly contained pure or mixed, low-attenuating lesions because of the replacement growth of the tumor, which manifests as areas of GGO on thin-section CT scans (20–25). Note that small areas of GGO do not always correlate with the histopathologic findings because of the partial volume effect on thick-section CT scans. To our knowledge, there is no appropriate terminology for low-attenuating lesions on thick-section low-dose CT scans; therefore, we referred to these lesions as areas of GGO in this study, and most of these lesions were subsequently identified as areas of GGO at diagnostic thin-section CT.
Kundel et al (26) classified lesion detection errors as those caused by scanning (ie, search) error, recognition error, or decision-making error. It was difficult in this study, however, to clearly determine the cause for each missed cancer. Kundel and Revesz (27) defined conspicuity as a measure of the visibility of a lesion on radiographic images. This conspicuity is defined as the ratio of lesion signal intensity to background structure noise; accordingly, low conspicuity may lead to reader error. In the detection error group in our study, faint lesions with GGO (corresponding to low signal intensity) accounted for 91% of the missed lesions, and high structured noise due to anatomic structures surrounding the lesion was associated with 83% of the missed lesions. Therefore, these small areas of GGO were considered to be subtle or very subtle owing to poor conspicuity. We believe that poor conspicuity probably was the main cause of the detection errors. However, a recognition or decision-making error might have occurred when a solid nodule, which typically is easy to detect, was missed, probably because of its similarity to a pulmonary vessel or benign lesion.
In addition, about 35% of the CT scans depicting missed cancers in the detection error group were reported as being missed on category C, D, and Ed low-dose CT scans because of the depiction of other lesions: a non–clinically important lung abnormality in six cases and a noncancerous suspicious lung lesion in two cases. Most of these abnormalities were relatively obvious compared with the missed cancers. Thus, some detection errors might be the result of satisfaction-of-search errors (28), and decision-making errors might be caused by other abnormalities. Therefore, we believe that the missed lesions in the detection error group were probably due to various types of error, including scanning, recognition, decision-making, and satisfaction-of-search errors, or to multiple combined factors.
In the interpretation error group, about half of the lesions were identified as having a complex background, such as pulmonary emphysema and focal or diffuse fibrosis, which may have been due to the advanced age and/or smoking history of the patients. For such cancers, uncommon findings, such as atypical tumor shape and tumor attenuation, might cause diagnostic errors at chest radiography or CT (13–15,29). In our interpretation error cases, 88% of the missed cancers, or the features of these tumors, mimicked benign lesions and/or were associated with underlying lung disease. Missed cancers with linear, triangular, and irregular patterns, similar to the patterns of benign lesions, were common findings, and the underlying lung diseases were due to other abnormalities, such as residual tuberculosis (including pleural thickening) or residual or new inflammatory lesions, emphysema, or lung fibrosis. The missed lesions in the interpretation error group were identified at review of the initial radiology reports, which described the main CT section number on which the lesions were depicted and the general position of lesions. Most of these missed cancers were solitary lesions on one section or continuous sections; thus, we attributed these failed identifications to decision-making error. In addition, satisfaction-of-search error might have contributed to the missed cancers in this group, because a few cases involved missed cancers together with relatively obvious noncancerous lesions on the same section and in the same hemithorax.
The mean diameter of the cancers missed in our study was substantially greater (9.8 and 15.9 mm in the detection error and interpretation error groups, respectively) than that in a previous report of seven cases (8 mm) by Kakinuma et al (15). The findings of our study help to explain this difference: The cancers missed owing to detection error generally appeared as faint nodules with GGO and to overlap with normal structures, whereas the cancers missed owing to interpretation error frequently occurred in a complex background caused by other diseases.
There was essentially no difference in low-dose CT image quality between the scans that depicted missed lesions and those that depicted detected lesions. Technical factors, such as low-dose x-ray exposure (50 or 25 mA), long exposure time (31-second breath hold), high pitch (2:1), and slight high-frequency enhancement, might have caused the increased background noise at CT imaging in the larger patients. Motion artifact also was present on some scans, especially those obtained in older individuals with chronic obstructive pulmonary disease; however, we do not believe that this contributed to cancers being missed. We recognize that the use of a relatively thick collimation (10 mm) may have impaired the detection of faint nodules with GGO in our study and that the use of thin collimation (5–8 mm) may be better for CT screening in the future. Such thin collimation might reduce the number of false-negative cases at CT screening for lung cancer.
Among all the lung cancers (in 83 patients) identified in the 3-year CT screening program, five (6%) were centrally located endobronchial lesions in or proximal to segmental bronchi: four squamous cell carcinomas and one adenocarcinoma. During this study, we followed up all screened subjects to determine if any had failed to return for screening because of the development of symptoms of cancer, and no such patients were found. The very low percentage of central endobronchial lesions in our study was probably related to the large fraction of nonsmokers (4,251 [54%] of 7,847 individuals) in the general population. Additionally, "overdiagnosis" (30) probably occurred among the intrapulmonary lung cancers. We studied the volume doubling time for these lung cancers and found a large number of slowly growing adenocarcinomas: Those with pure GGO had a mean volume doubling time of 813 days, and those with mixed GGO had a mean volume doubling time of 457 days (31). Thus, these cancers might have remained stage I tumors for a long time without causing clinical symptoms of lung cancer.
