Kwangil Yim

Ovarian Cancer Detection in Ascites Cytology with Weakly Supervised Model on Nationwide Data Set

Conventional ascitic fluid cytology for detecting ovarian cancer is limited by its low sensitivity. To address this issue, this multicenter study developed patch image (PI)-based fully supervised convolutional neural network (CNN) models and clustering-constrained attention multiple-instance learning (CLAM) algorithms for detecting ovarian cancer using ascitic fluid cytology. Whole-slide images (WSIs), 356 benign and 147 cancer, were collected, from which 14,699 benign and 8025 cancer PIs were extracted. Additionally, 131 WSIs (44 benign and 87 cancer) were used for external validation. Six CNN algorithms were developed for cancer detection using PIs. Subsequently, two CLAM algorithms, single branch (CLAM-SB) and multiple branch (CLAM-MB), were developed. ResNet50 demonstrated the best performance, achieving an accuracy of 0.973. The performance when interpreting internal WSIs was an area under the curve (AUC) of 0.982. CLAM-SB outperformed CLAM-MB with an AUC of 0.944 for internal WSIs. Notably, in the external test, CLAM-SB exhibited superior performance with an AUC of 0.866 compared with ResNet50's AUC of 0.804. Analysis of the heatmap revealed that cases frequently misinterpreted by AI were easily interpreted by humans, and vice versa. Because AI and humans were found to function complementarily, implementing computer-aided diagnosis is expected to significantly enhance diagnostic accuracy and reproducibility. Furthermore, the WSI-based learning in CLAM, eliminating the need for patch-by-patch annotation, offers an advantage over the CNN model.

Collision Tumor of the Ovary: Adult Granulosa Cell Tumor and Mesonephric-like Adenocarcinoma

Collision tumors of the ovaries are rare, with only a few reports in the literature. Adult granulosa cell tumors are a relatively common primary tumor component of previously reported collision tumors. The combination of serous and mucinous tumors with adult granulosa cell tumors has been reported in several cases. On the other hand, mesonephric-like adenocarcinomas are rare neoplasms that commonly arise in the uterine corpus and ovaries. In this report, we present the case of a collision tumor composed of an adult granulosa cell tumor and mesonephric-like adenocarcinoma of the ovary in a 63-year-old woman. The initial magnetic resonance imaging findings showed a cystic mass with an internal hemorrhage, which suggested an adult granulosa cell tumor, and a solid mass with different enhancements. Microscopically, the tumor had two distinct components: An adult granulosa cell tumor and a mesonephric-like adenocarcinoma. Recognizing collision tumors consisting of slow-growing and aggressive tumors may prove beneficial in future diagnostic and treatment strategies.

Tumour budding in pretreatment cervical biopsies: a prognosticator for personalised therapy in the era of precision oncology

Aims Tumour budding (TB) is a noteworthy morphologic indicator for tumour microenvironment (TME) especially because it is detectable with routine haematoxylin and eosin (H&E) staining. Its prognostic relevance has been demonstrated across various cancers, but its significance in pretreatment biopsy specimens of cervical cancer is unknown. This is the first study to investigate the prognostic value of TB in pretreatment cervical biopsy. Additional TME features identifiable with H&E such as cell nest size (CNS) were evaluated. Methods and results A retrospective review was conducted on the 2018 International Federation of Gynaecology and Obstetrics (FIGO) stage IIVA cervical cancer patients ( N  = 182) who had completed standard treatment. In multivariate analysis, TB (hazard ratio [HR], 2.06) and CNS (HR, 2.16) independently predicted overall survival. While TB (AUC, 0.7065) slightly outperformed CNS (AUC, 0.6975) in discriminating overall survival, the combination of TB and CNS demonstrated the highest performance (AUC, 0.7192) in time‐dependent receiver operating characteristic analysis. Conclusions This study is the first to suggest TB in pretreatment biopsy specimens as a reliable morphologic prognosticator in cervical cancer. TME features may enhance precision oncology by offering insights into the individual tumour biology. The fact that these morphologic features are available from routine H&E slides, reserving immunohistochemistry or molecular analysis for indeterminate cases, is of particular value in low‐resource settings where the burden of cervical cancer is most significant.