Automated Reading of Dual Stain Cytology,
AI-Based Image Analysis, Standardized Cancer Screening.
Cervical cancer screening is one of the most globally impactful cancer preventive measures and stands out due to its high degree of automation when primary HPV testing is used, which enables efficient and large-scale implementation but is lacking specificity. Therefore, effective triage and management of HPV-positive women is critical to avoid unnecessary colposcopy referrals and associated harms while maintaining high sensitivity for cervical precancer. Triage with p16/Ki-67 dual-stain (DS) testing has shown high sensitivity and specificity for detection of cervical precancers; however, its currently missing automation is hindering global implementation and adoption of the assay itself, in turn restricting the full potential of HPV-based cancer screeening. Dual-stain (DS) has already been included as an important component in the US Enduring Consensus Guidelines and the World Health Organization (WHO) cervical cancer screening guidelines as a recommended triage method for managing HPV-positive individuals. Compared with PAP cytology, dual-stain requires fewer subsequent colposcopies and detects more, and earlier, cervical intraepithelial neoplasia grade 3 or greater.
CYTOREADER™ is the first and only AI-driven solution for assisting with dual-stain cytology (p16/Ki-67) and ist built around a new AI-based algorithm integrated into a fully automated laboratory and analysis process. It is designed to most efficiently assist readers in detecting dual-stain events in very high throughput cervical cancer screening. The system allows for a fast and efficient analysis of all cells on a ThinPrep® or Surepath® liquid cytology slides stained with the CINtec PLUS® dual-stain assay. It creates a gallery of the 30 most relevant diagnostic image tiles based on computed likelihood for an image to contain a dual-stain positive cell. Readers thus get the decisive information necessary for fast diagnosis at immediate sight. Fully automatic reading of dual-stain slides using CYTOREADER has been evaluated in externallly validated, blinded scientific studies by the US National Cancer Institute.
*Cytoreader is not yet FDA or CE certified and is for research use only.
*The underlying AI technology is patent protected in the US (US 11,954,593 B2), European Union (IE20180171A) and China (CN112543934A).
Cytoreader-V2 was validated on 3,803 patients from the Kaiser Permanente Northern California (KPNC) Dual Stain Implementation Study (SurePathTM) [1], the NCI Biopsy Study (ThinPrep®) [2,3], the Improving Risk Informed HPV Screening study IRIS (SurePathTM) [4,5], the study STRIDES Studying Risk to Improve DisparitiES in Cervical Cancer in Mississippi (ThinPrep®) [6,5], and ACSS, the Anal Cancer Screening Study (ThinPrep®) [2,8]. Cytoreader-V2 has been evaluated to work with the Whole Slide Imaging (WSI) Scanners from Hamamatsu Photonics (Nanozoomer-Series) or Roche (DP-Series).
Cytoreader-V2 has unique robustness as each image tile is evaluated by an ensemble of AI algorithms simultaneously and has been validated on the dual-stain assay CINtec+® from Roche and on liquid cytology slides prepared with ThinPrep® and SurePath™. It comprises a complex, barcode-automated workflow management. Also multiple AI algorithms automatically check in-parallel quality of all slides for focus quality, cellularity and staining failures. This way correct dual-stain immunochemistry is AI supervised. Cytotechs or pathologists can make a diagnosis extremely fast in any web browser by displaying the 30 most likely dual-stain-containing tiles for each case.
Efficiently assess the 30 most relevant dual-stain events directly linked to the whole-slide image for quick diagnostics.
Streamline workflow with shortcut keys that enable rapid toggling through galleries, entering diagnoses, and transitioning between slides in under one minute.
Enhance slide analysis by annotating regions of interest, viewing computed likelihoods, and utilizing heatmaps for comprehensive insights.
Enable blinded analysis of computational results for training, quality control, and statistical monitoring of diagnostic performance.
Responsible for content:
Germany: H-Labs GmbH, Heckerstr. 9, 69124 Heidelberg, Germany
Internationally: Centauris med d.o.o., Rudarska 1, 52220 Labin, Croatia
Contact: Prof. Dr. Niels Grabe
Email: niels.grabe@stcmed.com