Muhm et al (10) reported that 36 central cancers (within 4 cm of the hilum) (39%) among 92 missed lung cancers grew rapidly, usually appearing as hilar or mediastinal enlargement after normal findings at radiography performed 4 months earlier. White et al (14) detected 10 central lung cancers (endobronchial lesions within 3 cm of the hilum) (67%) among 15 cancers missed at CT. A possible reason for these missed lesions is that radiologists tend to focus their visual search on structures other than the central airways. In the White et al (14) study, the prognosis for patients with central endobronchial cancers (mean diameter of missed lesions, 13 mm) that were missed at routine CT was very poor; most of these cancers were not curable at surgery performed several months later. All five of the central endobronchial cancers in our study were identified in the initial screening year: The lesions in four patients were detected at low-dose CT and were more advanced than stage III, whereas the cancer in one patient was detected at the sputum examination in the screening program and not at CT. There were no central cancers among the missed lung cancers, probably because the rapid growth of central cancer makes detection difficult at CT examination in the early stage.
However, we found four intrapulmonary adenocarcinomas, all of which were missed owing to detection error, in the perihilar region (within 2 cm of the hilum). Two of these cases involved subtle lesions with GGO at initial CT. Although these lesions were still small when they were detected at subsequent CT, the histopathologic findings indicated that they were of higher cancer stages. If these cancers had been found at a later period, it would have been difficult to distinguish the intrapulmonary cancers from the central lung cancers (10). Because of the large pulmonary vessels in the hilar region, anatomic structured noise is greater in this area than elsewhere in the lung.
The percentage of missed lung cancers (32/83 [39%]) in a general population in this study, including 19 (42%) of 45 lesions in the nonsmokers and 13 (34%) of 38 lesions in the smokers, was slightly higher than that in a previous study of missed lung cancers in a high-risk population of heavy smokers (seven [32%] of 22 cancers) (15). Most of the missed lung cancers in this study were judged to be stage IA tumors in the 2nd or 3rd years of the study. However, among these early-stage cancers, some aggressive (ie, fast-growing and fast-progressing) tumors, such as moderately or poorly differentiated adenocarcinoma, squamous cell carcinoma, and small cell carcinoma, were detected at repeat low-dose CT screening and had a mean volume doubling time of 150 days or less (31,32). These rapidly growing missed cancers have the potential to affect patient prognoses.
When an intrapulmonary tumor is near or overlapped with hilar structures, it tends to progress to an advanced stage in a short time, even for a well-differentiated adenocarcinoma. We believe that it is important to identify such aggressive cancers at an early stage by means of CT screening to potentially reduce the mortality of lung cancer. Although there were five missed cancer cases that represented moderately differentiated adenocarcinoma—three in nonsmokers and two in women—the more aggressive poorly differentiated adenocarcinoma in one case, squamous cell carcinoma in one case, and small cell carcinoma in two cases were all in men, three of whom were heavy smokers and one of whom was a nonsmoker with silicosis. In summary, radiologists interpreting CT scans at lung cancer screening must be especially attentive to identify possible lesions in the central region, including endobronchial lesions and lesions overlapping with hilar structures. Care must also be exercised when images also depict underlying lung disease, which is especially likely to exist in individuals who are at high risk of developing lung cancer because of their heavy smoking or occupational history.
Our study results suggest that there is a potential for more accurate lung cancer detection during assessment of the characteristics of malignant lesions that may be overlooked or misinterpreted by human observers. Objective and consistent information, however, may be generated by an automated computerized system that integrates image processing and computer visualization techniques for recognition of lung nodules (ie, potential lung cancers) on CT scans. Although the interpretations made by human observers are limited to those of the qualitative information conveyed on images, a computerized technique extracts the quantitative details on an image and applies mathematical and statistical algorithms to a detection task. Such a computer-aided diagnostic method can efficiently analyze the large amount of image data acquired during CT scanning and indicate to the radiologist the locations of suspicious regions (33–35).
A detailed understanding of the clinical, radiologic, and histopathologic characteristics of missed cancers and the incorporation of computerized detection results into the medical decision-making process have the potential to reduce the number of lung cancers missed by radiologists on CT scans.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Abbreviation: GGO = ground-glass opacity
Author contributions: Guarantors of integrity of entire study, F.L., S.S.; study concepts and design, F.L., S.S., K.D.; literature research, K.D.; clinical studies, F.L., S.S., H.A., H.M.; experimental studies, S.G.A., K.D.; data acquisition, F.L.; data analysis/interpretation, F.L., S.S., H.A., H.M.; statistical analysis, F.L.; manuscript preparation, definition of intellectual content, and editing, F.L., K.D.; manuscript revision/review, all authors; manuscript final version approval, F.L., K.D.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  G. M. Israel, S. Herlihy, A. N. Rubinowitz, D. Cornfeld, and J. Brink Does a Combination of Dose Modulation with Fast Gantry Rotation Time Limit CT Image Quality? Am. J. Roentgenol., July 1, 2008; 191(1): 140 - 144. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. M. Wahidi, J. A. Govert, R. K. Goudar, M. K. Gould, and D. C. McCrory Evidence for the Treatment of Patients With Pulmonary Nodules: When Is It Lung Cancer?: ACCP Evidence-Based Clinical Practice Guidelines (2nd Edition) Chest, September 1, 2007; 132(3_suppl): 94S - 107S. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. M. Park, J. M. Goo, H. J. Lee, C. H. Lee, E. J. Chun, and J.-G. Im Nodular Ground-Glass Opacity at Thin-Section CT: Histologic Correlation and Evaluation of Change at Follow-up RadioGraphics, March 1, 2007; 27(2): 391 - 408. [Abstract] |
