One of the limitations of this work is the difference in lesion size between the malignant and benign groups. The malignant lesions were on average larger than the benign lesions; 2.6 and 1.7 cm, respectively. Due to the partial volume effect, a larger lesion size may result in greater lesion to normal contrast than a smaller lesion. Furthermore, at similar depths, a larger lesion is more likely to be probed by the diffuse light field compared to a smaller lesion. To evaluate this effect, a subset of 12 lesions (6 benign, 6 malignant) with similar sizes and depths were selected from the cohort and analyzed separately. As was the case for the complete dataset, all DOSI parameters except lipid and saturation were found to be higher in the malignant group than in the benign group. These differences were statistically significant in L/N THb, water, and malignancy index, and nearly significant in deoxyhemoglobin and oxyhemoglobin concentrations. In the analysis of the entire population, statistical significance was also achieved for L/N contrast in deoxyhemoglobin, oxyhemoglobin, scatering power, and scattering amplitude. The loss in statistical differentiation is likely due to the smaller subgroup dataset. L/N THb and the malignancy index were found to be the best single predictors of malignancy, with AUCs of 0.94 (95% CI 0.82 to 1.00) and 1.00, respectively, in this subset. The findings were comparable between the subset with controlled lesion size and the complete population dataset.