Pediatr Gastroenterol Hepatol Nutr.
2024 May;27(3):168-175. 10.5223/pghn.2024.27.3.168.
A Multicenter Pilot Study of Biliary Atresia Screening Using Digital Stool Color Imaging
- Waitayagitgumjon K
1 - Poocharoen W2,3
- Trirongjitmoah S4
- Treeprapin K5
- Suttiwongsing A6
- Wirifai T7
- Trirongchitmoh C8
- Tangkabuanbutr P9
- Affiliations
-
- 1Department of Surgery, Bhumibol Adulyadej Hospital, Royal Thai Air Force, Bangkok, Thailand
- 2Department of Surgery, Queen Sirikit National Institute of Child Health, Ratchathewi, Bangkok, Thailand
- 3Department of Surgery, College of Medicine, Rangsit University, Bangkok, Thailand
- 4Department of Electrical and Electronics Engineering, Faculty of Engineering, Ubon Ratchathani University, Ubon Ratchathani, Thailand
- 5Department of Mathematics, Statistics and Computer, Faculty of Science, Ubon Ratchathani University, Ubon Ratchathani, Thailand
- 6Department of Surgery, Chiangrai Prachanukroh Hospital, Chiang Rai, Thailand
- 7Department of Surgery, Khon Kaen Hospital, Khon Kaen, Thailand
- 8Department of Surgery, Sunpasitthiprasong Hospital, Ubon Ratchathani, Thailand
- 9Department of Surgery, Ratchaburi Hospital, Ratchaburi, Thailand
- KMID: 2555824
- DOI: http://doi.org/10.5223/pghn.2024.27.3.168
Abstract
- Purpose
The presence of alcoholic stool in biliary atresia (BA) patients is the basis of a stool color card (SCC), a screening tool that has led to more patients receiving Kasai portoenterostomy earlier. This study aimed to evaluate the color image processing of stool images captured using smartphones. We propose that measuring digital color parameters is a more objective method for identifying BA stools and may improve the sensitivity of BA screening.
Methods
A prospective study was conducted in five hospitals in Thailand between October 1, 2020, and December 31, 2021. Stools from infants presenting with jaundice, acholic stool, or dark-colored urine were photographed. Digital image color analysis was performed, and software was developed based on the color on the original SCC. Sensitivity and specificity for predicting BA stools were compared between the SCC and the software.
Results
Of 33 infants eligible for data collection, 19 were diagnosed with BA. Saturation and blue were two potential digital color parameters used to differentiate BA stools. The receiver operating characteristic curve was used to determine the optimum cutoff point of both values, and when saturation ≤56 or blue ≥61 was set as a threshold for detecting BA stool, high accuracy was achieved at 81.8% and 78.8%, respectively.
Conclusion
Digital image processing is a promising technology. With appropriate cutoff values of saturation in hue, saturation, value and blue in red, green, blue color models, BA stools can be identified, and equivocal-colored stools of non-BA patients can be differentiated with acceptable accuracy in infants presenting with jaundice.
