A geometric modeling approach using just five cardiac MRI slices achieves over 90% concordance with full-stack imaging for measuring left ventricular ejection fraction and stroke volume in 342 patients. The 2-chamber long-axis view combined with five short-axis slices showed the highest accuracy, with ejection fraction bias of only 0.83% and a minimum detectable change of 7.01% — better than the 8.7% variability between expert human readers. This finding addresses a longstanding clinical challenge in cardiac imaging. Traditional cardiovascular MRI requires numerous breath-holds and extensive slice acquisition, limiting patient comfort and scanner efficiency. The reduced-slice protocol could dramatically streamline cardiac function assessment while maintaining diagnostic reliability. The geometric model exploits the heart's predictable elliptical geometry, using mathematical approximations to interpolate missing data between fewer measurement points. However, absolute volume measurements showed systematic negative bias, though this cancels out when calculating relative metrics like ejection fraction. As a preprint awaiting peer review, these results require validation across diverse patient populations and scanner platforms. The approach represents an incremental but practically significant advance that could make cardiac MRI more accessible and tolerable for patients requiring serial monitoring.