It’s been a while since my last blog Inside the Digital Society for APC. I’m glad to return and will be writing each month on an important issue for those concerned about the impact that digitalisatio...

Publicly available datasets used in this study can be accessed from their original repositories. TCGA diagnostic H&E WSIs and matched molecular and clinical annotations are available from the Genomic Data Commons portal (https://portal.gdc.cancer.gov/). The HEMIT mIF CRC dataset is available at https://github.com/BianChang/HEMIT-DATASET. The CODEX CRC dataset is available from The Cancer Imaging Archive at https://www.cancerimagingarchive.net/collection/crc_ffpe-codex_cellneighs/. The in-house G4X spatial proteomics cohorts, the private Orion LUAD cohort and the CRC and NSCLC immunotherapy cohorts were obtained and used under institutional ethical approvals and applicable data-use agreements. The patient-level data, including H&E images, spatial molecular imaging data and associated clinical information, were used with permission for the current study and are not publicly available. Any reuse requires approval from the corresponding ethics committee, and any dissemination must comply with applicable data privacy laws and the scope of patient consent and information documents. These documents do not authorize public release of the data. Removal of direct identifiers from processed datasets does not remove these governance and data-use restrictions; therefore, neither raw nor de-identified processed data may be redistributed through unrestricted public repositories. Access may be granted following institutional approval and execution of an appropriate data-use agreement, subject to applicable ethical, privacy and data-security...

Initial biomarker discovery was performed using tumor samples from patients with advanced NSCLC enrolled in TROPION-PanTumor01, a first-in-human, phase 1 study investigating Dato-DXd monotherapy in patients with solid tumors34,36. A total of 180 patients with NSCLC were enrolled (intention-to-treat (ITT) population) and received Dato-DXd at 4 mg kg−1, 6 mg kg−1 or 8 mg kg−1 (ref. 34); 115 patients (113 for visual scoring) with digitized TROP2 IHC-stained, WSIs of tumor samples comprised the TROPION-PanTumor01 biomarker-evaluable population (BEP; see Extended Data Table 1 for patient characteristics in the BEP and ITT population). Some patients in the BEP either harbored an AGA or had squamous histology and, of the latter, none had an objective response, limiting discovery efforts in this subpopulation. Analysis of TROP2 expression using a research-use-only (RUO) IHC assay was conducted by visual pathologist scoring34. TROP2 expression by manual scoring (H-score) was not associated with centrally confirmed objective response (Fig.1a), PFS (Fig. 1b) or reduction in tumor size from baseline (Fig. 1c), consistent with previously published data on TROP2 expression in NSCLC34,52. Furthermore, high TROP2 expression alone was not associated with response to Dato-DXd (Fig. 1d). Based on the QCS platform42, we developed QCS for TROP2—a fully supervised computational pathology...

Computational neurodegenerative neuropathology represents a transformative approach in the analysis and understanding of neurodegenerative diseases through utilization of whole slide images (WSIs) and advanced machine learning/artificial intelligence (ML/AI) techniques. This review explores the emerging field of computational neurodegenerative neuropathology, emphasizing its potential to enhance neuropathologic assessment, diagnosis, and research. Recent advancements in ML/AI technologies hav...

We utilized an in-house dataset, UWOV, with 152 patients (145 HGSOC and 7 other ovarian cancer patients) as well as two publicly available datasets: TCGAOV (81 HGSOC patients with 82 diagnostic WSIs) and PTRC (93 HGSOC patients) (Fig. 1a and Methods). For our model development, we focused on the UWOV cohort due to its high-quality, large sample size, and the comprehensive information from multiple slides of different anatomical locations per patient, and used TCGAOV and PTRC as independent validation sets where we only used diagnostic slides from TCGA. The UWOV dataset consists of 720 diagnostic whole tissue section slides from primary debulking resections of HGSOC. The specimens encompassed fallopian tube or ovary (tubo-ovarian: n = 480) and extra-ovarian anatomic locations: omentum (n = 151), abdomen (n = 29), Other Pelvic (n = 42), and other locations (n = 18) (Fig. 1b). HRDetect was used to generate ground truth HRD status for the UWOV dataset19 (“Methods”). 70 patients were classified as HRD−, while 82 were classified as HRD+ . These cancers were also assessed for HRD-related genomic alterations, including BRCA1 (32.9%), BRCA2 (25.7%), and RAD51C (2%) mutations (Fig. 1c and Methods). Methylation alterations were also assessed, where among the HRD+ cancers,...

AVIA, de la Fundación Cuidados de Salud, ganó el WSIS Prizes 2026 y muestra cómo la inteligencia artificial en salud puede prevenir el deterioro en adultos mayores

AVIA, de la Fundación Cuidados de Salud, ganó el WSIS Prizes 2026 y muestra cómo la inteligencia artificial en salud puede prevenir el deterioro en adultos mayores

Automated breast cancer classification from whole-slide images (WSIs) is hampered by the scarcity of expert-annotated patches and the domain variability inhe...

Whole-slide images (WSIs) are challenging for vision-language reasoning because diagnostically relevant morphology is sparse, heterogeneous, and distributed across gigapixel-scale images and multiple spatial resolutions. Existing WSI models and pathology agents can aggregate slide features or activ…

Whole-slide images (WSIs) are challenging for vision-language reasoning because diagnostically relevant morphology is sparse, heterogeneous, and distributed across gigapixel-scale images and multiple spatial resolutions. Existing WSI models and pathology agents can aggregate slide features or activ…

We develop end-user software in the areas of Computational & Digital Pathology and Spatial Biology. Brief summary of the select packages can be accessed via the left buttons. Latest codes are freely a...