capsule endoscopy: comparison of two different reading modes
TRANSCRIPT
ORIGINAL ARTICLE
Capsule endoscopy: comparison of two differentreading modes
Ute Günther & Severin Daum & Martin Zeitz &
Christian Bojarski
Accepted: 20 October 2011 /Published online: 9 November 2011# Springer-Verlag 2011
AbstractPurpose Capsule endoscopy (CE) is a very useful tool forthe evaluation of the small intestine, but it is timeconsuming. The aim of this study was to compareevaluation times and detection rates in two different readingmodes (single view at a speed of 10 frames per second (fps)and four images simultaneously, i.e., quadview mode at aspeed of 20 fps) to find the optimum setting mode forevaluation of CE videos.Methods CE videos of 70 patients performed for differentindications (obscure bleeding, n=50; suspected Crohn’sdisease, n=10; and suspected or complicated celiac disease,n=10) were reviewed by investigators A and B in the twodifferent reading modes.Results The mean evaluation time using single view at 10 fpswas 22 min (SD±9.1 min) and 11.9 min (SD±4.8 min) usingquadview mode at 20 fps. The detection rates ofangiodysplasias, erosions, small ulcers, and small polypswere only discreetly lower using the quadview mode at20 fps. In Crohn’s disease and celiac disease, theessential aspects of inflamed or atrophic mucosa segmentswere equally detected in both reading modes. In one case ofcomplicated celiac disease with severe erosive jejunitis, alymphoma-suspect lesion was overlooked in the quadviewmode at 20 fps.Conclusions It is often possible to read CE videos inquadview mode at a higher speed with even so a highdiagnostic yield in a shortened evaluation time.
Keywords Small intestine . Capsule endoscopy . Readingmode . Detection rate . Evaluation time
Introduction
Capsule endoscopy (CE) is a novel method for non-invasive evaluation of the small intestine. CE has beenproven to be a very useful diagnostic tool in the evaluationof obscure gastrointestinal bleeding and iron deficiencyanemia. Further indications for CE are: suspected orestablished Crohn’s disease, suspected or complicatedceliac disease, suspected small-bowel tumors, andsurveillance of patients with known polyposis syndromes[1–7]. CE as a method offers a lot of advantages. CE iseasily applicable. The patient just has to swallow the capsuleafter small-bowel preparation, and the procedure can even beperformed on an outpatient basis. Additionally, CE isconsidered to be an extraordinarily safe procedure.Over 340,000 CEs have been performed worldwidewith no reported deaths and very few side effects orcomplications [8–10].
As few drawbacks of the method, the costs and theexpenditure of time for the evaluation have to bementioned. CE is expensive—the cost of one capsule is atabout €600 in Europe. During one CE, a total number of40,000 to 60,000 images are recorded. The averagephysician time required for the evaluation is reported torange from 40 to 60 min, depending on the recording timeand the experience of the examiner [11, 12].
Over the last years, additional support systems have beenadded to the CE software to facilitate the evaluation andreduce the reading times [13]: a suspected blood indicator, aquickview mode, and a multi-viewing mode. The quick-view mode generates a preview of the entire CE recording
U. Günther (*) : S. Daum :M. Zeitz : C. BojarskiMedical Clinic I Gastroenterology, Infectious Diseases,Rheumatology, Charité - Campus Benjamin Franklin,Hindenburgdamm 30,12200 Berlin, Germanye-mail: [email protected]
Int J Colorectal Dis (2012) 27:521–525DOI 10.1007/s00384-011-1347-9
selecting the “images of interest” according to the analysisof colors and image patterns. But its diagnostic miss ratewas described as 8% and 10% [14, 15] in two studies; thus,the quickview mode cannot substitute for a completereading of the CE video.
