Skip to Main Content

An official website of the United States government

Deep Learning-Based Histopathological Model for Ovarian Cancer Prognosis Prediction

Principal Investigator
Name
Yangyang Zhang
Degrees
Ph.D.
Institution
fudan university
Position Title
Dr.
Email
About this CDAS Project
Study
PLCO (Learn more about this study)
Project ID
PLCOI-1787
Initial CDAS Request Approval
Jan 10, 2025
Title
Deep Learning-Based Histopathological Model for Ovarian Cancer Prognosis Prediction
Summary
Ovarian cancer is one of the most lethal gynecological malignancies, with a high incidence of late-stage diagnosis and poor prognosis. Traditional prognostic models rely on clinical and pathological factors, but their accuracy is often limited. This project aims to develop a deep learning model that leverages histopathological images to predict ovarian cancer prognosis more accurately. Using state-of-the-art convolutional neural networks (CNNs), the model will be trained on digitized tissue slides (WSI), extracting critical morphological and architectural features that correlate with survival outcomes. By integrating histopathological features with clinical data, the model will offer a comprehensive and personalized approach for predicting patient survival, improving early diagnosis, and facilitating targeted therapeutic interventions. The study will involve data from a large cohort of ovarian cancer patients, and external validation will be performed using independent datasets to assess the generalizability of the model.
Aims

Specific Aims:

Aim 1: Data Collection and Preprocessing
Collect histopathological images (WSI) and clinical data (age, stage, grade, etc.) from ovarian cancer patient cohorts.
Preprocess images and clinical data to ensure compatibility for deep learning model training.
Aim 2: Model Development and Training
Train a convolutional neural network (CNN) using the preprocessed histopathological images to learn critical features linked to ovarian cancer prognosis.
Integrate clinical features with the histopathological model to enhance prediction accuracy.
Aim 3: Model Evaluation and Validation
Evaluate the model's performance using internal and external validation datasets. Metrics like accuracy, AUC, and survival prediction accuracy will be assessed.
Compare the model’s performance with traditional prognostic methods to highlight improvements in prediction accuracy.
Aim 4: Clinical Implementation
Investigate the potential for implementing the model in clinical settings to support personalized therapeutic strategies for ovarian cancer patients.

Collaborators

qinhao guo，Ph.D., Department of Gynecologic Oncology, Fudan University Shanghai Cancer Center, Fudan University; Shanghai, 200032, China.