Kimura Fumikazu, Ohshima Kengo, Kanai Ryo, Urata Takumi, Kobayashi Takaki, Takeyama Saori, Yamaguchi Masahiro, Asaka Shiho, Ishii Keiko
Department of Biomedical Laboratory Sciences, School of Health Sciences, Shinshu University, Matsumoto, Japan.
Department of Information Processing, Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology, Yokohama, Japan.
Acta Cytol. 2025;69(2):131-143. doi: 10.1159/000543224. Epub 2024 Dec 26.
Although the widespread use of screening tests and HPV vaccines for squamous cell carcinoma has led to early detection and treatment, effectiveness is limited for cervical adenocarcinoma. Lobular endocervical glandular hyperplasia (LEGH) corresponds to gastric metaplasia but is regarded as a pathological condition with subtle morphological abnormalities. LEGH is a benign lesion and a precursor to gastric-type adenocarcinoma. We herein developed an objective and quantitative method by applying an image analysis technique to overcome the difficulties associated with the differential diagnosis of LEGH in uterine cervical cytology. This approach is expected to enable the early detection and accurate diagnosis of LEGH.
We extracted signal values for the nucleus and cytoplasm from microscopic images of cytological specimens of normal endocervical (EC) and LEGH cells. These values were then converted into CIELAB and sRGB values to create color distribution maps, and color unmixing techniques were applied to assess the spectral absorbance of each pigment.
The CIELAB signal values extracted from the nuclear images of LEGH cells exhibited lower values than those of EC cells. Furthermore, based on color distribution maps, the cytoplasm of EC cells exhibited shades from purple to pink, while LEGH cells showed a distribution toward yellow.
This study reveals that, compared to EC cells, LEGH cells exhibit lower nuclear signal values and increased nuclear chromatin content. Thus, assessing the relative difference in cytoplasmic color tones between them may become an effective indicator for distinguishing between EC and LEGH cells.
尽管鳞状细胞癌的筛查试验和HPV疫苗的广泛使用已实现早期检测和治疗，但对宫颈腺癌的有效性有限。宫颈小叶内腺增生（LEGH）与胃化生相对应，但被视为具有细微形态异常的病理状况。LEGH是一种良性病变，也是胃型腺癌的前体。我们在此开发了一种客观定量的方法，通过应用图像分析技术来克服宫颈细胞学中LEGH鉴别诊断的困难。这种方法有望实现LEGH的早期检测和准确诊断。
我们从正常宫颈内膜（EC）和LEGH细胞的细胞学标本的显微图像中提取细胞核和细胞质的信号值。然后将这些值转换为CIELAB和sRGB值以创建颜色分布图，并应用颜色分解技术评估每种色素的光谱吸收率。
从LEGH细胞核图像中提取的CIELAB信号值低于EC细胞。此外，基于颜色分布图，EC细胞的细胞质呈现从紫色到粉红色的色调，而LEGH细胞则呈现向黄色的分布。
本研究表明，与EC细胞相比，LEGH细胞的核信号值较低且核染色质含量增加。因此，评估它们之间细胞质色调的相对差异可能成为区分EC和LEGH细胞的有效指标。
