Yang Weixi, Tian Rong, Xue Tongqing
Department of Burn and Plastic Surgery, Huai'an First People's Hospital, Nanjing Medical University, Huai'an, 223300, China.
Tumour Biol. 2015 Aug;36(8):6125-31. doi: 10.1007/s13277-015-3294-5. Epub 2015 Mar 10.
We evaluated whether degrees of dysplasia may be consistently accessed in an automatic fashion, using different kinds of non-melanoma skin cancer (NMSC) as a validatory model. Namely, we compared Bowen disease, actinic keratosis, basal cell carcinoma, low-grade squamous cell carcinoma, and invasive squamous cell carcinoma. We hypothesized that characterizing the shape of nuclei may be important to consistently diagnose the aggressiveness of a skin tumor. While basal cell carcinoma is comparatively relatively benign, management of squamous cell carcinoma is controversial because of its potential to recur and intraoperative dilemma regarding choice of the margin or the depth for the excision. We provide evidence here that progressive nuclear dysplasia may be automatically estimated through the thresholded images of skin cancer and quantitative parameters estimated to provide a quasi-quantitative data, which can thenceforth guide the management of the particular cancer. For circularity, averaging more than 2500 nuclei in each group estimated the means ± SD as 0.8 ± 0.007 vs. 0.78 ± 0.0063 vs. 0.42 ± 0.014 vs. 0.63 ± 0.02 vs. 0.51 ± 0.02 (F = 318063.56, p < 0.0001, one-way analyses of variance). The mean aspect ratios were (means ± SD) 0.97 ± 0.0014 vs. 0.95 ± 0.002 vs. 0.38 ± 0.018 vs. 0.84 ± 0.0035 vs. 0.74 ± 0.019 (F = 1022631.931, p < 0.0001, one-way analyses of variance). The Feret diameters averaged over 2500 nuclei in each group were the following: 1 ± 0.0001 vs. 0.9 ± 0.002 vs. 5 ± 0.031 vs. 1.5 ± 0.01 vs. 1.9 ± 0.004 (F = 33105614.194, p < 0.0001, one-way analyses of variance). Multivariate analyses of composite parameters potentially detect aggressive variants of squamous cell carcinoma as the most dysplastic form, in comparison to locally occurring squamous cell carcinoma and basal cell carcinoma, or benign skin lesions.
我们评估了是否可以使用不同类型的非黑色素瘤皮肤癌（NMSC）作为验证模型，以自动方式一致地评估发育异常程度。具体而言，我们比较了鲍温病、光化性角化病、基底细胞癌、低级别鳞状细胞癌和浸润性鳞状细胞癌。我们假设，表征细胞核的形状对于一致地诊断皮肤肿瘤的侵袭性可能很重要。虽然基底细胞癌相对较为良性，但鳞状细胞癌的治疗存在争议，因为它有复发的可能性，并且在手术中对于切除边缘或深度的选择存在困境。我们在此提供证据表明，通过皮肤癌的阈值图像和估计的定量参数，可以自动估计渐进性核发育异常，从而提供准定量数据，进而指导特定癌症的治疗。对于圆形度，每组平均超过2500个细胞核，估计的均值±标准差分别为0.8±0.007、0.78±0.0063、0.42±0.014、0.63±0.02、0.51±0.02（F = 318063.56，p < 0.0001，单因素方差分析）。平均纵横比分别为（均值±标准差）0.97±0.0014、0.95±0.002、0.38±0.018、0.84±0.0035、0.74±0.019（F = 1022631.931，p < 0.0001，单因素方差分析）。每组2500多个细胞核的费雷特直径平均值如下：1±0.0001、0.9±0.002、5±0.031、1.5±0.01、1.9±0.004（F = 33105614.194，p < 0.0001，单因素方差分析）。与局部发生的鳞状细胞癌、基底细胞癌或良性皮肤病变相比，复合参数的多变量分析可能将鳞状细胞癌的侵袭性变体检测为发育异常最严重的形式。