The multi-viewing mode which simultaneously displaystwo or four images for evaluation is now being increasinglyused. To our knowledge, this mode has only been tested in onestudy so far [15], where it was used displaying four imagessimultaneously at a very high speed of 35 frames per second(fps) and proved to have a high diagnostic miss rate as well.
The aim of this study was to compare the CE evaluationtimes and detection rates in two different reading modes(single view at a speed of 10 fps and four imagessimultaneously, i.e., quadview mode at a speed of 20 fps)to find the optimum setting mode for reading with highdiagnostic yield in a shortened evaluation time.
Patients and methods
Patients
CEs performed in our hospital between 2009 and 2010were retrospectively reviewed, and 45 cases with clinicallysignificant localized lesions were selected. CEs have beencarried out on these 45 patients for the followingindications (obscure bleeding, n=25; suspected or estab-lished Crohn’s disease, n=10; and suspected or complicatedceliac disease, n=10). Additionally, 25 CEs performed forobscure bleeding in 2011 were prospectively viewed. OnlyCEs with completed small-bowel transit were chosen.
The main clinical data are summarized in Table 1. InCrohn’s patients with a history or actual clinical symptomsof a suspected small-bowel stenosis, a patency CE or MRenteroclysis was performed prior to the ingestion of thecapsule. Patients with known polyposis syndromes (familialadenomatous polyposis, Peutz–Jeghers syndrome, etc.)were excluded from our study.
CE recording
The PillCam™ SB2 capsule, the RAPID workstation, andthe RAPID 6.0 software (Given Imaging Ltd., Yoqneam,Israel) were used for CE according to standard techniques,as described previously [3, 4]. The procedure wasperformed after an overnight fast. For bowel preparation,the patients received 2 l of polyethylene glycol (Klean-Prep®)solution 2–4 h before examination and 15 ml simeticon(Sab Simplex®) before capsule ingestion.
Patients were allowed to drink clear fluids 2 h aftercapsule ingestion and to ingest a light meal after 4 h. Thedata recorder was removed after 8 h of recording.
Table 1 Clinical data, small-bowel transit times, and evaluation timesin two different reading modes
Case Age/sex SBTT (min) Investigatorevaluationtime (min)
Investigatorevaluationtime (min)
Single view,10 fps
Quadview,20 fps
1 70 f 175 A 30.7 B 15.9
2 51 m 151 A 10.8 B 11.6
3 80 f 269 A 35.9 B 18.7
4 68 f 332 A 33.6 B 21.7
5 68 m 237 A 12.5 B 9.2
6 63 m 258 A 22.3 B 11.7
7 76 f 243 A 24.2 B 15.9
8 76 f 341 A 19 B 10.2
9 74 m 227 A 24 B 18.2
10 71 f 250 A 30.4 B 19.2
11 79 f 259 A 17 B 11.8
12 72 m 206 A 12.2 B 8.8
13 69 m 274 A 32.4 B 27.4
14 54 m 179 A 20.5 B 11.2
15 72 m 205 A 19.3 B 8.4
16 65 m 131 A 13.7 B 7.2
17 53 m 212 A 13.8 B 14.3
18 47 m 316 A 32.8 B 16.8
19 89 m 280 A 28.1 B 13.8
20 83 m 205 A 17.6 B 10.2
21 75 m 149 A 13.4 B 7.1
22 60 m 62 A 7 B 5.2
23 40 m 344 A 28.2 B 13.1
24 56 m 169 A 11 B 5.2
25 75 m 339 A 27.9 B 13.4
26 77 f 186 B 19.2 A 11.1
27 50 f 111 B 10.6 A 7.1
28 48 m 266 B 25.6 A 14.4
29 29 f 184 B 18.2 A 11.4
30 67 m 286 B 22.8 A 10.4
31 81 m 215 B 21.2 A 12.6
32 41 f 164 B 15.6 A 8.1
33 39 m 57 B 8.6 A 4.5
34 66 m 217 B 29.8 A 14.1
35 66 m 151 B 14.6 A 8.3
36 74 f 203 B 14.9 A 8.6
37 27 m 231 B 13.8 A 9.1
38 67 f 86 B 16.8 A 8.6
39 81 f 244 B 26.9 A 11.8
40 75 f 145 B 16.6 A 8.4
41 68 m 198 B 23.8 A 9
42 74 f 218 B 27.5 A 11.7
43 82 f 136 B 11.1 A 5
44 51 m 229 B 26.6 A 9.7
