CT colonography: accuracy of initial interpretation by radiographers in routine clinical practice
ABSTRACT To investigate performance of computed-assisted detection (CAD)-assisted radiographers interpreting computed tomography colonography (CTC) in routine practice.
Three hundred and three consecutive symptomatic patients underwent CTC. Examinations were double-read by trained radiographers using primary two-dimensional/three-dimensional (2D/3D) analysis supplemented by "second reader" CAD. Radiographers recorded colonic neoplasia, interpretation times, and patient management strategy code (S0, inadequate; S1, normal; S2, 6-9 mm polyp; S3, > or = 10 mm polyp; S4, cancer; S5, diverticular stricture) for each examination. Strategies were compared to the reference standard using kappa statistic, interpretation times using paired t-test, learning curves using logistic regression and Pearson's correlation coefficient.
Of 303 examinations, 69 (23%) were abnormal. CAD-assisted radiographers detected 17/17 (100%) cancers, 21/28 (72%) polyps > or = 10 mm and 42/60 (70%) 6-9 mm polyps. The overall agreement between radiographers and the reference management strategy was good (kappa 0.72; CI: 0.65, 0.78) with agreement for S1 strategy in 189/211 (90%) exams; S2 in 19/27 (70%); S3 in 12/19 (63%); S4 in 17/17 (100%); S5 in 5/6 (83%). The mean interpretation time was 17 min (SD = 11) compared with 8 min (SD = 3.5) for radiologists. There was no learning curve for recording correct strategies (OR 0.88; p = 0.12) but a significant reduction in interpretation times, mean 14 and 31 min (last/first 50 exams; -0.46; p < 0.001).
Routine CTC interpretation by radiographers is effective for initial triage of patients with cancer, but independent reporting is currently not recommended.
- SourceAvailable from: Carsten Lauridsen
[Show abstract] [Hide abstract]
- "This result concurs with another study by Burling et al. , which showed an agreement between the reference standard (consensus between expert radiological review, colonoscopy data, and clinical follow-up) and computer aided detection (CAD)-assisted radiographers demonstrating the kappa value at 0.72 (95 % CI 0.65–0.78). "
ABSTRACT: OBJECTIVE: To compare computed tomographic colonography (CTC) performance of four trained radiographers with the CTC performance of two experienced radiologists. METHODS: Four radiographers and two radiologists interpreted 87 cases with 40 polyps ≥6 mm. Sensitivity, specificity, and positive predictive value (PPV) were assessed on a per-patient basis. On a per-polyp basis, sensitivity was calculated according to the respective size categories (polyps ≥6 mm as well as polyps ≥10 mm). RESULTS: Overall per-patient sensitivity for polyps ≥6 mm was 76.2 % (95 % CI 61.4-91.0) and 76.2 % (95 % CI 61.7-90.6), for the radiographers and radiologists, respectively. Overall per-patient specificity for polyps ≥6 mm were 81.4 % (95 % CI 73.7-89.2) and 81.1 % (95 % CI 73.8-88.3) for the radiographers and the radiologists, respectively. For the radiographers, overall per-polyp sensitivity was 60.3 % (95 % CI 50.3-70.3) and 60.7 % (95 % CI 42.2-79.2) for polyps ≥6 mm and ≥10 mm, respectively. For the radiologists, overall per polyp sensitivity was 59.2 % (95 % CI 46.4-72.0) and 69.0 % (95 % CI 48.1-89.6) for polyps ≥6 mm and ≥10 mm, respectively. CONCLUSION: Radiographers with training in CT colonographic evaluation achieved sensitivity and specificity in polyp detection comparable with that of experienced radiologists. MAIN MESSAGES : • The diagnostic accuracy of trained radiographers was comparable to that of experienced radiologists. • The use of radiographers in reading CTC examinations is acceptable, however radiologists would still be necessary for the evaluation of extracolonic findings. • Skilled non-radiologists may play a vital role as a second reader of intraluminal findings or by performing quality control of examinations before patient dismissal.Insights into Imaging 06/2013; 4(4). DOI:10.1007/s13244-013-0260-x
- [Show abstract] [Hide abstract]
