Background

Treatment of tricuspid regurgitation with direct annuloplasty with Cardioband ® from Edwards is already an established intervention.  Prior to intervention, right heart anatomy is analyzed with full-cycle cardiac CT. Size of the annulus itself as well as its proximity to right coronary artery is crucial for eligibility of patients for this procedure. We analyzed annulus and RCA proximity with cardiac CT´s using artificial intelligence (AI) and compared it with analysis from conventional analysis with 3Mensio.

Methods

We analyzed 38 Patients who received Cardioband implantation with heart.ai, an artificial intelligence software (AI) provided from Laralab from Munich, Germany. These measurements were compared with “conventional” measurements (CM) with 3Mensio. Mean, standard deviation, mean difference, standard deviation of difference (SDD), limits of agreement (LoA) and intraclass correlation coefficients (ICCs) were calculated as well as Bland-Altman plots were constructed using SPSS Statistics 27 from IBM.

Results

In the currently investigated patients (68 % female, mean age 77 years) mean tricuspid area using AI was 1768 ±275 mm² as compared to 1819 ±235 mm² in CM; SDD with LoA was 126 (177; -315); ICC 0.91. Perimeter was 149 ±12.1 mm as assessed with AI and 154 ±10.8 mm as assessed by CM; SDD 7.21 (8.44; -19.8); ICC 0.84. RCA-Proximity was calculated as smallest distance between annulus and RCA. Mean smallest distance was 4.7 ± 2 mm in AI-analysis and 5.2 ±2.6 mm in CM; SDD 1.86 (3.13; -4.15); ICC 0.803 with a systematic underestimation in distance by AI.

Discussion and Conclusion

In this retrospective trial artificial Intelligence software provided excellent accuracy in measurement of tricuspid area and perimeter in patients undergoing full-cycle CT scan to evaluate anatomical eligibility for percutaneous valvuloplasty of the tricuspid valve.