# UniCAS: A foundation model for cervical cytology screening

## TL;DR

UniCAS is a new AI model trained on thousands of cervical slides to improve cancer screening and reduce diagnostic time.

## Contribution

UniCAS introduces a unified foundation model for cervical cytology that handles multiple tasks efficiently with high accuracy.

## Key findings

- UniCAS achieves AUC values of 92.60%, 92.58%, and 98.39% for cancer screening, candidiasis testing, and clue cell diagnosis.
- The model reduces diagnostic time by 70% compared to conventional approaches.
- UniCAS enables multi-scale analysis for slide-level, region-level, and pixel-level tasks.

## Abstract

Cervical abnormality screening is pivotal for prevention and treatment. However, the substantial size of whole slide images (WSIs) makes examination labor-intensive and time-consuming. Current deep learning-based approaches struggle with the morphological diversity of cervical cytology and require specialized models for distinct diagnostic tasks, leading to fragmented workflows. Here, we present UniCAS, a cytology foundation model pre-trained on 48,532 cervical WSIs encompassing diverse patient demographics and pathological conditions. UniCAS enables various clinical analysis tasks, achieving state-of-the-art performance in slide-level diagnosis, region-level analysis, and pixel-level image enhancement. In particular, by integrating a multi-task aggregator for slide-level diagnosis, UniCAS achieves area under the curve (AUC) values of 92.60%, 92.58%, and 98.39% for cancer screening, candidiasis testing, and clue cell diagnosis, respectively, while reducing diagnostic time by 70% compared with conventional approaches. This work establishes a paradigm for efficient multi-scale analysis in automated cervical cytology, bridging the gap between computational pathology and clinical diagnostic workflows.

•UniCAS is a cytology foundation model pre-trained on 48,532 cervical slides•UniCAS outperforms other histopathology foundation models in various cytology tasks•Multi-task aggregator concurrently handles slide-level tasks without performance drop•A unified diagnostic pipeline can reduce the overall computational overhead by 70%

UniCAS is a cytology foundation model pre-trained on 48,532 cervical slides

UniCAS outperforms other histopathology foundation models in various cytology tasks

Multi-task aggregator concurrently handles slide-level tasks without performance drop

A unified diagnostic pipeline can reduce the overall computational overhead by 70%

Jiang et al. present UniCAS, a foundation model pre-trained on 48,532 whole slide images for cervical cytology screening. By functioning as a unified encoder for multi-scale analysis, UniCAS enables various clinical analysis tasks with state-of-the-art performance, thereby addressing the inefficient and fragmented workflows of conventional approaches.