522 Int J Colorectal Dis (2012) 27:521–525
CE reading
All CE videos included in the study showed a completeexamination of the small bowel. Two experienced endo-scopists (U.G. and C.B.) who had read more than 250 CEsinterpreted all videos. U.G. and C.B. were blinded to thepatients’ identification and clinical history as well to thefindings of the colleague. The small-bowel transit time(time between the first and the last small-intestinal image)was recorded for each patient. The small-bowel passagewas then reviewed in two different reading modes (singleview at a speed of 10 fps and quadview mode at a speed of20 fps) independently by U.G. and C.B. The evaluation
time included the pure reading time and the time theinvestigators spent to mark suspected lesions as thumb-nailphotographs.
Investigator A reviewed videos 1–25, 51–55, and 61–65 in single view at a speed of 10 fps and videos 26–50, 56–60, and 66–70 in quadview mode at a speed of20 fps (Table 1). Investigator B reviewed the videos viceversa (1–25, 51–55, and 61–65 in quadview mode at aspeed of 20 fps and videos 26–50, 56–60, and 66–70 insingle view at a speed of 10 fps). Evaluation times and dataresults were recorded and compared (Tables 1, 2, 3).
Statistical analysis
Quantitative data were quoted as mean values (± standarddeviation, SD). Statistical analysis regarding the number ofdetected or missed lesions in the two different reading
Table 1 (continued)
Case Age/sex SBTT (min) Investigatorevaluationtime (min)
Investigatorevaluationtime (min)
Single view,10 fps
Quadview,20 fps
45 66 m 173 B 19.6 A 6.4
46 64 m 77 B 9.5 A 4.1
47 47 m 230 B 24.5 A 11.5
48 68 m 237 B 37.9 A 12.4
49 57 m 296 B 20.2 A 11.1
50 35 f 340 B 37 A 17.1
51 73 f 129 A 20.6 B 10.1
52 48 m 236 A 18 B 13.9
53 68 m 297 A 36.9 B 19.6
54 43 m 342 A 23.6 B 16
55 59 f 245 A 20.4 B 12.9
56 37 f 59 B 8.3 A 3.3
57 38 f 166 B 15.1 A 7
58 55 f 365 B 54.4 A 24.1
59 51 f 89 B 14.5 A 6.8
60 37 m 330 B 38 A 15.4
61 72 m 261 A 30.7 B 18.2
62 50 f 369 A 26.3 B 19.1
63 60 f 227 A 18.4 B 9.6
64 55 f 287 A 23.5 B 15.7
65 65 f 152 A 11.2 B 7.5
66 18 f 237 B 31.8 A 13.6
67 66 m 298 B 38.1 A 14.2
68 49 f 226 B 20.5 A 11.4
69 28 f 155 B 19 A 10.2
70 61 f 251 B 19.8 A 10.4
CEs were performed on cases 1–50 for the indication obscurebleeding, on cases 51–60 for the indication Crohn’s disease, and oncases 61–70 for suspected or complicated celiac disease
SBTT small-bowel transit time in minutes, fps frames per second, mmale, f female, A investigator A, B investigator B
Table 2 Detected lesions in two different reading modes
Single view,10 fps
Quadview,20 fps
(a) Obscure bleeding
Angiodysplasias 87 72 p<0.05
Erosions 22 13 p<0.05
Ulcerations 8 7
Fresh blood 9 9
Duodenal varices 2 1
Lymphangiectasias 16 15
Lipomas 55 54
Small polyps 6 7
Tumorous lesions 2 3
(b) Suspected or established Crohn’s disease
Angiodysplasias 5 6
Erosions 33 23
Ulcerations 2 1
Inflamed mucosa segments 2 2
Fresh blood 1 1
Lymphangiectasias 8 8
Lipomas 12 10
Small polyps 1 1
Pseudopolyps 3 1
(c) Suspected or complicated celiac disease
Angiodysplasias 4 3
Erosions 7 7
Atrophic mucosa segments 10 10
Tumorous lesions 1 0
Lymphangiectasias 1 1
Lipomas 2 2
Multiple lymph follicles 5 5
Reading modes: single view at 10 fps and quadview at 20 fps;p values determined by two-sided t test
Int J Colorectal Dis (2012) 27:521–525 523
modes was performed using the paired two-sided t test.A p value <0.05 was considered statistically significant.
