Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

AI Assisted Cervical Cancer Screening for Cytology Samples in Developing Countries

Love Panta, Suraj Prasai, Karishma Malla Vaidya, Shyam Shrestha, Suresh Manandhar

TL;DR

Cervical cancer screening in developing countries is hampered by labor-intensive Liquid-Based Cytology and limited access to expert pathologists. The authors propose an end-to-end pipeline that combines a motorized, low-cost microscope with AI: an image-stitching module builds panoramic cytology views, a human-in-the-loop cellpose2.0 segmentation refines cell masks, and a Convolutional Vision Transformer (CvT-13) performs five-class cervical cell classification trained on the SIPaKMeD dataset. Key contributions include the CYTOCERVIX dataset collected in Nepal, the demonstrated effectiveness of CYTOCERVIX_cyto2 and combined_cyto2 segmentation variants, and a high-performing CvT-13 classifier achieving 99.68% accuracy with an AUC of approximately 1.0 on SIPaKMeD. Together, these results indicate a scalable, affordable screening approach suitable for low-resource settings, and the authors plan to publicly release their segmentation dataset for broader research use.

Abstract

Cervical cancer remains a significant health challenge, with high incidence and mortality rates, particularly in transitioning countries. Conventional Liquid-Based Cytology(LBC) is a labor-intensive process, requires expert pathologists and is highly prone to errors, highlighting the need for more efficient screening methods. This paper introduces an innovative approach that integrates low-cost biological microscopes with our simple and efficient AI algorithms for automated whole-slide analysis. Our system uses a motorized microscope to capture cytology images, which are then processed through an AI pipeline involving image stitching, cell segmentation, and classification. We utilize the lightweight UNet-based model involving human-in-the-loop approach to train our segmentation model with minimal ROIs. CvT-based classification model, trained on the SIPaKMeD dataset, accurately categorizes five cell types. Our framework offers enhanced accuracy and efficiency in cervical cancer screening compared to various state-of-art methods, as demonstrated by different evaluation metrics.

AI Assisted Cervical Cancer Screening for Cytology Samples in Developing Countries

TL;DR

Cervical cancer screening in developing countries is hampered by labor-intensive Liquid-Based Cytology and limited access to expert pathologists. The authors propose an end-to-end pipeline that combines a motorized, low-cost microscope with AI: an image-stitching module builds panoramic cytology views, a human-in-the-loop cellpose2.0 segmentation refines cell masks, and a Convolutional Vision Transformer (CvT-13) performs five-class cervical cell classification trained on the SIPaKMeD dataset. Key contributions include the CYTOCERVIX dataset collected in Nepal, the demonstrated effectiveness of CYTOCERVIX_cyto2 and combined_cyto2 segmentation variants, and a high-performing CvT-13 classifier achieving 99.68% accuracy with an AUC of approximately 1.0 on SIPaKMeD. Together, these results indicate a scalable, affordable screening approach suitable for low-resource settings, and the authors plan to publicly release their segmentation dataset for broader research use.

Abstract

Cervical cancer remains a significant health challenge, with high incidence and mortality rates, particularly in transitioning countries. Conventional Liquid-Based Cytology(LBC) is a labor-intensive process, requires expert pathologists and is highly prone to errors, highlighting the need for more efficient screening methods. This paper introduces an innovative approach that integrates low-cost biological microscopes with our simple and efficient AI algorithms for automated whole-slide analysis. Our system uses a motorized microscope to capture cytology images, which are then processed through an AI pipeline involving image stitching, cell segmentation, and classification. We utilize the lightweight UNet-based model involving human-in-the-loop approach to train our segmentation model with minimal ROIs. CvT-based classification model, trained on the SIPaKMeD dataset, accurately categorizes five cell types. Our framework offers enhanced accuracy and efficiency in cervical cancer screening compared to various state-of-art methods, as demonstrated by different evaluation metrics.
Paper Structure (25 sections, 10 figures, 4 tables)

This paper contains 25 sections, 10 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Literature Review
  3. Cell Segmentation
  4. Cell Classification
  5. Methodology
  6. Samples Collection Procedures
  7. Image-stitching Pipeline
  8. Cell Segmentation
  9. Cell Classification
  10. Convolutional Vision Transformer
  11. Experimental Results & Discussions
  12. Datasets Used
  13. Segmentation datasets
  14. Cx22:
  15. CYTOCERVIX:
  16. ...and 10 more sections

Figures (10)

  • Figure 1: Human-on-the-loop approach for training our segmentation model pachitariu2022cellpose
  • Figure 2: Convolutional Vision Transformer pipeline wu2021cvt for image classification in our cell Images
  • Figure 3: Cytology Samples of Cx22 dataset liu2022cx22
  • Figure 4: Cytology Samples of CYTOCERVIX
  • Figure 5: Cell samples of SIPaKMeD dataset plissiti2018sipakmed
  • ...and 5 more figures