Thyroid nodules are very common, and their incidence increase with age. The majority of thyroid nodules are benign, but approximately 5-15% are thyroid cancer. The cornerstone to evaluation of most thyroid nodules is a neck ultrasound followed by fine-needle aspiration (FNA) to evaluate for malignancy. Unfortunately, approximately 15-30% of FNAs are considered "indeterminate". In these cases, cytopathologist cannot determine if the thyroid nodule is benign or malignant based on FNA alone. Although the majority of these "indeterminate" nodules are ultimately benign, majority of these patients require thyroidectomies to rule out cancer. This puts the patient at unnecessary risk of surgical complications and increases health care costs. A better method is needed to help physicians determine the risk of malignancy in patients with indeterminate thyroid nodules. In recent years there has been much interest in the use of optical diagnostic in cancer detection. Recent investigations potentially suggest that Raman spectroscopy (RS) can be used as a clinical tool that could confer great patient advantage with minimally invasive, non-destructive, rapid and accurate diagnosis. In this study, we investigate the use of line-scan RS and imaging in combination with multivariate statistical analysis of the spectral data for objective identification and classification of single cells isolated from frozen samples of different types of human thyroid nodules. Preliminary results indicate a high sensitivity and specificity for identifying different cell types.