Hemochromatosis (iron overload in liver) is a condition that causes serious consequences for the patient through an increase in the body's iron stores. Diagnosis of the excess iron, which is often stored in the liver, requires an invasive biopsy. We are developing neutron stimulated emission computed tomography (NSECT) as a non-invasive alternative to measure liver iron concentration to diagnose hemochromatosis. This measurement is performed using an incident neutron beam that scatters inelastically with iron nuclei in the liver, causing them to emit characteristic gamma-rays. An energy-sensitive gamma-ray detector is used to detect these gamma-rays and quantify the iron in the liver. Preliminary experiments have demonstrated an implementation of NSECT to quantify concentrations of iron and potassium in bovine liver tissue. Due to the prohibitive nature of these experiments, it is not feasible to perform system evaluation and optimization at each step using a nuclear accelerator. Here we describe a GEANT4 simulation of NSECT as a feasible alternative to perform system evaluation for iron overload diagnosis using computing resources only. The simulation model uses a 5 MeV neutron beam to scan a human liver phantom with induced iron overload. The liver is modeled as a composite shape combining a half-cylinder and a polyhedron, and is housed in a human torso filled with water. Gamma-ray spectra are generated to show element concentration within the liver. To determine the lower limit of iron overload detection, the concentration of iron in the liver is reduced from an initial high value, and the p-value of detecting peaks corresponding to iron is calculated at each step. The lower limit of detection is defined as the concentration at which the p-value of peak detection exceeds 0.05. The limit of iron overload detection from this simulation was found to be 4 mg/g, which represents a clinically relevant value for iron overload.