摘要
Background: The “Suspected Blood Indicator”(SBI), a feature included in the software for interpretation of capsule endoscopy(CE), is designed to facilitate detection of bleeding lesions in the small bowel. This study evaluated the sensi tivity and the specificity of the SBI in patients who underwent CE for obscure G I bleeding (OGIB). Methods: CE recordings from patients with OGIB recruited in 7 centers were read by experts blinded to the red SBI tags. They classified lesio ns of interest as bleeding or as having a potential for bleeding that was high ( P2), low (P1), or absent (P0). The SBI tags then were marked by a another invest igator.Concordance was acknowledged when frames selected by the expert reader, a nd those tagged by the SBI had the same time code. Results: A total of 156 recor dings were evaluated.In 83, there was either. no lesion (n = 71) or a P0 lesion (n= 12); these CE recordings were regarded as normal. Among the 73 abnormal reco rdings, 114 P2 and 92 P1 lesions were identified. A total of 154 red tags were a nalyzed. Sensitivity,specificity, and positive and negative predictive values of SBI were 37%, 59%, 50%, and 46%, respectively, for detection of the presenc e of a P2 or P1 lesion in front of a red tag.Conclu- sions: SBI-based detection of intestinal lesions with the potential for bleed ing is of limited clinical value in practice and does not reduce the time requir ed for interpretation of CE.
Background: The “Suspected Blood Indicator”(SBI), a feature included in the software for interpretation of capsule endoscopy(CE), is designed to facilitate detection of bleeding lesions in the small bowel. This study evaluated the sensi tivity and the specificity of the SBI in patients who underwent CE for obscure G I bleeding (OGIB). Methods: CE recordings from patients with OGIB recruited in 7 centers were read by experts blinded to the red SBI tags. They classified lesio ns of interest as bleeding or as having a potential for bleeding that was high ( P2), low (P1), or absent (P0). The SBI tags then were marked by a another invest igator.Concordance was acknowledged when frames selected by the expert reader, a nd those tagged by the SBI had the same time code. Results: A total of 156 recor dings were evaluated.In 83, there was either. no lesion (n = 71) or a P0 lesion (n= 12); these CE recordings were regarded as normal. Among the 73 abnormal reco rdings, 114 P2 and 92 P1 lesions were identified. A total of 154 red tags were a nalyzed. Sensitivity,specificity, and positive and negative predictive values of SBI were 37%, 59%, 50%, and 46%, respectively, for detection of the presenc e of a P2 or P1 lesion in front of a red tag.Conclu- sions: SBI-based detection of intestinal lesions with the potential for bleed ing is of limited clinical value in practice and does not reduce the time requir ed for interpretation of CE.
