Clinical Scorecard: Radiographic Classification of Bone Tumours Using Recommender Systems: A Historical Perspective
At a Glance
Category
Detail
Condition
Primary bone tumours, including benign and malignant types
Key Mechanisms
Radiographic imaging combined with deep learning-based recommender systems to classify and diagnose bone tumours
Target Population
Patients with primary bone neoplasms, including aneurysmal bone cyst, chondroblastoma, chondrosarcoma, enchondroma, Ewing sarcoma, fibrous dysplasia, giant cell tumour, non-ossifying fibroma, osteochondroma, and osteosarcoma
Care Setting
Specialised musculoskeletal tumour centres with access to hospital information systems and picture archiving and communication systems
Key Highlights
Early diagnosis of bone tumours is critical for prognosis and curability but often delayed due to nonspecific symptoms and limited clinician experience.
Radiographs are recommended as the initial screening tool; CT and MRI provide additional information but should not delay care.
A deep learning-based recommender system leveraging historical imaging data can classify multiple bone tumour pathologies and recommend similar cases to aid early and specific diagnosis.
Guideline-Based Recommendations
Diagnosis
Use radiographs as the initial screening modality for suspected bone tumours.
Confirm malignant lesions by histopathology; benign and intermediate lesions verified by histopathology or multidisciplinary tumour board consensus.
Combine imaging, histopathologic findings, and clinical presentation for definitive diagnosis.
Refer patients promptly to specialised musculoskeletal tumour centres for detailed imaging and diagnosis.
Management
Initiate early treatment following specific diagnosis at specialised tumour centres.
Avoid delays in care by not postponing initial radiographic evaluation for advanced imaging.
Monitoring & Follow-up
Utilise hospital information systems and picture archiving systems to track patient imaging and clinical data.
Apply recommender systems to monitor and compare new cases with historical data to support diagnostic accuracy.
Risks
Delayed diagnosis due to nonspecific early symptoms and limited clinician experience can worsen prognosis.
Inadequate imaging or incomplete clinical data may compromise diagnostic accuracy.
Patient & Prescribing Data
Patients with primary bone tumours treated at a single musculoskeletal tumour centre between 2000 and 2021
Data-driven classification and diagnosis using radiographic features and recommender systems can enhance early detection and tailored treatment planning.
Clinical Best Practices
Ensure early radiographic screening for patients with suspected bone tumours.
Refer patients to specialised musculoskeletal tumour centres for comprehensive evaluation and management.
Incorporate advanced AI tools such as deep learning-based recommender systems to assist in tumour classification and diagnosis.
Maintain high-quality, curated imaging and clinical datasets to support AI model training and validation.
Use multidisciplinary tumour boards to verify diagnoses, especially for benign and intermediate lesions.
by Florian Hinterwimmer, Ricardo Smits Serena, Nikolas Wilhelm, Sebastian Breden, Sarah Consalvo, Fritz Seidl, Dominik Juestel, Rainer H. H. Burgkart, Klaus Woertler, Ruediger von Eisenhart-Rothe, Jan Neumann, Daniel Rueckert
