Dear Artificial Intelligence Insider,
Researchers from the Hospital of the University of Pennsylvania have developed a proof-of-concept system that employs advanced image-processing techniques, deep learning, and Bayesian networks to generate a differential diagnosis for brain MRI studies.
In testing, the system was found to perform at the level of academic neuroradiologists for 35 common and rare diseases. Our coverage of this automated image-processing pipeline is the subject of this issue's Insider Exclusive.
Traditionally, bone age has been assessed by radiologists on radiographs. But an artificial intelligence (AI) algorithm can also accurately estimate bone age using 3D hand MRIs, according to a new study. What's more, the technology was also effective when adapted for use with x-ray exams.
In other news, AI can enhance the performance of both general radiologists and breast imaging subspecialists in reading digital breast tomosynthesis studies. It can also accurately detect urinary tract stones on unenhanced CT studies and identify the presence of cancer and changes in disease burden described on radiology reports.
Consultant Michael J. Cannavo -- the PACSman -- has produced the latest edition of his popular Practical Considerations of AI series. In part 5, Cannavo provides a reality check on the current state for radiology AI technology and offers his thoughts on how vendors can stand out in a crowded marketplace.
Research on mammography computer-aided detection software has largely focused on diagnostic accuracy, with very few studies addressing how radiologists use and perceive the technology, according to a recent study. Also, a deep-learning model can identify patients who face a higher long-term mortality risk by analyzing their chest radiographs.
In addition, an AI model that simulates a clinician's diagnostic process by analyzing both neuroimaging results and clinical cognitive impairment test scores can yield impressive results for diagnosing Alzheimer's disease. A team of researchers has also concluded that a deep-learning algorithm shows promise for helping to improve the management of thyroid nodules.
A combination of AI algorithms and augmented reality technology can enable real-time pain assessment on virtual 3D brain models. Advances in AI can also make cardiac MRI more accessible, according to contributing writer Dr. Melany B. Atkins of Fairfax Radiological Consultants.
Not surprisingly, AI was the driving force behind the most important recent advances in imaging informatics, according to a presentation at the Society for Imaging Informatics in Medicine (SIIM) annual meeting in Denver. In other coverage from SIIM 2019, AI was found to be capable of boosting triage efficiency for aortic tears on CT angiography, spotting and classifying tuberculosis on chest x-rays, finding lung nodules on CT by analyzing both the reconstructed image and the raw sinogram data, and detecting diseases of the cerebral hemispheres on brain MRI.
In case you missed it, we also have video interviews at the meeting with Drs. Jim Whitfill on AI highlights; Howard Chen on recent advances in AI; and Charles Kahn on developing safe, effective, and humane AI.
Do you have an idea for a story you'd like to see covered in the Artificial Intelligence Community? Please feel free to drop me a line.
