Liver segmentation in living liver transplant donors: comparison of semiautomatic and manual methods

Purpose: In our study, the right liver lobe volumes calculated with two different CT volumetry programmes, one of which was a manual software and had been used in routine practice for a long time and the other a semi-automatic CT volumetry programme developed specifically for the liver, were compared with the actual right liver lobe volumes measured intraoperatively to evaluate the validity of these programmes. Material and Method: Two radiologists evaluated preoperative abdominal CT angiography examinations of 109 liver donors who underwent living donor liver transplantation (LDLT) surgery between January 2009 and May 2018. Radiologists calculated the total and right lobe (segment V-VI-VII-VIII) volumes of the livers using two different CT volumetry programmes on portal venous phase axial images of CT examinations. The interaction times were noted in each CT volumetric measurement. The right liver lobe volumes calculated with CT volumetry programmes were compared with the actual right liver volumes measured intraoperatively. The inter-observer and inter-programme agreement were evaluated in all CT volumetric measurements. Results: Right liver lobe volumes calculated in both manual and semi-automatic programmes were found to be higher than the actual graft weights. The right lobe volumes calculated with CT volumetry programmes were 9.5% - 17.4% higher in the manual programme and 10.4% - 13.1% higher in the semi-automatic programme compared to the actual graft weight. All differences were statistically significant (p <0.001). We found excellent agreement between semi-automated and manual volumetric measurements for both total liver and the right lobe,. Inter-observer agreements were also excellent in both CT volumetry programmes. The difference between the two observers in the volumetric measurements of the right liver lobe was found to be less in the semi-automatic software than in the manual programme. Although experienced observer was faster, there was no significant difference between observers in terms of interaction time in both programmes. The less experienced observer measured higher volumes of the right lobe in both programs. Semiautomated liver volumetry was three to four times faster than manual software. Conclusion: We thought that the main reason for the volume over-estimation with CT volumetry programmes was disturbed perfusion and blood loss of the liver graft after resection. According to our study, conversion coefficients or formulas can be used after CT volumetric measurements to make more accurate volume estimates. Additionally, the semi-automatic programme was found to be more efficient since it provides volumetric measurements with similar accuracy to the manual programme in shorter time periods.