AI Models for Diagnosing Spine Cancer and Compression Fractures
Author Information
Author(s): Jeong Woosik, Baek Chang-Heon, Lee Dong-Yeong, Song Sang-Youn, Na Jae-Boem, Hidayat Mohamad Soleh, Kim Geonwoo, Kim Dong-Hee
Primary Institution: Gyeongsang National University
Hypothesis
Can artificial intelligence models accurately differentiate between metastatic spine cancer and spinal compression fractures using MRI images?
Conclusion
The study developed AI models that achieved high accuracy and sensitivity in distinguishing metastatic spine cancer from spinal compression fractures using MRI data.
Supporting Evidence
- The CNN model demonstrated high sensitivity (1.00) and accuracy (0.98) in identifying metastatic spine cancer.
- Preprocessing methods improved the performance of AI models in detecting cancerous cases.
- The study included MRI data collected from 503 patients, with 248 patients analyzed after exclusions.
- Models were able to classify cancer more accurately with high sensitivity scores.
- AI-assisted diagnosis can potentially reduce the misdiagnosis rate of spinal conditions.
Takeaway
Researchers created computer programs that can tell the difference between cancer in the spine and broken bones using special pictures called MRIs.
Methodology
The study used convolutional neural networks (CNN) and support vector machines (SVM) to analyze MRI images, applying preprocessing techniques like Otsu’s binarization and Canny edge detection.
Potential Biases
Exclusion of patients with metallic implants may introduce bias in the dataset.
Limitations
The dataset was acquired from a single institution, which may limit the generalizability of the findings.
Participant Demographics
The study included 248 patients, with a mean age of 64.48 years, consisting of 138 males (55.6%) and 110 females (44.4%).
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website