Table 1 Overview of the clinical tasks achieved through artificial intelligence (AI) systems in five fields of musculoskeletal imaging
From: AI applications in musculoskeletal imaging: a narrative review
Clinical setting | Main clinical tasks | Imaging modalities | Examples from literature |
|---|---|---|---|
Trauma | Fracture detection | Radiography/CT | Fractures around the hip [7], spine [8], multiple anatomic sites [9] |
Fracture classification | Radiography/CT | Fractures of the calcaneus [10], femur [11], humerus [12], spine [13], around the knee [14], and ankle [15] | |
Detection of ligament or meniscal tears | MRI | Anterior cruciate ligament and meniscal tears [16] | |
Bone age | Bone age estimation | Radiography | BoneXpert [17] and VUNO Med-BoneAge [18] for hand radiographs |
Osteoarthritis | Grading | Radiography | Grading of knee osteoarthritis [19] |
Cartilage lesion detection | MRI | Detection of knee cartilage lesions [20] | |
Prediction of progression | Radiography | Progression of knee osteoarthritis [21] | |
Bone and soft-tissue tumors | Benign/malignant discrimination | Radiography/CT/MRI | Primary bone tumors [22] |
Grading | CT/MRI | Bone chondrosarcoma [23,24,25] and soft-tissue sarcomas [26] | |
Prediction of outcomes (recurrence, survival, therapy response) | CT/MRI | ||
Orthopedic implants | Identification and classification | Radiography | Spinal hardware [31], knee [32] and shoulder [33] arthroplasty |
Implant positioning and measurements | Radiography | Acetabular component positioning after hip arthroplasty [34] | |
Implant-related complications | Radiography/MRI | Knee or hip arthroplasty loosening [35] |