Publications by authors named "Kazumasa Tanabe"
Article Synopsis
- The study focused on creating an automated image classification model using deep learning to classify lung cancer from cytology images.
- Liquid-based cytology samples from a variety of lung cancer types were used to train the model, utilizing 9,141 image patches and employing techniques like Densenet-121 and 5-fold cross-validation.
- The results demonstrated high sensitivity and accuracy in predicting both general malignancy and specific types of lung cancer, suggesting that this model could enhance clinical diagnosis practices.
Introduction: The morphology of high-grade squamous intraepithelial lesion (HSIL) on Papanicolaou (Pap) smears widely varied, including syncytial aggregates, sheets, and scattered single cells, and no particular cellular pattern is consistently observed. Therefore, this study aimed to determine whether the menstrual cycle affects the cellular pattern of HSILs, an effort to avoid false negatives due to the oversight of scattered small single HSIL cells in the cytological triage of human papillomavirus-positive women.
Methods: A total of 147 HSIL samples of liquid-based cytology (LBC) in patients with cervical intraepithelial neoplasia grade 2 or 3 were obtained, and then, the relationship between cellular patterns, such as single-cell-like and syncytial aggregates, and menstrual cycles classified into six phases was analyzed.
Objectives: False-negatives on cytology may be observed during follow-ups for patients with persistent cervical intraepithelial neoplasia (CIN); however, the underlying reasons are unknown, and the relationship between the intra-individual variability of false-negatives and the menstrual cycle phase has not been elucidated. Therefore, this study aimed to determine whether the menstrual cycle influences cytological results to maximise the accuracy of such tests.
Methods: A total of 154 liquid-based cytological (LBC) samples were obtained during follow-ups for 26 patients with CIN, and the relationship between cytological results and the menstrual cycle, which was classified into six phases, was analysed.
Human papillomavirus (HPV) testing using self-collected vaginal specimens is the preferred choice to increase screening uptake. Although the HPV testing results of these samples depend on the cells that naturally exfoliate from the cervical lesion and drift into the vagina, the mechanism of when and how these exfoliated cells mix with the self-collected sample remains unclear. Hence, the study aimed to clarify the relationship between the vaginal drift of HPV-infected cells exfoliated from the cervix, and the menstrual cycle.
View Article and Find Full Text PDF