Baba Justin S, Chung Jung-Rae, DeLaughter Aimee H, Cameron Brent D, Coté Gerard L
Texas A&M University, Biomedical Engineering Program, College Station, Texas 77843-3120, USA.
J Biomed Opt. 2002 Jul;7(3):341-9. doi: 10.1117/1.1486248.
The high fatality rate associated with the late detection of skin cancer makes early detection crucial in preventing death. The current method for determining if a skin lesion is suspect to cancer is initially based on the patient's and physician's subjective observation of the skin lesion. Physicians use a set of parameters called the ABCD (asymmetry, border, color, diameter) rule to help facilitate diagnosis of potential cancerous lesions. Lesions that are suspicious then require a biopsy, which is a painful, invasive, and a time-consuming procedure. In an attempt to reduce the aforementioned undesirable elements currently associated with skin cancer diagnosis, a novel optical polarization-imaging system is described that has the potential to noninvasively detect cancerous lesions. The described system generates the full 16-element Mueller matrix in less than 70 s. The operation of the system was tested in transmission, specular reflection, and diffuse reflectance modes, using known samples, such as a horizontal linear polarizer, a mirror, and a diffuser plate. In addition, it was also used to image a benign lesion on a human subject. The results of the known samples are in good agreement with their theoretical values with an average accuracy of 97.96% and a standard deviation of 0.0084, using 16 polarization images. The system accuracy was further increased to 99.44% with a standard deviation of 0.005, when 36 images were used to generate the Mueller matrix.
皮肤癌晚期检测所带来的高死亡率使得早期检测对于预防死亡至关重要。目前用于确定皮肤病变是否疑似癌症的方法最初基于患者和医生对皮肤病变的主观观察。医生使用一组称为ABCD（不对称性、边界、颜色、直径）规则的参数来辅助潜在癌性病变的诊断。可疑病变随后需要进行活检，这是一个痛苦、有创且耗时的过程。为了减少目前与皮肤癌诊断相关的上述不良因素，描述了一种新型光学偏振成像系统，它有可能非侵入性地检测癌性病变。所描述的系统在不到70秒的时间内生成完整的16元素穆勒矩阵。该系统的操作在透射、镜面反射和漫反射模式下进行测试，使用已知样本，如水平线性偏振器、镜子和漫射板。此外，它还用于对人体受试者的良性病变进行成像。使用16幅偏振图像时，已知样本的结果与其理论值高度吻合，平均准确率为97.96%，标准差为0.0084。当使用36幅图像生成穆勒矩阵时，系统准确率进一步提高到99.44%，标准差为0.005。