Results
CEs have been performed on 70 patients for differentindications (obscure bleeding case 1–50, suspected or knownCrohn’s disease case 51–60, and suspected or complicatedceliac disease case 61–70). The CE videos have beenretrospectively (case 1–13, 39–70) or prospectively (case14–38) viewed by investigator A and investigator B in twodifferent reading modes (Table 1). The mean evaluation time(including pure reading time and the time the investigatorsspent to mark suspected lesions as thumb-nail photographs)using single view at 10 fps was 22 min (SD±9.1 min) and11.9 min (SD±4.8 min) using quadview mode at 20 fps—leaving the evaluation time in single view at 10 fps nearlytwice as long as the evaluation time in quadview at 20 fps.
Reviewing the CE videos of patients with obscurebleeding, numerous angiodysplasias, some erosions andulcers, one jejunal tumor as well as areas of fresh bloodwere detected in both setting modes (Table 2). Using singleview at 10 fps, 87 angiodysplasias and 22 erosions weredetected, whereas in quadviewmode only, 72 angiodysplasiasand 13 erosions could be captured (p<0.05). Using singleview at 10 fps, 8 angiodysplasias were missed, whereas inquadview at 20 fps, 18 angiodysplasias, 8 erosions, and 1ulcer have not been picked up by the investigators (Table 3).
In all of these cases, other angiodysplasias or inflammation-associated lesions have been discovered. The medicaljudgment has not been changed by the overlooked lesions,and the recommended access for a subsequent double-balloon enteroscopy was the same. In the quadview mode,duodenal varices were overlooked in one patient (Table 3),which had already been spotted during standard gastroscopy.Using single view at 10 fps, in one patient (case 49) one oftwo tumorous lesions was overlooked, which turned out tobe neuroendocrine tumors in the subsequent intraoperativeendoscopy and surgical resection.
Reviewing the CE videos of patients with suspected orestablished Crohn’s disease, inflamed mucosa segments andareas with fresh blood were detected in both setting modes(Table 2). Using single view at 10 fps, 33 erosions werepicked up, whereas in quadview, only 23 erosions havebeen found. Using single view at 10 fps, two erosions havenot been picked up; using quadview at 20 fps, 12 erosions,1 ulcer, and 2 pseudopolyps were missed (Table 3). In allthese cases, other erosions, ulcers, or pseudopolyps weredetected; thus, the overlooked lesions did not change themedical diagnoses.
Reviewing the CE videos of patients with suspected orcomplicated celiac disease, the essential aspect of atrophicmucosa segments was detected in both setting modes—aswell as erosions and areas of multiple lymph follicles.Additional lesions (as angiodysplasias, lymphangiectasias,lipomas) were captured in both reading modes (Tables 2, 3).Using quadview, however, in one patient (case 67) withcomplicated celiac disease and severe erosive jejunitis inthe adjacent jejunal segments, one tumorous lesion(suspected lymphoma) was missed which showed up inonly a single image (Tables 2, 3).
