top of page

AI Performance

Reaching highest precision

Proven by extensive performance testing against renowned clinical experts¹

Exceeds state-of-the-art performance of 0.89 ¹ ² ³ ⁴

0.98

Accuracy*

Left Atrium

CE_edited.png

★

Exceeds state-of-the-art performance of 0.90 ¹ ² ³ ⁴

0.93

Accuracy*

Left Ventricle

CE_edited.png

★

Exceeds state-of-the-art performance of 0.85 ¹ ² ³ ⁴

0.96

Accuracy*

Right Atrium

CE_edited.png

★

Exceeds state-of-the-art performance of 0.87 ¹ ² ³ ⁴

0.89

Accuracy*

Right Ventricle

CE_edited.png

★

Exceeds state-of-the-art performance of 0.83 ¹ ² ³ ⁴

0.90

Accuracy*

Left Ventricular Muscle

CE_edited.png

★

Category

* Accuracy measured in Dice Score

° Accuracy measured in Mean Surface Distance

References

[1]  Khalique, O. (2025) Deep Learning Automation of Complete Heart Segmentation for Planning Multivalvular Structural Heart Interventions. SCAI 2025

[2] Habijan, M., H. Leventić, G. Irena and B. Danilo (2020). "Neural Network based Whole Heart Segmentation from 3D CT
images." International journal of electrical and computer engineering systems 11 (1): 25-31.

[3] Park, S., & Chung, M. (2021). Cardiac Segmentation on CT Images through Shape-Aware Contour Attentions

[4] Sundgaard J. V,. Juhl K.A., Kofoed K.F., Paulsen R.R., (2020) "Multi-planar whole heart segmentation of 3D CT images
using 2D spatial propagation CNN," Proc. SPIE 11313, Medical Imaging 2020: Image Processing

[5] Habijan, M., H. Leventić, G. Irena and B. Danilo (2020). "Neural Network based Whole Heart Segmentation from 3D CT
images." International journal of electrical and computer engineering systems 11 (1): 25-31.

[6] Sundgaard J. V,. Juhl K.A., Kofoed K.F., Paulsen R.R., (2020) "Multi-planar whole heart segmentation of 3D CT images
using 2D spatial propagation CNN," Proc. SPIE 11313, Medical Imaging 2020: Image Processing

★ We are continuously evaluating our performance against the state-of-the-art.
If you find a new publication, please write to us at ai-benchmark@laralab.de

bottom of page