Daisuke Komura

This study investigates how label noise in pathology segmentation affects AI model performance, finding that boundary-altering noise leads to overestimation of model accuracy.

This study uses single-cell RNA sequencing to create a large cell atlas of the human stomach, revealing new insights into cell diversity and developmental features, including potential stem cell markers and fibroblast distribution patterns in gastric and metaplastic mucosa.

Viola extracts biologically meaningful structural variant signatures from cancer data and provides tools for custom classification and annotation.

This study uses supervised machine learning to classify BCR/Ig sequences from normal and tumor tissues, finding distinct sequence-level features between the two environments.

Luigi retrieves histopathological images using deep texture features and a pre-trained CNN, enabling fast and accurate similarity searches across cancer cases.