RSS-Feed abonnieren
DOI: 10.1055/s-0040-1704621
THE NEW GENERATION OF EXPRESS VIEW IS HIGHLY ACCURATE AND IS EFFECTIVE TO REDUCE CAPSULE ENDOSCOPY READING TIME
Publikationsverlauf
Publikationsdatum:
23. April 2020 (online)
Aims Capsule Endoscopy (CE) is a first-line diagnostic test in patients with suspected small bowel bleeding (SSBB). Since CE reading is time-consuming, innovative types of reading software have been developed. The new generation reading algorithm ‘Express-View’(EV) by MiroCam aims to reduce CE reading time by removing images based upon their similarity, without affecting the diagnostic yield.
Primary aim: evaluation of EV accuracy in patients with SSBB.
Secondary aim: evaluation of the reading time with the EV compared to Standard Reading (SR).
Methods This interim analysis included 48 pts with SSBB prospectively enrolled to undergo small bowel-CE. SR was initially performed according to the ESGE guidelines.
A second reading was performed using the EV mode by an external,blinded reader. Visualization time,nature,and relevance of lesions were collected.
EV accuracy was calculated before and after a consensus performed by two experienced investigators to re-classify dicordant cases. Reading time was also evaluated.
Results 47/48 patients performed a complete VCE examination. Cecal visualization was achieved in 43 pts(91.5%).
In 33/47 pts (70.2%) both SR and EV were in agreement for type of lesion and localization, while in the remaining 14 pts (29.8%) SR and EV had different results. After consensus, in 12/14 cases EV was reclassified to properly detect the lesion, being the prevoius disagreement considered related to a reader-misinterpretation. In the remaining 2pts (4.2%), EV mode missed relevant lesions.
Pre-consensus and post-consensus accuracy parameters are shown in the table.
(values in %) |
Sensitivity |
Specificity |
PPV |
NPV |
Pre-consensus |
71.9 |
66.7 |
82.1 |
52.6 |
Post-consensus |
94.6 |
100 |
100 |
83.3 |
Mean reading time at SR and EV was 84.13 ± 53.63 min and 14.79 ± 11.2 min, respectively (p< 0.001).
Conclusions Preliminary results suggest that EV algorithm shows excellent accuracy and significantly reduces CE reading time.
#