Related Experiment Video
Updated: Jun 15, 2025
13:44
Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns
Published on: August 30, 2013
42.8K
Improved rank-based recursive feature elimination method based ovarian cancer detection model via customized deep
Namani Deepika Rani1, Mahesh Babu1
1Department of Computer Science and Engineering, Koneru Lakshmaiah Education Foundation, Hyderabad, 500075, Telangana, India.
Computer Methods and Programs in Biomedicine
|August 27, 2024
Summary
A novel ovarian cancer detection model, CCLSTM, demonstrates high accuracy and consistency. This deep learning approach significantly improves early diagnosis, offering better patient outcomes for this lethal gynecological cancer.
Area of Science:
- Oncology
- Biomedical Engineering
- Data Science
Background:
- Ovarian cancer (OC) is a leading cause of gynecological cancer mortality, often diagnosed at advanced stages.
- Challenges in OC management include rapid metastasis and genetic predispositions, necessitating precise early diagnosis.
- Timely and accurate diagnosis is crucial for effective treatment planning and patient support.
Purpose of the Study:
- To develop and evaluate a novel deep learning model for accurate and consistent ovarian cancer detection.
- To improve upon existing methods for early-stage diagnosis of ovarian cancer.
- To enhance treatment planning and patient outcomes through precise diagnostic capabilities.
Main Methods:
- A four-stage approach: preprocessing (Improved Two-step Data Normalization), feature extraction (statistical, entropy, raw, mutual information), feature selection (Improved Rank-based Recursive Feature Elimination - IR-RFE), and detection using a Convolutional-Recurrent Neural Network (CCLSTM) model.
- Extraction of diverse features from normalized data to capture complex patterns.
- Utilizing IR-RFE for optimal feature subset selection to enhance model performance.
Main Results:
- The proposed CCLSTM model achieved a high sensitivity of 0.948.
- CCLSTM outperformed other methods, including ALO-LSTM + ALOCNN, Bi-GRU, LSTM, RNN, KNN, CNN, and DCNN, in sensitivity.
- The model demonstrated superior accuracy and consistency in ovarian cancer detection compared to existing strategies.
Conclusions:
- The integration of Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) techniques in the CCLSTM model yields a highly accurate and consistent ovarian cancer detection system.
- The developed CCLSTM technique represents a significant advancement over current ovarian cancer diagnostic strategies.
- This enhanced diagnostic capability holds promise for improving patient management and outcomes in ovarian cancer care.
