MSGID: 1:317/3 6402c968   
   PID: hpt/lnx 1.9.0-cur 2019-01-08   
   TID: hpt/lnx 1.9.0-cur 2019-01-08   
    Artifical intelligence approach may help detect Alzheimer's disease from   
   routine brain imaging tests    
    The tool may help clinicians identify patients who would benefit from   
   treatment    
      
     Date:   
         March 3, 2023   
     Source:   
         Massachusetts General Hospital   
     Summary:   
         Researchers have developed and validated a deep learning-based   
         method to detect Alzheimer's disease based on routinely collected   
         clinical brain images.   
      
      
         Facebook Twitter Pinterest LinkedIN Email   
   FULL STORY   
   ==========================================================================   
   Although investigators have made strides in detecting signs of   
   Alzheimer's disease using high-quality brain imaging tests collected   
   as part of research studies, a team at Massachusetts General Hospital   
   (MGH) recently developed an accurate method for detection that relies   
   on routinely collected clinical brain images. The advance could lead to   
   more accurate diagnoses.   
      
      
   ==========================================================================   
   For the study, which is published in PLOS ONE, Matthew Leming,   
   PhD, a research fellow at MGH's Center for Systems Biology and an   
   investigator at the Massachusetts Alzheimer's Disease Research Center,   
   and his colleagues used deep learning -- a type of machine learning   
   and artificial intelligence that uses large amounts of data and complex   
   algorithms to train models.   
      
   In this case, the scientists developed a model for Alzheimer's disease   
   detection based on data from brain magnetic resonance images (MRIs)   
   collected from patients with and without Alzheimer's disease who were   
   seen at MGH before 2019.   
      
   Next, the group tested the model across five datasets -- MGH post-2019,   
   Brigham and Women's Hospital pre- and post-2019, and outside systems pre-   
   and post-2019 -- to see if it could accurately detect Alzheimer's disease   
   based on real-world clinical data, regardless of hospital and time.   
      
   Overall, the research involved 11,103 images from 2,348 patients at risk   
   for Alzheimer's disease and 26,892 images from 8,456 patients without   
   Alzheimer's disease. Across all five datasets, the model detected   
   Alzheimer's disease risk with 90.2% accuracy.   
      
   Among the main innovations of the work were its ability to   
   detect Alzheimer's disease regardless of other variables, such as   
   age. "Alzheimer's disease typically occurs in older adults, and so   
   deep learning models often have difficulty in detecting the rarer   
   early-onset cases," says Leming. "We addressed this by making the deep   
   learning model 'blind' to features of the brain that it finds to be   
   overly associated with the patient's listed age."  Leming notes that   
   another common challenge in disease detection, especially in real-world   
   settings, is dealing with data that are very different from the training   
   set. For instance, a deep learning model trained on MRIs from a scanner   
   manufactured by General Electric may fail to recognize MRIs collected   
   on a scanner manufactured by Siemens.   
      
   The model used an uncertainty metric to determine whether patient data   
   were too different from what it had been trained on for it to be able   
   to make a successful prediction.   
      
   "This is one of the only studies that used routinely collected brain   
   MRIs to attempt to detect dementia. While a large number of deep   
   learning studies for Alzheimer's detection from brain MRIs have been   
   conducted, this study made substantial steps towards actually performing   
   this in real-world clinical settings as opposed to perfect laboratory   
   settings," said Leming. "Our results -- with cross-site, cross-time, and   
   cross-population generalizability -- make a strong case for clinical use   
   of this diagnostic technology."  Additional co-authors include Sudeshna   
   Das, PhD and, Hyungsoon Im, PhD.   
      
   This work was supported by the National Institutes of Health and by the   
   Technology Innovation Program funded by the Ministry of Trade, Industry   
   and Energy, Republic of Korea, managed through a subcontract to MGH.   
      
       * RELATED_TOPICS   
             o Health_&_Medicine   
                   # Alzheimer's_Research # Healthy_Aging #   
                   Diseases_and_Conditions # Today's_Healthcare   
             o Mind_&_Brain   
                   # Alzheimer's # Disorders_and_Syndromes # Dementia #   
                   Caregiving   
       * RELATED_TERMS   
             o Alzheimer's_disease o Dementia_with_Lewy_bodies o   
             Energy_(healing_or_psychic_or_spiritual) o Pilates o   
             Deep_brain_stimulation o Early_childhood_education o   
             Personalized_medicine o Confocal_laser_scanning_microscopy   
      
   ==========================================================================   
   Story Source: Materials provided by Massachusetts_General_Hospital. Note:   
   Content may be edited for style and length.   
      
      
   ==========================================================================   
   Journal Reference:   
      1. Matthew Leming, Sudeshna Das, Hyungsoon Im. Adversarial confound   
         regression and uncertainty measurements to classify heterogeneous   
         clinical MRI in Mass General Brigham. PLOS ONE, 2023; 18 (3):   
         e0277572 DOI: 10.1371/journal.pone.0277572   
   ==========================================================================   
      
   Link to news story:   
   https://www.sciencedaily.com/releases/2023/03/230303105255.htm   
      
   --- up 1 year, 4 days, 10 hours, 50 minutes   
    * Origin: -=> Castle Rock BBS <=- Now Husky HPT Powered! (1:317/3)   
   SEEN-BY: 15/0 106/201 114/705 123/120 153/7715 226/30 227/114 229/111   
   SEEN-BY: 229/112 113 307 317 400 426 428 470 664 700 292/854 298/25   
   SEEN-BY: 305/3 317/3 320/219 396/45   
   PATH: 317/3 229/426