Related Experiment Video
Updated: May 24, 2025
04:23
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
Published on: April 21, 2023
1.7K
Development and validation of automated three-dimensional convolutional neural network model for acute appendicitis
Minsung Kim1, Taeyong Park2, Jaewoong Kang2
1Department of Surgery, Hallym University Medical Center, Hallym Sacred Heart Hospital, Hallym University College of Medicine, 22 Gwanpyeong-ro 170 beon-gil, Pyeongan-dong, Dongan-gu, Anyang, Gyeonggi-do, Republic of Korea.
Scientific Reports
|March 5, 2025
Summary
A new deep learning model accurately diagnoses appendicitis using 3D CT scans. This automated framework aids surgical decisions by classifying non-, simple, and complicated appendicitis, improving diagnostic speed and accuracy.
Area of Science:
- Radiology
- Artificial Intelligence
- Medical Imaging
Background:
- Accurate preoperative diagnosis of appendicitis is crucial for timely surgical intervention in emergency care.
- Current diagnostic methods for appendicitis can be time-consuming and may lead to delays in treatment.
- The integration of artificial intelligence in medical imaging holds promise for enhancing diagnostic efficiency.
Purpose of the Study:
- To develop and validate a fully automated diagnostic framework for appendicitis using a 3D convolutional neural network (CNN).
- To assess the model's ability to differentiate between non-appendicitis, simple appendicitis, and complicated appendicitis.
- To evaluate the model's performance in terms of accuracy, sensitivity, specificity, and negative appendectomy rate.
Main Methods:
- A deep learning model, termed Information of Appendix (IA), was developed using a 3D CNN architecture (ResNet, DenseNet, EfficientNet).
- The model automatically extracted the volume of interest (VOI) corresponding to the appendix from contrast-enhanced abdominopelvic CT images.
- A two-stage binary algorithm with transfer learning was employed to classify appendicitis severity.
Main Results:
- The IA model utilizing DenseNet169 achieved 79.5% accuracy and 70.1% sensitivity in Stage 1 (non-appendicitis vs. appendicitis), with a 12.4% negative appendectomy rate.
- In Stage 2, the model demonstrated 76.1% accuracy and 82.6% sensitivity in differentiating simple from complicated appendicitis.
- The model achieved an Area Under the Curve (AUC) of 0.865 in Stage 1 and 0.827 in Stage 2, indicating strong diagnostic performance.
Conclusions:
- The developed IA model provides a reliable, automated diagnostic tool for appendicitis using CT imaging.
- This framework has the potential to improve surgical decision-making in emergency care by offering rapid and accurate diagnostic information.
- The model's generality and reproducibility within the VOI suggest its clinical utility in appendicitis diagnosis.
