Comprehensive Summary
The study by Santaló-Corcoy et al., investigates the effectiveness of the software application TAVI-PREP for pre-TAVI procedures such as evaluating anatomical measurements from the aortic valve taken by CT scans. In order to train the TAVI-PREP software to generate exact anatomical measurements based on pre-TAVI CT scans, various applications were used. This includes the MM-WHS dataset, MeshDeformNet for generating 3D surface meshes based off of provided pre-TAVI CT scan data, and a 3D Residual U-Net for landmark detection. Through this, TAVI-PREP software is able to analyze 22 measurements from the aortic valve based on CT scans. After a comparison between an expert cardiologist and the TAVI-PREP algorithm, a high Pearson correlation coefficient was found with a small mean absolute relative error for all measurements except for the left and right coronary height. With the accuracy in measurements found by the TAVI-PREP software, the researchers determined that it can provide reliable and useful measurements of the aortic valvular complex within an efficient amount of time.
Outcomes and Implications
Due to the invasiveness of open-heart surgery, TAVI, a less intensive method that works to help serious cases of aortic stenosis, proves to be a more well suited option for many patients. To minimize any potential procedural issues, having a reliable preoperative framework is essential. The researchers emphasized the effectiveness and high amount of support that the TAVI-PREP algorithm (or simply AI in general) can have in preoperative planning for TAVI procedures and the influence that this successful program can have on the widening of TAVI procedures.