| Autor: |
Erlöv T; Department of Biomedical Engineering, Faculty of Engineering, Lund University, 22363 Lund, Sweden., Hawez T; Pediatrics, Department of Clinical Sciences, Faculty of Medicine, Lund University, 22185 Lund, Sweden., Granéli C; Pediatrics, Department of Clinical Sciences, Faculty of Medicine, Lund University, 22185 Lund, Sweden.; Department of Pediatric Surgery, Skåne University Hospital Lund, 22185 Lund, Sweden., Evertsson M; Pediatrics, Department of Clinical Sciences, Faculty of Medicine, Lund University, 22185 Lund, Sweden.; Department of Clinical Sciences Lund, Biomedical Engineering, Lund University, 22363 Lund, Sweden., Jansson T; Department of Clinical Sciences Lund, Biomedical Engineering, Lund University, 22363 Lund, Sweden.; Clinical Engineering Skåne, Digitalisering IT/MT, RegionSkåne, 22185 Lund, Sweden., Stenström P; Pediatrics, Department of Clinical Sciences, Faculty of Medicine, Lund University, 22185 Lund, Sweden.; Department of Pediatric Surgery, Skåne University Hospital Lund, 22185 Lund, Sweden., Cinthio M; Department of Biomedical Engineering, Faculty of Engineering, Lund University, 22363 Lund, Sweden. |
| Abstrakt: |
Ultra-high-frequency ultrasound (UHFUS) has a reported potential to differentiate between aganglionic and ganglionic bowel wall, referred to as histoanatomical differences. A good correlation between histoanatomy and UHFUS of the bowel wall has been proven. In order to perform more precise and objective histoanatomical morphometrics, the main research objective of this study was to develop a computer program for the assessment and automatic calculation of the histoanatomical morphometrics of the bowel wall in UHFUS images. A computer program for UHFUS diagnostics was developed and presented. A user interface was developed in close collaboration between pediatric surgeons and biomedical engineers, to enable interaction with UHFUS images. Images from ex vivo bowel wall samples of 23 children with recto-sigmoid Hirschsprung's disease were inserted. The program calculated both thickness and amplitudes (image whiteness) within different histoanatomical bowel wall layers. Two observers assessed the images using the program and the inter-observer variability was evaluated. There was an excellent agreement between observers, with an intraclass correlation coefficient range of 0.970-0.998. Bland-Altman plots showed flat and narrow distributions. The mean differences ranged from 0.005 to 0.016 mm in thickness and 0 to 0.7 in amplitude units, corresponding to 1.1-3.6% and 0.0-0.8% from the overall mean. The computer program enables and ensures objective, accurate and time-efficient measurements of histoanatomical thicknesses and amplitudes in UHFUS images of the bowel wall. The program can potentially be used for several bowel wall conditions, accelerating research within UHFUS diagnostics. |