Discussion
Since CE is a time-consuming procedure, a multi-viewingmode simultaneously displaying four images is now beingincreasingly used for evaluation. Using quadview mode at20 fps halved the evaluation times compared to single viewat 10 fps. In our study, the detection rates of angiodys-plasias, erosions, small ulcers, and small polyps were onlydiscreetly lower using the quadview mode at 20 fps. Incases, where angiodysplasias or erosions were missed,similar lesions have been identified in the same jejunalsegments, and the consecutive medical conclusions orrecommended access for a subsequent double-balloonenteroscopy did not change.
In Crohn’s disease and celiac disease, the essentialaspects of inflamed or atrophic mucosa segments wereequally detected in both reading modes. Nevertheless, inone case of complicated celiac disease, a lymphoma-
Table 3 Missed lesions in two different reading modes
Single view, 10 fps Quadview, 20 fps
(a) Obscure bleeding
Angiodysplasias 8 18
Erosions 0 8
Ulcerations 0 1
Duodenal varices 0 1
Lymphangiectasias 2 3
Lipomas 11 14
Small polyps 1 0
Tumorous lesions 1 0
(b) Suspected or established Crohn’s disease
Angiodysplasias 1 0
Erosions 2 12
Ulcerations 0 1
Lipomas 1 3
Pseudopolyps 0 2
(c) Suspected or complicated celiac disease
Angiodysplasias 1 2
Tumorous lesions 0 1
Reading modes: single view at 10 fps and quadview at 20 fps
524 Int J Colorectal Dis (2012) 27:521–525
suspect lesion which showed up in only a single image wasoverlooked in the quadview mode at 20 fps. Due to thesevere erosive jejunitis in the adjacent jejunal segments,further diagnostic steps would have been initiated—evenwithout the detection of the lymphoma-suspect lesion.
Each physician should decide the reading settings based onhis own experience with CE. We would not recommend thequadview mode for trainees in CE. Trainees should beginusing the single view mode at 5 to 10 fps. However, forexperienced investigators, it is often possible to read CEvideos in quadview mode at a higher speed with even so ahigh diagnostic yield in a shortened evaluation time. Thequadview mode can be especially useful in CEs with longsmall-bowel recording times. In CEs with short small-bowelrecording times, a further reduction of the evaluation time isprobably not of such importance in the daily clinical routine.
Another approach to reduce the physician expenditure oftime is the use of trained endoscopy nurses for theinterpretation of CE videos. In previous trials, the meanreading time of the nurses was slightly longer than that ofthe physicians [15, 16]. However, the trained endoscopynurse detected 93–94% of the clinically significant lesionsseen by the physician [12, 16]. Thus, trained nurses couldpre-read the CE videos allowing the gastroenterologist onlyto review the marked abnormalities and to formulatemedical recommendations [17, 18].
CE as the only non-invasive diagnostic tool for detection ofsmall-bowel lesions has meanwhile become a standardprocedure in clinical gastroenterology. After a basic learningcurve in an experienced clinical setting, the quadview readingmode at 20 fps seems to be a good compromise for reading CEvideos at higher speedwith even so a high diagnostic yield in ashortened evaluation time. At least in our study, single missedfindings in the quadview mode at 20 fps did not alter thepatients’ clinical management. However, since CE is anexpensive method, which often represents the last investiga-tion in a row of endoscopic procedures, the extra time forsingle view might be justifiable.
