Causal discovery aims to uncover causal relationships among variables from observational data, serving as a crucial method for understanding various phenomena and changes in natural, social, and technological systems. A mainstream approach for causal discovery is a constraint-based algorithm, which determines the causal structure among variables by examining their conditional independence. However, data collection in the real world often faces challenges such as limited sample sizes and high variance among nodes due to resource or technical constraints. In these scenarios, the accuracy of conditional independence tests is greatly affected, leading to erroneous deletion of causal edges of some variables in learned causal graphs, thereby impacting the accuracy of the algorithm’s output. To address this issue, this study proposes an enhanced method for conditional independence testing, which focuses on minimizing the interference of irrelevant external noise on the variables being tested, thereby improving the accuracy of conditional independence tests. Based on this enhanced method, the paper introduces a structure learning algorithm based on heuristic search, which iteratively searches for mistakenly deleted causal edges on a graph with an initial structure. This algorithm reconstructs the causal structure by combining enhanced conditional independence tests with score optimization. Experimental results show that, compared to existing methods, the proposed algorithm significantly improves both the F1 score and the structural Hamming distance (SHD) on simulated, Bayesian network, and real data, demonstrating its ability to more accurately reveal underlying causal structures in observational data with limited samples and high-variance nodes.