Researchers are developing AI and machine learning (ML) tools for a host healthcare applications, from robotic-assisted surgeries to EHR analysis to medical image interpretations. It is vital that the output of medical devices be trustable. As of August 7, 2024, the U.S. FDA has authorized 950 AI/ML-enabled medical devices. More than 100 of these were authorized in 2024, which collect accurate input metrics from wearable devices. AI can now effectively determine demographic features from medical images, which can facilitate disease diagnosis and prognosis.
AI in medical imaging can help healthcare professionals identify problem areas or details that may be missed by the human eye. For instance, AI-powered medical imaging can analyze data points in a medical report to distinguish disease from healthy tissue, and signals from noise.
Can We Trust Medical Images?
Trusting medical images generated by CT and MRI scans is crucial for accurate diagnosis and treatment. CT and MRI scans are popular methods of imaging internal body parts. They have similar uses but produce images in different ways. CT scans use X-rays, while MRI scans use strong magnets and radio waves. A CT scan is generally good for larger areas, while an MRI scan produces a better overall image of specific tissue under examination.
Creating AI-Generated CT Scan Images from MRI Scans
Researchers are developing AI technology to generate CT images based on MRI images, which can further enhance the accuracy and reliability of medical imaging. Transcranial-focused ultrasound can be used to treat degenerative movement disorders, intractable pain, and mental disorders by delivering ultrasound energy to a specific area of the brain without opening the skull. This treatment must be performed with an image-based technology that can locate the brain lesions. Doctors typically use CT to obtain information about a patient's skull that is difficult to identify with MRI alone and to accurately focus the ultrasound on the lesions through the skull. However, there have been concerns about the safety of CT scans, during which radiation exposure is inevitable, especially in paediatric and pregnant patients.
Bionics Research Centre at the Korea Institute of Science and Technology (KIST) has proved that CT images obtained by artificial intelligence have clinical utility. Efforts have been made to obtain cranial information from MRI images, but special coils for the MRI that are not widely available in the medical field are required. The KIST research team developed a 3D, conditional adversarial generative network that learns the nonlinear CT transformation process from T1-weighted MRI images—one of the most commonly used images in the medical field. The team devised a loss function that minimizes the Hounsfield unit pixel variation error of the CT images. This method also optimizes neural network performance by comparing changes in the quality of synthetic CT images according to the normalization methods of MRI image signals. Consider Z-score normalization and partial linear histogram matching normalization as examples.
For safe and effective ultrasound treatment, it is imperative to understand each patient's skull density ratio and skull thickness in advance. When these skull factors were obtained via the synthetic CT, both factors showed >0.90 correlation with the actual CT. There was no statistically significant difference. Moreover, when simulated ultrasound treatment was performed, the ultrasound focal distance had an error of less than 1 mm. The intracranial peak acoustic pressure had an error of approximately 3.1%. The focal volume similarity was approximately 83%. This demonstrated that a transcranial-focused ultrasound treatment system can be performed with only the MRI image.
Responsible Data for AI Makes All the Difference
By having trusted data as input to medical devices, the medical community can expect best possible results as image outputs from CT scans. In the scenarios discussed (radiation concerns of CT scans in specific portions of the brain), we will continue to uncover safer options where AI can provide reasonable accuracy of image generation from MRI scan images. This is one example where responsible data for AI plays a significant role in diagnosis, treatment, and patient care.
