The emerging paradigm of Direct-to-Satellite Internet of Things (DtS-IoT) involves Earth surface nodes communicating directly with Low Earth Orbit (LEO) satellites, utilizing standard Low-Power Wide Area Networks (LPWAN) protocols. One of the core challenges faced in this paradigm is scaling the Medium Access Control (MAC) from a limited number of nodes to potentially thousands within the satellite's coverage area. To address this issue, medium access control schemes can utilize a priori information on the number of nodes the satellite will cover along its orbit. However, developing technically viable solutions for network size estimation that are both precise and accurate remains an open research challenge. This work presents the implementation, parameter selection, and evaluation of the first LoRa/LoRaWAN-compatible network size estimation protocol that leverages the onboard Optimistic Collision Information (OCI) estimator. Our solution, LoRa-OCI (L-OCI), was integrated into FLoRaSat, a C++ discrete-event DtS-IoT simulator that integrates realistic orbital and LoRa/LoRaWAN communication models. Through an extensive simulation campaign, we can determine appropriate LoRa configurations to achieve low root mean square error (RMSE) and low power consumption. Additionally, our results indicate that the approach is relatively insensitive to LoRa parameters when assessing the aggregated throughput of a Slotted ALOHA Game (SAG) protocol throttled by L-OCI.