Abstract: Conventional, manual identification and scoring of a whole slide image (WSI) to detect Cytokeratin 17 (K17) expression has limitations in terms of lower accuracy in the medium grade cases, slower turn-around time, and high inter and intra-observer variability. Each WSI can take a pathologist six to ten minutes to examine. This is because the brown color staining indicating the presence of K17 may be one of thousands or tens of thousands or other expressions present on the slide. Apparatus and methods are provided for an artificial intelligence (AI) solution that identifies and quantifies positively stained urothelial cells in WSI of urine cytology to aid the diagnosis of urinary bladder carcinoma. An illustrative AI solution may include one or computer models that are programmed to run one more machine learning (ML) models. The ML models may be trained to recognize subtle and complex patterns and features associated with specific diseases.

Abstract: Conventional, manual identification and scoring of a whole slide image (WSI) to detect Cytokeratin 17 (K17) expression has limitations in terms of lower accuracy in the medium grade cases, slower turn-around time, and high inter and intra-observer variability. Each WSI can take a pathologist six to ten minutes to examine. This is because the brown color staining indicating the presence of K17 may be one of thousands or tens of thousands or other expressions present on the slide. Apparatus and methods are provided for an artificial intelligence (AI) solution that identifies and quantifies positively stained urothelial cells in WSI of urine cytology to aid the diagnosis of urinary bladder carcinoma. An illustrative AI solution may include one or computer models that are programmed to run one more machine learning (ML) models. The ML models may be trained to recognize subtle and complex patterns and features associated with specific diseases.