AI Model Promises New Way to Detect Brain Malignancies
Central nervous system and brain cancers accounted for 248,500 deaths globally in 2022. In America, roughly 90,000 brain tumors are diagnosed annually, with data from the American Cancer Society showing that over 25,000 of these tumors are cancerous. It is projected that there’ll be an estimated 25,400 new cases of central nervous system and brain cancers in 2024.
Estimates also show that roughly one million individuals in the U.S. live with a primary brain tumor, with about 28% of all brain tumors being malignant. Glioblastoma, a fatal form of brain cancer, accounts for 50% of all malignant brain tumors in the country. This cancer has an 8-month median survival rate and a 5-year relative rate of survival of 6.9%.
Now new research has shown how reusing camouflage animal detection and combining it with artificial intelligence can help identify tumors in the brain.
Dr. Arash Yazdanbakhsh, the lead author, stated that this study was the first of its kind to combine deep neural network training and camouflage animal transfer learning to a tumor classification and detection task.
The researchers theorized that there was a similarity between cancerous cells that blend in with surrounding healthy tissue and animals hiding using natural camouflage. They argued that an artificial intelligence network which was trained to find animals in camouflage could also be used to detect tumors from image data from MRI brain scans.
For their study, a repurposed neural network was pretrained to pick out camouflaged animals into 2 artificial intelligence models, one dubbed T2Net to categorize T2-weighted MRIs and the other dubbed T1Net to categorize T1-weighted MRIs.
The brain scan data used for the study was obtained mainly from the Kaggle public databases and the Cancer Imaging Archive. They also used data from normal MRIs to train the neural network as a control.
Gliomas made up a huge share of the brain scan data, with other tumor categories including astrocytomas, oligoastrocytomas, and oligodendrogliomas. The researchers found that transfer learning from the camouflage detection strengthened the AI model’s ability to categorize brain tumors, particularly astrocytomas.
In their report, the researchers revealed that the model demonstrated a strong ability to differentiate between normal and cancerous brains.
They added that this approach to neural network training held potential, particularly when using T2-weighten MRI images, as the model demonstrated the highest improvement in accuracy. The research’s findings were reported in Biology Methods & Protocols.
Other researchers involved in the study include Haluk Ogmen, Faris Rustom, Pedram Parva, and Ezekiel Moroze.
As the detection of brain cancers becomes easier and more accurate, patients can have a chance to quickly get started on the medications being commercialized by companies like CNS Pharmaceuticals Inc. (NASDAQ: CNSP) targeting different central nervous system malignancies.
NOTE TO INVESTORS: The latest news and updates relating to CNS Pharmaceuticals Inc. (NASDAQ: CNSP) are available in the company’s newsroom at https://ibn.fm/CNSP
Please see full terms of use and disclaimers on the BioMedWire website applicable to all content provided by BMW, wherever published or re-published: http://BMW.fm/Disclaimer
Central nervous system and brain cancers accounted for 248,500 deaths globally in 2022. In America, roughly 90,000 brain tumors are diagnosed annually, with data from the American Cancer Society showing that over 25,000 of these tumors are cancerous. It is projected that there’ll be an estimated 25,400 new cases of central nervous system and brain cancers in 2024.
Estimates also show that roughly one million individuals in the U.S. live with a primary brain tumor, with about 28% of all brain tumors being malignant. Glioblastoma, a fatal form of brain cancer, accounts for 50% of all malignant brain tumors in the country. This cancer has an 8-month median survival rate and a 5-year relative rate of survival of 6.9%.
Now new research has shown how reusing camouflage animal detection and combining it with artificial intelligence can help identify tumors in the brain.
Dr. Arash Yazdanbakhsh, the lead author, stated that this study was the first of its kind to combine deep neural network training and camouflage animal transfer learning to a tumor classification and detection task.
The researchers theorized that there was a similarity between cancerous cells that blend in with surrounding healthy tissue and animals hiding using natural camouflage. They argued that an artificial intelligence network which was trained to find animals in camouflage could also be used to detect tumors from image data from MRI brain scans.
For their study, a repurposed neural network was pretrained to pick out camouflaged animals into 2 artificial intelligence models, one dubbed T2Net to categorize T2-weighted MRIs and the other dubbed T1Net to categorize T1-weighted MRIs.
The brain scan data used for the study was obtained mainly from the Kaggle public databases and the Cancer Imaging Archive. They also used data from normal MRIs to train the neural network as a control.
Gliomas made up a huge share of the brain scan data, with other tumor categories including astrocytomas, oligoastrocytomas, and oligodendrogliomas. The researchers found that transfer learning from the camouflage detection strengthened the AI model’s ability to categorize brain tumors, particularly astrocytomas.
In their report, the researchers revealed that the model demonstrated a strong ability to differentiate between normal and cancerous brains.
They added that this approach to neural network training held potential, particularly when using T2-weighten MRI images, as the model demonstrated the highest improvement in accuracy. The research’s findings were reported in Biology Methods & Protocols.
Other researchers involved in the study include Haluk Ogmen, Faris Rustom, Pedram Parva, and Ezekiel Moroze.
As the detection of brain cancers becomes easier and more accurate, patients can have a chance to quickly get started on the medications being commercialized by companies like CNS Pharmaceuticals Inc. (NASDAQ: CNSP) targeting different central nervous system malignancies.
NOTE TO INVESTORS: The latest news and updates relating to CNS Pharmaceuticals Inc. (NASDAQ: CNSP) are available in the company’s newsroom at https://ibn.fm/CNSP
Please see full terms of use and disclaimers on the BioMedWire website applicable to all content provided by BMW, wherever published or re-published: http://BMW.fm/Disclaimer
