PV26 Speakers

Subject to change.

 

 

image

Tien-Jen Liu, MD

Co-Founder & COO, AIxMed, Inc.


Tien-Jen Liu, MD, is an experienced attending physician with a proven history in the biotechnology industry. Dr. Liu is skilled in clinical research, medical device development, clinical study, and business development. He is also the director of the biomedical innovation program at a medical university.

 

 

SESSIONS

Autopilot for cytology slide digitization: A novel QC workflow for high-quality whole-slide imaging
   Sun, Oct 18
   1:35 PM - 1:55 PM PT
  Seaport H

Introduction: Digital cytology requires consistently high-quality, well-focused whole-slide images (WSIs), yet cytology specimen scanning is challenged by uneven cell distribution, variable preparations, and focal complexity. Workflow inefficiencies result when these issues are detected only after initial manual review. We developed and evaluated novel AI-powered software integrated into the digitization workflow to automate quality control (QC) and rescan slides that fail QC through intelligent focus-point adjustment.

Design: We prospectively tested the performance of 120 ThinPrep cytology slides across different specimen types from 80 urine, 20 cervical, and 20 thyroid cases. Slides were digitized using a Hamamatsu S360MD scanner with a cytology-optimized scan profile derived from prior studies. We compared manual QC (MQC), in which a cytologist classified WSIs as pass or fail based on image focus and rescanned failed slides after manual focus-point adjustment until QC passed; and automated CytoQC (ACytoQC), in which dedicated software assessed image quality using AI and unattended rescanning of failed slides using AI-suggested focus points until QC passed. Outcomes included QC pass rate, rescan frequency, and total time to achieve 100% QC pass.

Results: After the first scan, QC pass rates for 120 WSIs were 91% with MQC and 87% with ACytoQC (Table 1). Both workflows reached 97% after the second scan of failed slides and 98% after the third scan. ACytoQC achieved 100% QC pass by the fourth scan, while MQC required a fifth scan to achieve the same result. ACytoQC also reduced downstream QC and rescan time by 68% versus MQC (0.9 vs 2.8 hours), saving 1.9 hours of manual labor (Fig. 1).

Discussion: ACytoQC provides a novel approach to cytology WSI quality assessment that improves digitization efficiency. In this study, it reduced QC and rescan time by 1.9 hours compared with the manual workflow and offers a scalable solution for unattended after-hours slide digitization.

Learning Objectives:

  1. Recognize key challenges in cytology slide digitization and the need for automated WSI quality control.
  2. Compare automated CytoQC-guided rescanning with manual QC for efficient cytology slide digitization.
  3. Explain how automated CytoQC removes manual QC bottlenecks and enables unattended cytology digitization.
Chat bot