Deep learning for early detection of osteopenia and osteoporosis in lumbar vertebrae based on T۱-weighted lumbar MRI
محل انتشار: اولین کنگره بین المللی هوش مصنوعی در علوم پزشکی
سال انتشار: 1402
نوع سند: مقاله کنفرانسی
زبان: انگلیسی
مشاهده: 204
نسخه کامل این مقاله ارائه نشده است و در دسترس نمی باشد
- صدور گواهی نمایه سازی
- من نویسنده این مقاله هستم
این مقاله در بخشهای موضوعی زیر دسته بندی شده است:
استخراج به نرم افزارهای پژوهشی:
شناسه ملی سند علمی:
AIMS01_110
تاریخ نمایه سازی: 1 مرداد 1402
چکیده مقاله:
Background and aims: Osteoporosis and osteopenia are conditions characterized by a decreasein bone density, leading to an increased risk of fractures. These conditions affect millions of individualsworldwide and are associated with significant morbidity and mortality. Early detectionand diagnosis are crucial for the effective management of these conditions. The aim of this studywas to explore the potential of deep learning in detecting osteopenia and osteoporosis in lumbarvertebrae based on lumbar MRIs .Methods:۱۹۰ women with an average age of ۶۰.۳۰ (SD = ۹.۹۶) were selected. All subjects hadsagittal T۱-weighted spin-echo lumbar MRIs and bone mineral density examinations done bythe DEXA method. Based on the WHO classification of DEXA reports, ۶۵ subjects were normalwhile ۷۷ and ۴۸ subjects had osteopenia and osteoporosis, respectively. Five slices were selectedfrom each T۱ sequence, resulting in a total of ۹۵۰ slices. Of these, ۸۵۰ were used for training, ۵۰for validation, and ۵۰ for testing. A Densenet۱۲۱ neural network pre-trained on the Imagenet۱Kdataset was used as a base model for training on the dataset.Results: The precision, recall, and f۱-score for the normal population were ۰.۷۸۹۵, ۰.۷۵۰۰, and۰.۷۶۹۲ respectively. For those with osteopenia, the corresponding scores were ۰.۶۹۵۷, ۰.۸۰۰۰,and ۰.۷۴۴۲ respectively while for those with osteoporosis, they were ۱.۰۰۰۰, ۰.۸۰۰۰, and ۰.۸۸۸۹respectively. The overall accuracy of the model was found to be at a rate of about .۷۸۰۰ while theMCC was calculated to be at about .۶۵۴۹.Conclusion: In conclusion, this study demonstrated the potential of using deep learning and neuralnetworks to identify osteopenia or osteoporosis in patients based on lumbar MRIs. The resultssuggest that this technology could improve diagnosis and treatment decisions in clinical practice.However, further research and the collection of more data are needed to fully explore its potential.
کلیدواژه ها:
نویسندگان
Mohammad Tabaresfani
Mazandaran university of medical sciences, Mazandaran, Iran
Pooria Sobhanian
Mazandaran university of medical sciences, Mazandaran, Iran
Arash Ziaee
Mazandaran university of medical sciences, Mazandaran, Iran
Hossein Alavi
Mazandaran university of medical sciences, Mazandaran, Iran.
Amirhossein Arab
Mazandaran university of medical sciences, Mazandaran, Iran.
Misagh Shafizad
Mazandaran university of medical sciences, Mazandaran, Iran.