The rapid development of quantum computing would pose a threat to traditional public-key cryptography. Post-quantum cryptography has gained significant attention to address this threat, with lattice-based schemes playing a central role. This paper focuses on CTRU [21], an NTRU-based key encapsulation mechanism (KEM), which is now in the standardization process in China. CTRU is the first NTRU scheme with ciphertext compression, high efficiency in both computation and bandwidth, security stronger than that of {R, M}LWE-based KEMs, and versatile choices in parameters and underlying algebraic structures. However, efficiently implementing CTRU on embedded platforms such as the ARM Cortex-M4 remains challenging due to its costly polynomial operations and the limited computational resources available. This work presents the first optimized implementation of the complete parameter sets of CTRU on the Cortex-M4, enhancing the efficiency of post-quantum cryptography in resource-constrained environments. The main contributions include: The experimental results show that CTRU-768 achieves speedups of 60.07%, 65.26%, and 91.68% in key generation, encapsulation, and decapsulation, respectively, over the reference implementation. This work provides a comprehensive and efficient implementation of CTRU across all parameter sets on the Cortex-M4 platform, making CTRU a viable candidate for secure post-quantum applications in embedded systems.