ABSTRACT: In the absence of official guidance for the management of colonic wall thickening identified by computed tomography (CT), a common clinical dilemma surrounds the volume of colonoscopies subsequently performed. To identify whether colonic wall thickening identified at CT consistently warrants colonoscopy, consecutive colonoscopies performed at Leeds Teaching Hospitals Trust in 2008 and recorded as "possible colonic lesion on cross-sectional abdominal CT" in an endoscopic database were retrospectively analyzed. Clinical, radiologic, colonoscopic, and histologic data were obtained from medical records. Of 4,702 colonoscopies, 94 (2%) had a full data set meeting the inclusion criteria. The primary diagnoses were normal condition (n = 11, 11.7%), adenocarcinoma (n = 25, 26.6%), adenoma (n = 23, 24.5%), diverticular disease (n = 12, 12.8%), nonspecific colitis (n = 6, 6.4%), Crohn's disease (n = 4, 4.3%), and hyperplastic polyp (n = 3, 3.2%). Computed tomography and colonoscopy were concordant for specific pathology in 79.8% of the cases (n = 75). Compared with diagnosis after histology, colonoscopy alone correctly identified specific pathology in 18.1% of the cases (n = 17), and CT alone was correct in 4.3% of the cases (n = 4)), whereas both were incorrect in 3.2% of the cases (n = 3). Computed tomography had a sensitivity of 72.3% (95% confidence interval [95% CI], 61.9-80.8%), a specificity of 96.5% (95% CI, 94.9-97.6%), a positive predictive value of 72.3%, and a negative predictive value of 96.5%. In 63.8% of the cases (n = 60), CT identified pathology necessitating further intervention at the time of colonoscopy or afterward, and in 28.7% of the cases (n = 27), CT identified pathology requiring no additional intervention. In the remaining 7.4% of the cases (n = 7), CT detected no new pathology. Computed tomography is highly predictive of colonic pathology compared with final outcome after colonoscopy and biopsy. For patients without a pre-existing diagnosis, colonic wall thickening demonstrated at CT warrants further investigation with colonoscopy.Surgical Endoscopy 02/2011; 25(8):2586-91. DOI:10.1007/s00464-011-1591-7 · 3.31 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Colonic polyps are frequently encountered in clinics. Computed tomographic colonography (CTC), as a painless and quick detection, has high values in clinics. In this study, we evaluated the application value of computer-aided detection (CAD) in CTC detection of colonic polyps in the Chinese population. CTC was performed with a GE 64-row multidetector computed tomography (MDCT) scanner. Data of 50 CTC patients (39 patients positive for at least one polyp of ≥ 0.5 cm in size and the other 11 patients negative by endoscopic detection) were retrospectively reviewed first without computer-aided detection (CAD) and then with CAD by four radiologists (two were experienced and another two inexperienced) blinded to colonoscopy findings. The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of detected colonic polyps, as well as the areas under the ROC curves (Az value) with and without CAD were calculated. CAD increased the overall sensitivity, specificity, positive predictive value, negative predictive value and accuracy of the colonic polyps detected by experienced and inexperienced readers. The sensitivity in detecting small polyps (5 - 9 mm) with CAD in experienced and inexperienced readers increased from 82% and 44% to 93% and 82%, respectively (P > 0.05 and P < 0.001). With the use of CAD, the overall false positive rate and false negative rate for the detection of polyps by experienced and inexperienced readers decreased in different degrees. Among 13 sessile polyps not detected by CAD, two were ≥ 1.0 cm, eleven were 5 - 9 mm in diameter, and nine were flat-shaped lesions. The application of CAD in combination with CTC can increase the ability to detect colonic polyps, particularly for inexperienced readers. However, CAD is of limited value for the detection of flat polyps.Chinese medical journal 02/2011; 124(3):380-4. · 1.02 Impact Factor