References
1. Ladas SD, Triantafyllou K, Spada C, Riccioni ME, Rey JF, Niv Y,Delvaux M, de Franchis R, Costamagna G, the ESGE ClinicalGuidelines Committee, (2010) European Society of GastrointestinalEndoscopy (ESGE): recommendations (2009) on clinical use ofvideo capsule endoscopy to investigate small-bowel, esophageal andcolonic diseases. Endoscopy 42(3):220–227
2. Mishkin DS, Chuttani R, Croffie J, Disario J, Liu J, Shah R,Somogyi L, Tierney W, Song LM, Petersen BT (2006) ASGEtechnology status evaluation report: wireless capsule endosopy.Gastrointest Endosc 63(4):539–545
3. Ell C, Remke S, May A, Helou L, Henrich R, Mayer G (2002)The first prospective controlled trial comparing wireless capsuleendoscopy with push enteroscopy in chronic gastrointestinalbleeding. Endoscopy 34(9):685–689
4. Lewis BS, Swain P (2002) Capsule endoscopy in the evaluation ofpatients with suspected small intestinal bleeding: results of a pilotstudy. Gastrointest Endosc 56(3):349–353
5. Herrerias JM, Caunedo A, Rodriquez-Tellez M, Pellicer F,Herrerías JM Jr (2003) Capsule endoscopy in patients withsuspected Crohn’s disease and negative endoscopy. Endoscopy35(7):564–568
6. Rondonotti E, Pennazio M, Toth E, Menchen P, Riccioni ME, DePalma GD, Scotto F, De Looze D, Pachofsky T, Tacheci I,Havelund T, Couto G, Trifan A, Kofokotsios A, Cannizzaro R,Perez-Quadrado E, de Franchis R (2008) Small-bowel neoplasmsin patients undergoing video capsule endoscopy: a multicenterEuropean study. Endoscopy 40(6):488–495
7. Günther U, Bojarski C, Buhr HJ, Zeitz M, Heller F (2010)Capsule endoscopy in small-bowel surveillance of patients withhereditary polyposis syndromes. Int J Colorectal Dis 25(11):1377–1382
8. Pennazio M (2006) Capsule endoscopy: where are we after 6 yearsof clinical use? Review. Dig Liv Dis 38(12):867–878
9. Cave D, Legnani P, de Franchis R, Lewis BS (2005) ICCEconsensus for capsule retention. Endoscopy 37(10):1065–1067
10. Karagiannis S, Faiss S, Mavrogiannis C (2009) Capsule retention:a feared complication of wireless capsule endosopy. Scand JGastroenterol 44(10):1158–1165
11. Cave DR (2004) Reading wireless capsule endoscopy.Gastrointest Endosc Clin N Am 14(1):17–24
12. Levinthal GN, Burke CA, Santisi JM (2003) The accuracy ofan endoscopy nurse in interpreting capsule endoscopy. Am JGastroenterol 98(12):2669–2671
13. Spada C, Riccioni ME, Costamagna G (2007) Raid access real-time device and rapid access software: new tools in thearmamentarium of capsule endoscopy. Expert Rev Med Devices4(4):431–435
14. Westerhof J, Koornstra JJ, Weersma RK (2009) Can we reducecapsule endoscopy reading times? Gastrointest Endosc 69(3Pt1):497–502
15. Shiotani A, Honda K, Kawakami M, Murao T, Matsumoto H,Tarumi K, Kusunoki H, Hata J, Haruma K (2011) Evaluation ofRAPID® 5 Access software for examination of capsuleendoscopies and reading of the capsule by an endoscopynurse. J Gastroenterol 46(2):138–142
16. Riphaus A, Richter S, Vonderach M, Wehrmann T (2009) Capsuleendoscopy interpretation by an endoscopy nurse—a comparativetrial. Z Gastroenterol 47(3):273–276
17. Niv Y, Niv G (2005) Capsule endoscopy examination—preliminary review by a nurse. Dig Dis Sci 50(11):2121–2124
18. Bossa F, Cocomazzi G, Valvano MR, Andriulli A, Annese V(2006) Detection of abnormal lesions recorded by capsuleendoscopy. A prospective study comparing endoscopist’s andnurse’s accuracy. Dig Liver Dis 38(8):599–602
Int J Colorectal Dis (2012) 27:521–525 525