HEALTH AI · CELL IMAGING · MORPHOLOGY

Development of an Advanced Artificial Intelligence Model using Morphological Features to Enhance Cervical Cancer Screening Efficiency

Combine image evidence with cell structure to study whether a model that sees both appearance and nucleus-to-cytoplasm morphology can better distinguish abnormal cells.

Project approved · Fiscal year 2027 · Awaiting budget authorization before commencement

THE RESEARCH STORY

Begin with the decision—not the model.

Pap-smear review requires specialist attention across large numbers of cells. This proposal studies an assistive model that does not treat imagery only as a black box, but also calculates interpretable morphological evidence such as nucleus and cytoplasm boundaries and the N/C ratio.

RESEARCH INSIGHT

Health AI becomes useful when it clarifies evidence rather than merely producing an impressive score. The proposal therefore connects cell segmentation and the N/C ratio to outputs specialists can inspect.

PRIMARY RESEARCH QUESTION

What must the study resolve?

Can combining image-based representations with morphological features improve accuracy, interpretability and cross-dataset robustness in cervical-cell classification?

OBJECTIVES

  1. 01

    Develop instance segmentation of nuclei and cytoplasm

  2. 02

    Extract interpretable N/C ratio and morphological features

  3. 03

    Study image–morphology fusion and evaluate cross-dataset robustness

METHOD ARCHITECTURE

From input to accountable support

The method is organized as an evidence chain. Every transformation should preserve enough context to explain what entered the system, what the model inferred, how confidence was assessed and when a person must review the case.

  1. 01

    Cell-image preparation and quality control

  2. 02

    Instance segmentation of nucleus and cytoplasm

  3. 03

    N/C ratio and morphological feature extraction

  4. 04

    Fusion with deep image features

  5. 05

    Specialist and external-dataset evaluation

WORK PACKAGES

Four connected research responsibilities

Segmentation

Develop nucleus/cytoplasm instance segmentation and measure boundary error.

Feature fusion

Combine learned image representations with auditable morphological features.

Validation

Use data made available by the National Cancer Institute and the public SIPaKMeD dataset subject to relevant permissions.

Clinical usefulness

Evaluate sensitivity, specificity, calibration, subgroup performance and error patterns requiring escalation.

EVIDENCE PLAN

What the project should produce

A research project is stronger when its outputs can be inspected independently of the final model score. The dossier therefore separates artifacts from evaluation.

Expected research assets

  • Nucleus and cytoplasm instance-segmentation prototype
  • Defined and auditable morphological feature set
  • Fusion model combining image and morphological representations
  • Cross-dataset and subgroup robustness report
  • Research pathway for specialist-supervised screening support

Evaluation dimensions

  • Dice/IoU and boundary accuracy
  • Sensitivity and specificity
  • Precision, recall and clinically relevant error analysis
  • Calibration and confidence intervals
  • External-dataset and subgroup performance

RESEARCH TRANSLATION

From prototype to a credible working system

Translation requires collaboration with pathologists, cytotechnologists, hospitals, screening services, data custodians and regulators from question design through validation. A research prototype must not enter clinical use before appropriate approval.

01Laboratory evidence
02Relevant-environment validation
03Human and governance review
04Controlled operational pilot

LIMITATIONS AND RESPONSIBLE FRAMING

The boundary is part of the research design.

This is a research proposal, not an approved diagnostic device. Model output must not replace slide interpretation or diagnosis by qualified healthcare professionals and requires research ethics, data protection and clinical validation.

This dossier communicates the documented project design and intended evaluation. It does not convert a proposal, funded status or prototype into a claim of validated operational performance. Numerical results should be cited only from a resulting peer-reviewed publication or verified project report.

PROJECT RECORD AND CITATION

A stable record for this research direction

FORMAL PROJECT TITLE

Development of an Advanced Artificial Intelligence Model using Morphological Features to Enhance Cervical Cancer Screening Efficiency

Suggested citation for this public dossier:

Arayaphan, W. (2027). Development of an Advanced Artificial Intelligence Model using Morphological Features to Enhance Cervical Cancer Screening Efficiency. Research project dossier, Warin Arayaphan.Use the original project report or resulting publication when citing methods, data or findings.

STUDENT PATHWAYS

Where students can enter the research

Student directions include biomedical segmentation, morphology extraction, multimodal fusion, calibration, explainable AI, dataset shift and responsible health AI.

Bring a research question.

We can shape it into a dataset, experiment, model, system component or responsible deployment study.

Talk to the lab →