MIT Researchers Build Machine Learning Model that Quickly Generates Brain Scan Templates to Aid Diagnosis
By MedImaging International staff writers Posted on 31 Dec 2019 |
Image: Brain scan templates of various ages (Photo courtesy of Massachusetts Institute of Technology)
A team of researchers from the Massachusetts Institute of Technology (Cambridge, MA, USA) have devised a method that accelerates the process for creating and customizing templates used in medical-image analysis, to guide disease diagnosis.
Medical image analysis is used to crunch datasets of patients’ medical images and capture structural relationships that may indicate the progression of diseases. In many cases, analysis requires use of a common image template, called an “atlas,” that’s an average representation of a given patient population. Atlases serve as a reference for comparison, for example to identify clinically significant changes in brain structures over time. However, building a template is a time-consuming, laborious process, often taking days or weeks to generate, especially when using 3D brain scans. To save time, researchers often download publicly available atlases previously generated by research groups, although these fail to fully capture the diversity of individual datasets or specific subpopulations, such as those with new diseases or from young children. Ultimately, the atlas cannot be smoothly mapped onto outlier images, producing poor results.
The MIT researchers devised an automated machine-learning model that generates “conditional” atlases based on specific patient attributes, such as age, sex, and disease. By leveraging shared information from across an entire dataset, the model can also synthesize atlases from patient subpopulations that may be completely missing in the dataset. The researchers hope clinicians can use the model to build their own atlases quickly from their own, potentially small datasets.
“The world needs more atlases,” says first author Adrian Dalca, a former postdoc in the Computer Science and Artificial Intelligence Laboratory (CSAIL) and now a faculty member in radiology at Harvard Medical School and Massachusetts General Hospital. “Atlases are central to many medical image analyses. This method can build a lot more of them and build conditional ones as well.”
Related Links:
Massachusetts Institute of Technology
Medical image analysis is used to crunch datasets of patients’ medical images and capture structural relationships that may indicate the progression of diseases. In many cases, analysis requires use of a common image template, called an “atlas,” that’s an average representation of a given patient population. Atlases serve as a reference for comparison, for example to identify clinically significant changes in brain structures over time. However, building a template is a time-consuming, laborious process, often taking days or weeks to generate, especially when using 3D brain scans. To save time, researchers often download publicly available atlases previously generated by research groups, although these fail to fully capture the diversity of individual datasets or specific subpopulations, such as those with new diseases or from young children. Ultimately, the atlas cannot be smoothly mapped onto outlier images, producing poor results.
The MIT researchers devised an automated machine-learning model that generates “conditional” atlases based on specific patient attributes, such as age, sex, and disease. By leveraging shared information from across an entire dataset, the model can also synthesize atlases from patient subpopulations that may be completely missing in the dataset. The researchers hope clinicians can use the model to build their own atlases quickly from their own, potentially small datasets.
“The world needs more atlases,” says first author Adrian Dalca, a former postdoc in the Computer Science and Artificial Intelligence Laboratory (CSAIL) and now a faculty member in radiology at Harvard Medical School and Massachusetts General Hospital. “Atlases are central to many medical image analyses. This method can build a lot more of them and build conditional ones as well.”
Related Links:
Massachusetts Institute of Technology
Latest Industry News News
- Bayer and Google Partner on New AI Product for Radiologists
- Samsung and Bracco Enter Into New Diagnostic Ultrasound Technology Agreement
- IBA Acquires Radcal to Expand Medical Imaging Quality Assurance Offering
- International Societies Suggest Key Considerations for AI Radiology Tools
- Samsung's X-Ray Devices to Be Powered by Lunit AI Solutions for Advanced Chest Screening
- Canon Medical and Olympus Collaborate on Endoscopic Ultrasound Systems
- GE HealthCare Acquires AI Imaging Analysis Company MIM Software
- First Ever International Criteria Lays Foundation for Improved Diagnostic Imaging of Brain Tumors
- RSNA Unveils 10 Most Cited Radiology Studies of 2023
- RSNA 2023 Technical Exhibits to Offer Innovations in AI, 3D Printing and More
- AI Medical Imaging Products to Increase Five-Fold by 2035, Finds Study
- RSNA 2023 Technical Exhibits to Highlight Latest Medical Imaging Innovations
- AI-Powered Technologies to Aid Interpretation of X-Ray and MRI Images for Improved Disease Diagnosis
- Hologic and Bayer Partner to Improve Mammography Imaging
- Global Fixed and Mobile C-Arms Market Driven by Increasing Surgical Procedures
- Global Contrast Enhanced Ultrasound Market Driven by Demand for Early Detection of Chronic Diseases