The incidence of esophageal adenocarcinoma has continued to rise substantially, but the current standard of care for detecting cellular dysplasia and biopsy guidance has remained stagnant. Current white light endoscopy procedures often miss signs dysplasia, a common precursor to esophageal cancer. Optical scattering properties of tissue can provide quantitative details of tissue morphology and offer the possibility of improving the detection of pre-cancerous states. We present an implementation of Spatial Frequency Domain Imaging (SFDI) for characterizing optical scattering and absorption properties for pig and human esophagus samples. Through structured illumination, optical properties were extracted for various tissue types in carcinoma patients for improved tissue identification. Additionally, polarization gating and high spatial-frequency illumination techniques were implemented to quantify and visualize phase function information for tissue microstructure. During optical property extraction, SFDI implements corrections on reflectance measurements based on variation of sample height and angles measured by profilometry. This correction utilizes a phase measurement technique that is ideal for low spatial-frequency features and smooth surfaces but suffers from poor accuracy of high spatial-frequency structures due to noise limitations. We present the application of photometric stereo (PS) imaging, which illuminates a surface from four directions to calculate normal maps with improved accuracy on high frequency details. Additionally, we combine the low frequency components from profilometry with high frequency components from PS and validate our approach in silicone hemisphere phantoms and a pig colon. Through analysis of the frequency power spectra and comparing to a CT scan ground truth of the pig colon, we observe a 35.6%, 45.6%, and 51.0% improvement in accuracy when measuring the x, y, and z components, respectively, for the combined SFDI and PS normal maps compared to the traditional profilometry measurement. This method would therefore improve optical property extraction over complex topographies and remove the necessity of first flattening samples for accurate measurements.