Terahertz wave technique is very prospective in the biomedical application field, where there exists a practical problem that it is hard to register a terahertz time-domain spectroscopy (THz-TDS) detection point and its accurate corresponding position in pathology image. In this paper, complex human gastric tissue is taken as the experimental subject to search for accurate detection approach. This paper uses tissue microarray technology to process paraffin-embedded gastric tissue, and then contrasts the Terahertz absorption coefficient and refractive index spectrum of detection subjects which are divided into registered experimental group and unregistered control group using principal component analysis method. Meanwhile, this paper contrasts the gastric tubular adenocarcinoma THz-TDS discrimination methods using support vector machine and logistic regression respectively. The research result reveals that the detection accuracy brought by tissue microarray technology helps better discriminate tumor samples and normal samples.